文艺复兴科技

是的，是规模化实战经验。这正是关键所在，也让他们两人可能成为Rentec史上最重要的招聘——甚至超越了之前所有杰出的学者。因为他们不仅数学能力强，还拥有大型系统经验。吉姆和尼克明白，由于数据量和市场复杂度，要进军股票领域就必须建立更复杂的系统。

Yes. At operational scale. And this is what's so important and why the two of them become probably the most critical hires in Rentec's history, even including all the great academics that came before them. Because they're good on the math side, but they have this large systems experience. And Jim and Nick know that if they're gonna move into equities because of the volume of data and because of how much more complex that market is, they need more complex systems.

而RENTECH现有的学术背景人才既没有相关经验，也从未构建过类似系统。

And the current talent at RENTECH coming from academia has just never experienced that or built anything like it.

他们所处的世界正以惊人的速度在复杂性和维度上爆炸式增长。具体来说，他们挖掘和分析的是股票间的逐笔交易数据，试图描绘每只股票与其他所有股票的关系网络——不仅限于当下时刻，还包括历史与未来的每个时间点。关键在于，当算法识别出这些模式时——请注意是算法而非人类在识别。

And the world that they're entering is just exploding in complexity and dimensionality. And when I say that, here's what I mean. The data that they are mining, that they're looking for, is this intraday tick data between every stock trading. So they're in this sort of trying to map the relationship between one stock and every other stock, not just at that moment in time, but every time before it and every time after it. They're also once they do identify patterns, which this is key, the algorithms identify the patterns.

这不是靠人类直觉说'油价上涨时航空股会受冲击'，而是计算机通过机器学习发现数据规律。第二步则是根据发现的概率，实际部署哪些交易才能盈利？这些企业间错综复杂的权重关系意味着你不可能只做单一交易。

It's not a human with a hunch saying, I think when oil prices go up, the airline prices are gonna get hit. It's computers doing machine learning to discover the patterns in the data. Then there's the second piece of, well, what trades do you actually put on to be profitable from the probabilities that you just discovered? All these weights of relationships between all of these different companies. You're not just putting on one trade.

你需要同时进行10笔、100笔甚至上千笔交易，既作为对冲又能锁定特定目标变量——注意是计算机在寻找这些变量。交易规模必须精确控制以避免扰动市场。这本质上是个超高维度的复合难题：既要处理数据输入，又要制定反应策略。所有计算都必须极速完成，因为行动窗口不是以毫秒计——

You're putting on 10, a 100, thousands of simultaneous trades both to hedge to be able to isolate some particular variable that you're looking for. Again, not you, but a computer is looking for. And you also need to do it in such specific bite sizes so that you don't move the market. So you're looking for a super multivariate, multidimensional problem both on the data ingestion side and on the how do I actually react to it side. And all of this computation can't take a long time because you must act, you know, not in milliseconds.

但也不是那种抢跑市场的高频交易——很多人误解他们是做这个的（稍后会详谈）。他们实际需要在分钟级时间内快速响应，因此需要极其高效的计算机系统。

It's not a high frequency trading that's front running the market. That's not actually what they do. A lot of people think it is, but we'll get to that later. But they do need to act with reasonable quickness, probably on the order of minutes. So these need to be really efficient computer systems too.

没错。股票市场的维度性和关联性复杂得多。全球货币种类有限，具备足够流动性的就更少。但股票市场有成千上万支流动性充足的标的，它们彼此之间都存在某种相关性。

Yep. And the universe of equities is so much more multidimensional and interrelated. There are only so many currencies in the world, and there are especially only so many currencies that are large enough trading markets that you can operate in. There's not infinite, but thousands and thousands of equities in the world that are deep enough markets that you can operate in. And to some degree, they're all correlated with one another.

复杂层级还在持续叠加。许多股票在多个交易所上市，因此你还需要捕捉同一支股票在不同市场、不同时点的价差。需要分析、关联并采取行动的多维因素简直层出不穷。

And just keep adding layers of complexity here. Keep adding new things to multiply by. Many of these are traded on multiple exchanges. So you might also be looking for pricing disparities on the same equity on different markets at different points in time. So there's just dimensions upon dimensions of things to analyze, correlate, and act upon.

于是帕特森和西蒙斯像乔布斯'窃取'施乐PARC技术那样挖来了IBM的彼得、鲍勃及程序员同事大卫·马格曼。他们在Rentec开始构建股票模型，结果不仅大获成功，影响更远超预期——因为鲍勃和彼得发现：我们其实应该建立统一的全能模型。

So Patterson and Simons go raid IBM. They're like Steve Jobs raiding Xerox PARC. They bring Peter and Bob and one of their programming colleagues, David Magerman, over from IBM into Rentec, and they get started on building the equities model. But it turns out, a, they're obviously very successful at that. But the impact that they have and what they build is even bigger because Bob and Peter realized that, hey, actually, we should just have one model for everything here.

在货币、大宗商品、股票等所有领域，一切都相互关联。所有数据都是信号。股票市场并非完全独立于货币市场或大宗商品市场的动态，它们之间处处存在关联。我们真正需要的只是一个统一的模型。

For currencies, for commodities, for equities, everything is correlated. Everything is a signal. It's not like the equities market is wholly independent and separate from what's happening in currencies or what's happening in commodities. There are relationships everywhere. We really want just one model.

这简直是一项天方夜谭般的壮举，特别是在九十年代早中期。

This is like a fantastical undertaking, especially in the early to mid nineties.

没错。但若能实现，就意味着能做出些有趣的事——比如某个新兴市场数据匮乏，但它与我们掌握数据的某个领域高度相似。通过统一模型，就能将既有经验迁移应用到这个数据稀缺的新领域。正因我们采用独此一家的统一模型，才能发现世人未知的规律。

Right. But if you can nail it, it means that you can do interesting things like, hey. We don't have a lot of data on this particular market, but it looks a lot like something we do have data on. So if it's all part of the same model, we can kinda just apply all the learnings from this other thing onto this brand new thing that we're looking at with little data for the first time. And because we're putting it all in one model and no one else in the world is, we can discover patterns that no one else knows about.

事实证明，这正是鲍勃和彼得为Rendtech带来的第二大创新——单一模型的实际产品表现。最重要的是：单一模型和基础设施让全公司人员协同工作，这对聚集全球顶尖人才的机构尤为关键。此前Rentec内部模型割裂，团队间的洞见创新无法互通。

It turns out that this was actually the second most important innovation that Bob and Peter bring to Rendtech, the actual product and performance of having one model. The most important thing is that if you have only one model, one infrastructure, everybody in the firm is working on that same model. You can all collaborate all together, which is especially important when you have the smartest people in the entire world all in one building. Before this, there were separate models within Rentec. So insights and innovations and work that one team was doing on one model wouldn't get applied or translate over to work that was happening by another team on another model.

他们确实建立了鼓励知识共享的文化，但需要有人牺牲午餐时间学习他人成果，再移植到自己的版本中。这存在滞后性，甚至可能无法落实。

They did have the cultural element where it was encouraged that you share your learnings, but someone would have to take the time during their lunch break and go learn from you about those and then implement it in their version. There's a lag, and it may actually not get implemented.

是的，这完全称得上颠覆性创新。当今没有任何大型投资机构——尤其是量化公司——采用单一模型运作。其他机构的组合经理和跨策略团队存在内部竞争，更不用说城堡集团这边的工作根本影响不到另一边的业务。

Yep. This is wholly unique and revolutionary. No other at scale investment firm, period, and especially quant firm, operates this way today with just one model. Their portfolio managers and teams and multi strategy people are culturally competitive with one another. But even if they're not, the work that you're doing on this side of Citadel is not impacting the work that you're doing on that side of Citadel.

正是如此。鲍勃和彼得将Rentec所有资源整合归一，形成万众一心的合力。

Right. What Bob and Peter do is they unify everything at Rentec. So all the wood is going behind one arrow.

是的。现在正是感谢我们节目的好朋友ServiceNow的好时机。我们曾向听众讲述过ServiceNow惊人的创业故事，以及他们如何成为过去十年表现最佳的公司之一，但有些听众对ServiceNow具体做什么提出了疑问。所以今天，我们就来解答这个问题。

Yes. Now is a great time to thank good friend of the show, ServiceNow. We have talked to listeners about ServiceNow's amazing origin story and how they've been one of the best performing companies the last decade, but we've gotten some questions from listeners about what ServiceNow actually does. So today, we are gonna answer that question.

首先，最近媒体经常用一句话来形容ServiceNow，即所谓的'企业级AI操作系统'。但具体来说，ServiceNow二十二年前成立时只专注于自动化。他们最初将实体文书工作转化为软件流程，服务于企业内部的IT部门。仅此而已。随着时间的推移，他们在这个平台上构建了更强大、更复杂的任务处理能力。

Well, to start, a phrase that has been used often here recently in the press is that ServiceNow is the, quote, unquote, AI operating system for the enterprise. But to make that more concrete, ServiceNow started twenty two years ago focused simply on automation. They turned physical paperwork into software workflows initially for the IT department within enterprises. That was it. And over time, they built on this platform going to more powerful and complex tasks.

他们的服务范围从IT部门扩展到人力资源、财务、客户服务、现场运营等其他部门。在过去二十年的发展过程中，ServiceNow已经完成了连接企业各个角落并实现自动化所需的所有繁琐基础工作。

They were expanding from serving just IT to other departments like HR, finance, customer service, field operations, and more. And in the process over the last two decades, ServiceNow has laid all the tedious groundwork necessary to connect every corner of the enterprise and enable automation to happen.

所以当AI技术出现时——从定义上来说AI本身就是高度复杂的任务自动化——而谁已经建立了支持这种自动化的平台和企业连接网络？正是ServiceNow。因此要回答'ServiceNow现在做什么'这个问题，他们自称'连接并赋能每个部门'绝非虚言。

So when AI arrived well, AI kinda just by definition is massively sophisticated task automation. And who had already built the platform and the connective tissue with enterprises to enable that automation? ServiceNow. So to answer the question, what does ServiceNow do today? We mean it when they say they connect and power every department.

IT和人力资源部门用它来管理全公司的人员、设备和软件许可证。客户服务部门使用ServiceNow处理诸如检测支付失败并路由到内部正确的团队或流程来解决问题。供应链组织则用它进行产能规划，整合其他部门的数据和计划以确保所有人步调一致。不再需要在不同应用程序间切换并重复输入相同数据。最近，ServiceNow还推出了AI助手，任何岗位的员工都可以创建AI助手来处理繁琐事务，让人力专注于更具战略性的工作。

IT and HR use it to manage people, devices, software licenses across the company. Customer service uses ServiceNow for things like detecting payment failures and routing to the right team or process internally to solve it. Or the supply chain org uses it for capacity planning, integrating with data and plans from other departments to ensure that everybody's on the same page. No more swivel chairing between apps to enter the same data multiple times in different places. And just recently, ServiceNow launched AI agents so that anyone working in any job can spin up an AI agent to handle the tedious stuff, freeing up humans for bigger picture work.

ServiceNow去年入选了《财富》全球最受赞赏公司榜单和《快公司》最佳创新者工作场所，正是源于这一愿景。如果您想在业务的每个角落利用ServiceNow的规模和速度优势，请访问servicenow.com/acquired，只需告诉他们是本和大卫推荐您的。

ServiceNow was named to Fortune's world's most admired companies list last year and Fast Company's best workplace for innovators last year, and it's because of this vision. If you wanna take advantage of the scale and speed of ServiceNow in every corner of your business, go to servicenow.com/acquired and just tell them that Ben and David sent you.

谢谢ServiceNow。那么，大卫，关于股票机器...

Thanks, ServiceNow. So, David, the equities machine.

是的。这确实是一台机器。彼得和鲍勃分别在1993、1994和1995年加入。他们正在构建这个系统。Rentech开始涉足股票市场。

Yes. And indeed, a machine it is. So Peter and Bob come in in 1993 and 1994, 1995. They're building this. Rentech is getting into equities.

想象一下你们在1994和1995年使用的电脑。令人震惊的是，他们用当时的技术实时分析这些市场，实现了如此复杂的计算、协调和成果。

And, yeah, just imagine the computers that you were using during 1994 and 1995. It is astonishing the level of computational complexity and coordination and results that they are pulling off, again, in real time analyzing these markets with the technology that was available during those years.

没错。令人惊叹的是，回报率可能略有下降，当然比1994年的爆发性增长有所回落，但每年仍保持在30%以上，多数年份超过40%。在进入这个更为复杂的市场时，他们能保持这样的表现，同时还在扩大管理资产规模，这简直难以置信。

Yes. And here's what's amazing. Returns go down maybe slightly, certainly a bit from the blowout year that 1994 was, but they're still above 30% every single year. Most years above 40%. This is unbelievable that they're maintaining this performance as they're going into this hugely more complex market, and they're scaling assets under management.

因此，到九十年代末，Medallion的管理资产已接近20亿美元，通过进入股票市场保持了大致相同的表现。这非常了不起。

So by the end of the nineteen nineties, Medallion has almost $2,000,000,000 in assets under management while maintaining roughly the same performance by getting into equities. This is huge.

是的。大卫，如果你看看这些数据并计算一下：1994年他们的管理资产是2.76亿美元，增长了93%。第二年达到4.62亿美元，增长了52%。再下一年是6.37亿美元。

Yep. And, David, if you just kind of look at this and do the math okay. So '94, their AUM was 276,000,000, and they grew 93%. And then their AUM the next year was 462,000,000, and then they grew 52%. And their AUM the next year was 637,000,000.

你大概很快就能明白我的意思，他们不是通过引入新投资者来扩大管理资产规模的。

You kinda quickly get where I'm going here, which is, oh, they're scaling AUM not by bringing in new investors.

对。它对新投资者是封闭的。

Right. It's closed to new investors.

这一切都在不断累积。这是他们在1993年拥有的相同资本，从年初的1.22亿美元增长到1999年的15亿美元。

It's all just compounding. This is the same capital that they had in 1993 that has gone from 122,000,000 at the beginning of that year to 1999 being 1,500,000,000.0.

没错。然后在二月份那年，他们彻底爆发了。毛回报率128%，扣除费用后净回报率98.5%。这简直疯狂。

Yes. And then in the year February, they just totally blow the doors off. A 128% gross returns, net returns after fees of 98.5%. This is bananas.

他们将基金规模从19亿美元增长到38亿美元的管理资产。再次强调，完全依靠投资收益，而非吸纳新投资者。那一年正是科技泡沫破裂的时候。

They grow the fund from 1,900,000,000.0 to 3,800,000,000.0 of assets under management. Again, purely by investing gains, not by getting any new investors. The year the tech bubble burst.

是的。当整个市场其他部分都大幅下跌时，Medallion基金当年毛收益却上涨了128%。这成为了一个规律——高波动时期正是Medallion真正闪耀的时刻。

Yes. While the whole rest of the market is down big time, Medallion is up a 128% gross on the year. And this becomes a theme. High volatility is when medallion really shines.

看这里，不相关性。他们在此获得了最终认证：我们不仅是印钞机，而且是能在任何市场环境下运行的印钞机，无论大盘状况如何。正如大卫你所说，波动性实际上让他们的算法运作得更好。因为他们到底在做什么？

And here you go. Uncorrelated. They have their final stamp of approval right here of not only are we a money printing machine, we are a money printing machine in all environments regardless of the state of the broad market. And, David, as you said, volatility actually makes their algorithms work even better. Because what are they doing?

他们在寻找市场行为反常的场景，从而利用人们做出错误决策的机会。每当投资者面临压力时，Medallion就能获得些许优势——它会判断：哦，你现在是恐慌性抛售？我能判定你是否应该恐慌抛售。如果我认为你不该抛售该资产，我就会从你手中买下它。

They're looking for scenarios where the market's gonna act erratically, and they can take advantage of people making decisions that they shouldn't. And anytime any investors are under pressure, there's a little bit of edge that's gonna accrue to a medallion that's saying, oh, okay. You're fear selling right now? Well, I can determine if you should be fear selling or not. And if I determine that you shouldn't be dumping that asset, I'm buying it from you.

关于这个有个很有趣的故事，充分展现了吉姆管理公司和人员的才华，也说明这一年他们真正领悟了这一点。科技泡沫破裂的头几天，Medallion其实遭受了多次重大亏损。部分原因可能是模型尚未调试妥当，因为Rentec公司没人见过市场出现这类行为。也可能那几天确实表现不佳。对所有人来说那都是压力巨大的时期。

So there's a really fun story around this that really illustrates Jim's genius in managing the firm and the people and how this year was when they really figured this out. So the first couple days of the tech bubble bursting, Medallion actually takes a bunch of large losses. And part of it might be that the model wasn't tuned right yet because nobody at Rentec had seen this type of behavior in the market before. Part of it might also be too that it didn't perform well for those couple days. It's a really stressful time for everybody.

要知道，所有人都在吉姆的办公室里。吉姆抽着烟，整个房间烟雾缭绕，大家正讨论该怎么办。吉姆最终决定降低风险——他担心公司会垮掉。

You know, everybody's in Jim's office. Jim's smoking his cigarettes. It's a cloud of smoke, and they're debating what to do. And Jim makes the call to take some risk off. He's worried about blowing up.

当时距离长期资本管理公司出事还没多久。模型可能显示我们应该继续做多，但别把公司搞垮了。果然，事后彼得·布朗因这几天的亏损向吉姆提出辞职。吉姆却说：你在说什么胡话？

We're not very far removed at this point from long term capital management. The model may be saying we should stay long here, but let's not blow up the firm. Yep. After this goes down, Peter Brown comes to Jim and offers to resign given the losses that they incurred over these couple days. And Jim says, what are you talking about?

你当然不该辞职。经历过这次事件后，你对公司更有价值了，现在你明白了不能在所有情况下都百分百相信模型。

Of course, you shouldn't resign. You are way more valuable to the firm now that you've lived through this, and you now know not to 100% trust the model in all situations.

真有意思，这个洞察太精辟了。正好展现了吉姆的领导力。

It's fascinating. It's such a good insight. That illustrates Jim as a leader right there.

确实如此。还有个类似的故事：2009年吉姆退休时，彼得和鲍勃接任联席CEO。就在此前一年左右，发生了所谓的'量化地震'——类似科技泡沫破裂，市场上所有量化公司突然遭遇巨额回撤，连任特克也未能幸免。当时彼得极力主张相信模型、保持风险敞口，说这将是我们大赚的时机。但吉姆踩了刹车，亲自干预降低了风险。

It totally does. There's a parallel story when Jim ultimately does retire in 2,009, and Peter and Bob take over as co CEOs, where a year or so before the, quote, unquote, quant quake had happened, where similar to the tech bubble bursting, there was all of a sudden very large drawdowns among all quantitative firms in the market, and Rentec gets hit. And during that period, Peter argued very strenuously that we should trust the model, stay risk on. This is gonna be an incredibly profitable time for us. And Jim pumped the brakes and stepped in, intervened, and took risk off.

后来在CEO交接时，彼得又去找吉姆：'吉姆，你不担心我接手后会太激进，哪天把公司搞砸吗？'吉姆回答：'完全不担心。我知道你当时那么激进是因为有我在制衡你。'

And Peter goes to Jim again around the CEO transition and says, hey, Jim. Aren't you worried that with me running the place now, I'm gonna be too aggressive and blow it up one of these days? And Jim says, no. I'm not worried at all. I know you were only so aggressive in that moment because I was there pushing back on you.

等你真正坐上这个位置，反而不会那么激进了。他真是洞察人性的大师。

And when you're in the seat, you're gonna be less aggressive. He's just such a master at insight into human behavior.

确实如此。我甚至发现自己会自然而然地站在对面人的对立面。如果有人表现得有些强势，我会发现自己采取的立场，在停下来反思时，会意识到：我没想到在这次对话中会采取这种立场。但你知道，你本能地想要扮演另一方，来平衡坐在你对面的那个人。

It is so true, though. I even find this about myself that I will naturally take the position of the foil to the person across from me. So if somebody's being pushy in some way, I'll find myself taking a position where if I pause and reflect, I'm like, I don't think I expected to take this position coming into this conversation. But, you know, you naturally want to sort of play the other side to balance out the person sitting across from you.

没错。回到2000年和那次惊人表现，本，就像你之前提到的非相关收益。他们不仅在那年大获全胜，而且是在市场低迷时做到的。我们现在必须引入夏普比率这个概念，金融界的听众都知道，但对其他人来说，这是个非常重要的概念。

Yep. So back to the year 2000 and this incredible performance, Ben, to what you were saying earlier about uncorrelated returns. Not only did they shoot the lights out that year, they're doing it when the market is down. We gotta introduce this concept of a Sharpe ratio now, which for all of you listeners that are in the finance world, you'll know this. But for everybody else, this is a really important concept.

我认为人们凭直觉就能理解。我们在本期节目中多次提到这个概念，好吧。很棒。拥有一只基金一年翻25倍，或者投资回报率达到100%，或者昨天买的比特币一夜翻倍，这很惊人。但这能让你成为世界上最好的投资者之一吗？

And I think people grasp it intuitively. We've mentioned this concept a couple times this episode where, okay. Great. It's amazing to have a fund that 25 x's or a year where you have a 100% investment return, or I bought Bitcoin yesterday and it doubled overnight. Does that make you one of the best investors in the world?

我们都直觉地知道。不，不能。因为那可能只是运气。也许你承担了极大的风险。

We all intuitively know. No. It doesn't. Because maybe that was a fluke. Maybe you're taking on an extreme amount of risk.

问题总是：在考虑你所承担的风险后，能否在计入风险的情况下产生超额回报？因此，作为基金经理，你基本上可以通过两种方式为投资者创造价值：要么跑赢市场，要么完全与市场不相关但仍获得市场回报。或者像RENTECH那样两者兼顾——既与市场不相关又大幅跑赢，这实际上是资金管理的圣杯。

And then the question is always adjusting for the risk that you're taking, can you produce a superior return taking the risk into that account? And so you basically can provide value to investors as a fund manager in two ways. You can outperform the market, or you can be entirely uncorrelated with the market and get market returns. Or what you can do as RENTECH is both. You can be uncorrelated and massively outperform, which is effectively the holy grail of money management.

是的。因此夏普比率就是结合这两个概念的衡量指标。

Yes. And so the Sharpe ratio is a measurement combining these two concepts.

没错。它以经济学家威廉·F·夏普命名，于1966年首创。它实质上是衡量基金表现相对于无风险利率的指标。如果你当年的收益率是15%，而无风险利率是3%，那么分子就是12%，并将其与波动率或标准差进行比较——技术上来说就是标准差。

Exactly. So it's named after the economist William f Sharpe. It was pioneered in 1966. It is effectively the measure of a fund's performance relative to the risk free rate. So if you performed at 15% that year and the risk free rate was 3%, then, you know, your numerator is gonna be 12%, and it is compared against the volatility or the standard deviation is technically what it is.

但是，实际上，过去x年里你的波动性有多大？通常，我们会看三年夏普比率、五年夏普比率或十年夏普比率。夏普比率代表投资者每承担一单位额外风险所获得的超额回报。所以，大卫，你开始抛出数字了。低夏普比率是不好的。

But, effectively, how volatile have you been the last x years? And, typically, it's looked at as a three year sharp or a five year sharp or a ten year sharp. The sharp ratio represents the additional amount of return that an investor receives per unit of an increase in risk. And so, David, you're starting to throw out numbers. Low sharp ratios are bad.

负夏普比率更糟，因为这意味着你的表现低于无风险利率。高夏普比率是好的，因为它意味着你产生了大量回报，而你的方差或标准差，或者说风险很低。所以1990年，他们的夏普比率是2.0，是标普500基准的两倍。太棒了。是的。

Negative sharp ratios are worse because that means you're underperforming the risk free rate. High Sharpe ratios are good because it means that you're producing lots of returns and your variance or your standard deviation or your sort of risk is low. So in 1990, they had a Sharpe of two point o, which was twice that of the S and P 500 benchmark. Awesome. Yep.

很好。1995年到二月，夏普比率2.5，真的开始变得难以置信了。

Good. 1995 to February, Sharpe ratio of 2.5, really starting to Pretty unbelievable.

很好。我要去哪里签约投资？

Good. Where do I sign up to invest?

在某个阶段，他们加入了外国市场，实现了6.3的夏普比率，这是最佳量化公司的两倍。这家公司几乎不可能亏钱，至少历史上是这样，并且大幅跑赢

At some point, they added foreign markets and achieved a Sharpe ratio of 6.3, which is double the best quant firms. This is a firm that has almost no chance of losing money, at least historically, and massively outperforms

市场，且相关性很低。我相信，如果我的研究没错的话，在2004年2月，他们实际上实现了7.5的夏普比率。令人震惊。你知道，再次回到我们的体育类比。这些不是名人堂级别的数字。

the market on an uncorrelated basis. And I believe, if I have my research right, in 02/2004, they actually achieved a Sharpe ratio of 7.5. Astonishing. You know, again, back to our sports analogy here. These aren't hall of fame numbers.

这些简直就像是，我不知道，让汤姆·布雷迪看起来像个三流球员。

These are like, I don't know, make Tom Brady look like a third stringer.

是的，正是如此。

Yes. Exactly.

于是在2000年及其后的上涨背景下，次年2001年2月，他们将基金的附带权益从原来的20%或25%提高到了36%。请注意，此时基金已对新投资者关闭，所以仍有外部投资者在基金中，但没有新投资者加入。接着在2002年2月，他们又将附带权益提高到44%。能拿到这样的条件真是厉害。

So on the back of 2000 and this rise, the next year in 02/2001, they raise the carried interest on the fund to 36% up from either 20 or 25%, whatever it was before. Now remember, they've already closed the fund to new investors. So there's still outside investors in the fund, but no new investors are coming in. And then the next year in 02/2002, they raised the carry to 44%. I mean, great work if you can get it.

但作为对比，像红杉这样的标杆企业，其附带权益高得离谱也就30%。44%是前所未有的。

But for context, the Sequoias, the benchmarks out there, they have obscene carry of 30%. 44 is unprecedented.

这可以从两个有趣的角度来看。其一，他们可能故意把条件提到极高，以此变相清退现有投资者——虽然不明说赶人，但基金长期不接纳新业务，暗示投资者该自行退出。其二，我认为更合理的解释是：投资者本质是套利者，他们发现了定价偏差。

There's two interesting ways to look at this. One, they're just trying to jack it up so high that they just purge their existing investors out, where they're saying, we're not gonna kick anyone out yet, but we've been closed to new business for a long time now. You should see yourself out at some point. The other way to look at this, which I think is probably the right way to look at it, is investors are arbitragers. They see a mispricing.

他们进入市场纠正这种偏差。就像当某种货币在两个交易所存在价差时，投资者就会介入套利，赚取微薄利润作为回报，同时促使市场价格回归真实价值——让市场成为称重机而非投票机。某种程度上，文艺复兴公司此刻对自身或投资者做的正是这件事。他们实际上在说：

They come into the market. They fix that mispricing. So anytime that there's an opportunity to bring the way that a currency is trading on two different exchanges closer together, investors are serving their purpose of coming in, arbitraging that difference, taking a little bit of profit as a thank you, and then sort of fixing the market to make the market a true weighing machine, not a voting machine, but making it so that all prices reflect the value of what something is actually worth. And in some ways, that's what Renaissance is doing here to themselves or to their investors. They're coming in and saying, look.

这太荒谬了。我们明显跑赢市场这么多。即便我们拿走更大份额，你们仍会接受这笔交易，因为当前定价显然不合理。这个产品不该只值25%的附带权益，它理应获得高得多的分成比例——而你们依然会趋之若鹜。

This is obscene. We so clearly outperformed the market. You're still gonna take this deal even if we take more of this because there's just a mispricing here. This product should not be priced at 25% carry. This product should be priced at a much higher carried interest, and you're still gonna love it.

对于年化15%回报的机构，你们支付20%附带权益；而我们带给你们50%的年化回报。

You should pay 20% carry for a firm that delivers you 15% annual returns. We're delivering you 50% annual returns.

确实。所以我猜现有投资者对此肯定不满，但他们手握太多筹码，事情会怎么发展还真不好说。

Totally. So I have to imagine it didn't go over well with the existing investors, but they just have so much leverage that what's gonna happen.

好的。再次向听众们道歉，我得说请再等一分钟，关于分成部分我还有个观点要补充，但我想先讲完这个故事。2001年2月，他们将分成比例提高到36%。

Okay. Once again, I'm sorry, audience. I have to say, hold on one more minute for another perspective that I have to offer on the carry element, but I wanna finish the story first. Okay. So 02/2001, they raised the carry to 36%.

2002年2月涨到44%。到了2003年2月，他们直接表态：听着，我们没法用基金外的激励措施留住你们，我们要清退外部投资者了。所以从2003年2月起，所有非Rentec家族成员的外部投资者——就是既非现雇员也非公司校友的——全被请出局。

02/2002, they raised it to 44%. And then in 02/2003, they actually say, hey. We can't incentivize you out of the fund outside investors. We are gonna kick you out. So starting in 02/2003, everybody who's an outside investor who's not part of the Rentec family, you know, current employee or alumni of the firm, gets kicked out.

而且不是所有校友都能留下，只有精选的资深校友获得保留资格。是的，那么问题来了——

And not all alumni get to stay. There's select alumni that get grandfathered in. Yes. Now why did we

为什么要这么做？我马上会讲一个原因。但有个理由显而易见：如今Medallion基金的管理资产已达50亿美元，即便在股票市场，他们也面临滑点问题了。

do this? I'm gonna talk about one reason in a minute. But one reason is super obvious. The Medallion fund is now at $5,000,000,000 in assets under management that they're trading. Even in the equities market, they are now hitting up against slippage.

没错。要想维持这种惊人的超高回报率，规模确实不能再扩大了。

Yep. And so if they wanna maintain this crazy, crazy performance, they just can't get that much bigger.

这正是巴菲特整天念叨的问题，也是为什么他只能增持苹果股票，而不是去收购下一家优质家族企业。对他们而言能产生实质影响的标的实在太大了。当体量达到这种级别，进入任何市场都会引发震荡。没错，RENTECH现在采用的策略，在超过50亿美元规模时就完全行不通了。

This is the problem that Warren Buffett talks about all the time and why he has to basically just increase his position in Apple rather than going and buying the next great family owned business. The things that move the needle for them are so big that that's really all they can do. And when you are big, you're gonna move any market that you enter into. Yep. And the strategy that RENTECH is employing right now, they're just deeming doesn't work at north of $5,000,000,000.

于是在2003年2月，他们开始将外部投资者全部清出Medallion基金。但显然，市场对文艺复兴公司的投资需求依然旺盛。那他们怎么应对呢？

So in 02/2003, they start kicking all the outside investors out of Medallion. But, clearly, there's still lots of institutional demand to invest with Renaissance. So what do they do?

好吧，是时候成立新基金了。于是他们创立了文艺复兴机构股票基金。这里需要补充几点背景来理解这个决策：首先，有时市场会出现比Medallion现有资金能承载的更赚钱策略，但他们不确定这些策略能否持续。如果能确保Medallion始终能管理1015亿或2025万亿美元，他们早就扩容了。

Well, time to start another fund. So they start the Renaissance Institutional Equities Fund. And there's a couple things to add a little bit of context to really why they decide to do this. Well, the first one is sometimes there's just more profitable strategies than they had the capital to take advantage of in Medallion, but they weren't sure it would be on a durable basis. If they were sure that they could manage $10.15, $2,025,000,000,000 in Medallion all the time, then they would grow to that.

但有些策略只是阶段性出现，他们不愿承诺大幅扩大基金规模后却无法持续获得这类机会。另一个原因是，很多这类策略并不符合Medallion的运作模式，需要更长持仓周期。这带来些弊端——由于要预测更远期退出价格，这些新策略的预测准确性会降低。但他们仍认为...

But if just sometimes there's these strategies that appear, oh, we don't wanna commit to a much higher fund size and then not always have those strategies available. The other thing is that a lot of the times, those strategies aren't really what Medallion is set up to do. They require longer hold times. And so there's a little bit of downside to that because these new strategies, the predictive abilities are less because they have to predict further into the future to understand what the exit prices will be on these longer term holds. But they still figure, hey.

虽然不如短期交易那么得心应手，但做这个应该也能赚钱。

Even though it's not quite our bread and butter with the short term stuff, we should be able to make some money doing it.

有意思的是，彼得·布朗讲过个故事：有天吉姆走进他办公室说'彼得，做个思维实验——如果你娶了洛克菲勒家族的人，会建议他们把大部分财富投入标普500吗？'彼得回答'当然不会'。

Yeah. There's a fun story around this that Peter Brown tells of Jim came into his office one day and said, Peter, I got a thought exercise for you. If you married a Rockefeller, would you advise the family that they should invest a large portion of their wealth in the S and P 500? And Peter says, no. Of course not.

那根本算不上风险调整后的好收益。

That's not a great risk adjusted return.

这些人早就习惯了远超标普指数的夏普比率。

And these guys are very used to Sharpe ratios that are far better than the S and P.

没错。所以吉姆说，是的，完全正确。现在开始着手设计他们应该投资的产品吧。

Right. And so Jim says, yes. Exactly. Now get to work on designing the product that they should invest in.

对。所以他们基本上得出的结论是：我们能否创造出类似标普500但夏普比率更高的产品？我们能否每年比市场表现高出几个百分点，或者坦白说，即便只是与市场持平，但波动性比购买指数基金更低？你可以想象这对谁非常有吸引力——养老金、大型机构、那些希望以市场或略高于市场的速度复利增长，但又不愿承担巨额回撤风险的公司，或者坦白说，就是那些可能提前撤资而不愿承受大幅波动的机构。投资对冲基金相比风投基金的一个好处是你可以赎回资金。

Right. And so that's basically what they come up with is, can we create something that's like an S and P 500 with a higher Sharpe ratio? Can we beat the market by a few percentage points or, frankly, even match the market each year with lower volatility than if they were buying an index fund? And you can see who this would be very attractive to, pensions, large institutions, firms that want to compound at market or slightly above market rate, but don't wanna risk these massive drawdowns or, frankly, just big volatility in general should they need to pull the capital earlier. And the nice thing about being invested in a hedge fund versus a venture fund is you can do redemptions.

比如，如果你看13F文件，即文艺复兴机构股票基金定期向SEC提交的文件，会发现它每个季度都在变化，因为有新资金流入，也有人赎回。所以这是个相当不错的产品，至少其理论依据是提供一个对标普500而言风险更低但收益相似的产品。

Like, if you look at the 13 f's, the SEC documents that the Renaissance Institutional Equities Fund files over time, it changes every quarter because there's new people putting money in, there's people doing redemptions. So it's a pretty good product, or at least the theory behind it is a pretty good product of a lower risk similar return thing to the S and P 500.

而且营销是内置的。根本不缺想与RENTECH合作的外部资金需求。

And the marketing is built in. It's not like there's any lack of demand of outside capital that wants to invest with RENTECH.

没错。特别有意思的是，有很多故事说他们的营销文件明确写着：这不是奖章基金。我们不承诺奖章基金那样的回报。实际上，我们也没按奖章基金的标准收费。

Right. It's really funny. There's all these stories about how the marketing documents literally say, this is not the Medallion Fund. We don't promise returns like the Medallion Fund. In fact, we're not charging for it like the Medallion Fund.

大卫，你提到奖章基金的费用和提成涨到了5%和44%对吧？而这个机构基金的费用是1%和10%。每年只收1%的管理费和10%的业绩提成。

You know, David, you said that the fees and carry on Medallion went up to, what, 5 and 44. Well, on the institutional fund, the fees are one in 10. You're only taking 1% annual fee and 10% of the performance.

显然，这是完全不同的产品。

Clearly, this is a very different product.

但人们并未意识到这一点。他们非常兴奋，认为这是文艺复兴时期的产物。还是那些分析师，用着他们那些花哨的电脑。

But people did not perceive that. People were very excited. It's a Renaissance product. It's the same analysts. They're using all their fancy computers.

我确信我们会获得这种疯狂的超额收益。但归根结底，这是完全不同的投资工具。

I'm sure we're gonna get this crazy outperformance. And at the end of the day, it is an extremely different vehicle.

是的。它的表现远不及奖章基金的表现。

Yep. That has not performed anywhere near how Medallion has performed.

没错。它达到目的了吗？是的。但它像奖章基金那样吗？不。

Correct. Has it served its purpose? Yeah. But is it medallion? No.

它不像奖章基金那样特别。是的。关于机构基金还有几件有趣的事。我花了很多时间翻阅过去十年奖章基金提交的13F文件，它们都来自——我想他们有两个机构基金。

It's not special in the way that medallion is special. Yes. A couple other funny things on the institutional fund. So I spent a bunch of time scrolling through 13 f's over the last decade from the medallion filings, and they're all from I think they have two institutional funds.

对。有机构股票基金和多元化阿尔法基金。

Yep. There's institutional equities and diversified alpha.

最有趣的是他们提交的这些13F文件。大卫和我经常看节目中那些对冲基金经理朋友的13F文件，他们曾作为嘉宾来过，或者任何你想了解他们本季度买卖情况的投资者。通常你会看到十五、二十五，也许五十个不同的名字。但文艺复兴的13F文件里有4300支股票，都是些零碎的小仓位，而且季度间还有点持续性。

So the funniest thing is they file these 13 f's. And David and I are very used to looking at the 13 f's of friends of the show who run hedge funds, who we've had on as guests, or perhaps really just any investor where you wanna see, like, or what are they buying and selling this quarter. And usually, you see fifteen, twenty five, maybe 50 different names on there. Well, the 13 f for Renaissance has 4,300 stocks in these tiny little chunks. And there's a little bit of persistence quarter to quarter.

举个例子，奇怪的是，诺和诺德一直是他们最大的持仓之一。我说最大，大概在1%到2%左右。这已经是连续几个季度他们最大的持仓了。

For example, weirdly, Novo Nordisk has been one of their biggest holdings. Biggest, I say, at, like, one to 2%. That's their biggest position for several quarters in a row.

嘿，他们一直在听《Acquired》节目。

Hey. They've been listening to acquired.

没错。

That's right.

这是底部的信号之一。

That's one of the signals in the bottom.

从这些文件中你可以感觉到，这些东西当时到处乱飞，这只是他们决定拍个快照记录在纸上的那个瞬间。尽管这是季度末他们持股情况的文件，但如果早一天或一周记录，可能看起来就完全不同了。

You kinda get the sense from looking at these filings that these things were flying all over the place, and this was just the moment in time where they decided to take a snapshot and put it on a piece of paper. And even though this is the end of quarter filing of what their ownership was, if you had taken it a day or a week earlier, it could look completely different.

是的。我们交谈过的一些人这样描述机构基金和Medallion的区别：Medallion的平均持仓时间大概是一天，也许一天半。而机构基金的平均持仓时间是几个月左右。所以组合里的4300只股票每天都有大量交易活动，但具体到每只股票上交易速度比Medallion慢得多。对。

Yes. The way that some folks we talked to described the difference between the institutional funds and Medallion to us is that Medallion's average hold time for their trades and positions is, call it, like, a day, maybe a day and a half. Whereas the average hold time for the institutional funds positions is, like, a couple months. So across 4,300 stocks in the portfolio, there's a lot of trading activity that happens on any given day, but it's a lot slower in any given name than Medallion would be. Yep.

这很合理。又回到滑点这个概念。如果你管理更大的基金，投资金额更大——机构基金正是如此——你就不能交易太频繁，否则所有收益都会从指缝中溜走。

Which makes sense. Again, it gets back to this slippage concept. If you have a bigger fund and you're investing larger amounts, which the institutional funds are, you can't be trading as frequently or all of your gains are gonna slip away.

是的。坦白说，它看起来和标普500指数非常相似。比如，截至11月23日的数据显示，全年12个月中有11个月都实现了8.6%的增长。听起来就像是典型的指数回报。

Yep. And frankly, it just looks a lot like the S and P 500. Like, when you look at as of November 23, so 11 of the twelve months of the year had happened, they were up 8.6%. Okay. That sounds like an index type return.

看看2020年前四个月，疫情导致市场暴跌后，它们下跌了10.4%。虽然跌幅小于整体市场，但依然像是整体市场的一面镜子。所以我认为他们的机构基金RIEF确实如预期般运作。不，它不是奖章基金。如果单看这个基金本身，我们根本不可能在《Acquired》节目中报道它背后的机构。

You look at the first four months of twenty twenty right after the crazy dip from the pandemic, they were down 10.4%. Less than the broader market, but they still were sort of a mirror of the broader market. So I think the RIEF, their institutional fund, yes, it works as expected. No, it's not Medallion. And if it were standing on its own, there's zero chance that we would be covering the organization behind it on acquired.

0%的可能性。说到让我们在本期节目中报道这家公司的原因——那只基金，我们在科技泡沫破裂时就发现，市场波动时期正是奖章基金大放异彩的时候。而2007年和2008年2月堪称最动荡的时期。没错，2007年2月，奖章基金实现了136%的总回报。

0% chance. Speaking of the fund that is the reason why we are covering this company on this show, We set up during the tech bubble crash that volatility is when Medallion really shines. Well, there's no more volatile periods than 2007 and 02/2008. Yep. 02/2007, Medallion does a 136% gross.

2008年2月，奖章基金更是创下152%的总回报率。简直离谱！要知道这可是2008年，当时整个金融世界都在崩塌。

02/2008, Medallion does a 152% gross. Like, get out of here. Crazy. This is 2008 while the rest of the financial world is melting down.

这确实揭示了他们的盈利来源，以及这些交易的另一方是谁。是那些情绪化行事的人。他们拥有这些极其稳健、高度理性的模型，能够进行超级复杂、涉及多种证券的押注，并精确执行一系列交易以实现系统所要求的风险和敞口。而交易的另一方是谁？是恐慌性抛售者。

And so this really does illustrate where do they make their money from, who is on the other side of these trades. It's people acting emotionally. They have effectively these really robust models that are highly unemotional, that are making these super intricate, multi security bets, and they are putting on exactly the right set of trades to achieve the risk and exposure that the system wants them to have. And who is on the other side of those trades? It's panic sellers.

是牙医们。是对自己计算机系统不信任的对冲基金，他们会说‘啊，糟糕，我们得降低风险，尽管这对我们来说是负期望值的举动’。他们本质上是在与人性对抗。重要的是，与我们在这里讨论的其他行业或大多数其他行业不同，这个行业是真正的零和博弈。

It's dentists. It's hedge funds who don't trust their computer systems and are like, ah, crap. We gotta just take risk off even though it's a negative expected value move for us. They're basically trading against human nature. And importantly, in this business versus every other business that we cover here on Acquired or most other businesses, this is truly zero sum.

不像他们所处的行业是一个增长型行业，许多竞争者可以采取不同策略，整个市场规模在不断扩大，所以我不在乎。但在这里，你们是在争夺一块固定大小的蛋糕。我在与其他人交易，我赢，他们就输。

It's not like they're here in an industry that's a growth industry and lots of competitors can take different approaches, but the whole pie is growing so much that I don't care if no. You're fighting over a fixed pie here. I'm trading against someone else. I win, they lose.

是的。不过这里有个细微差别，虽然我不确定其说服力有多大。辩护者的观点可能是：沃伦·巴菲特可能站在交易的另一边，而Medallion基金可以在一天半的时间跨度内从这笔交易中获利，而沃伦则能在五十年的时间跨度里赚钱。这非常公平。

Yes. Well, there's one slight nuance to that, but I don't know how much it holds water. And the apologist nuance would be, well, Warren Buffett could be on the other side of the trade, and Medallion could make money on that trade with Warren over its time horizon of a day and a half. And Warren could make money over his time horizon of, you know, fifty years. Super fair.

但我认为反驳的论点是：Medallion在一天半后以更低价格卖给了其他买家。因此在这条链条的某个环节，损失最终会转嫁给某人。Medallion的直接对手方和整个量化行业可能不会承担损失，但总有人要为此买单。正如你所说，这是零和游戏。

So I think the argument against that, though, is that Medallion sold after a day and a half to somebody else who bought at that lower price. And so somewhere along the chain, that loss is getting offloaded to somebody. The direct counterparty of Medallion and the quant industry writ large might not take the loss, but somebody is gonna take the loss along the way. It is, as you say, a zero sum game.

没错。但关键问题是：你和对手能否双赢？在这个案例中，你和交易对手确实追求不同的目标结果。比如我能否通过这次交易从沃伦·巴菲特那里赚到一美分？当然可以。

Yeah. But I think the important thing is, can you and your adversary both benefit? And I think in this case, you and your counterparty, the person you're trading against, yes, you have two different objective outcomes. Like, can I get a penny over on Warren Buffett by managing to take him on this one trade? Sure.

但他的投资策略决定了这无关紧要。

But his strategy is such that that is irrelevant.

正如之前提到的，在金融危机期间创下历史性表现后，吉姆于2009年退休，彼得和鲍勃在2010年成为联席CEO。他们接手时将投资组合规模从吉姆任期最后几年保持的50亿提升到100亿，且未影响收益——我猜这意味着RenTec的算法模型更成熟了，否则他们早就该扩容了。

So after the historic performance during the financial crisis, as I alluded to earlier, Jim retires at the 2009, and Peter and Bob become co CEOs, co heads of the firm in 2010. They take the portfolio size up to $10,000,000,000 when they take over. It had been at five for the last few years of Jim's tenure. They take it up to 10 and really with no impact, which I assume means that Rentec was getting better and the models were getting better. Because otherwise, they would have gone to 10 before.

对。他们确信有足够多盈利交易来支撑规模扩张而不稀释回报。很可能他们本可以更早行动，只是缺乏信心。但我敢打赌他们非常清楚策略的最大有效规模。

Right. They gained confidence that they had enough profitable trades they could make that they could raise the capacity without dampening returns. Yes. And perhaps they could have done it earlier, and they just didn't have the confidence that it would work at larger size. But I bet they're very good at knowing how large can our strategy work up to before it starts having diminishing returns.

重要的是，在2020年这样的波动高峰期，Medallion依然表现惊艳。根据我们找到的数据，2020年其总收益149%，净收益76%。魔力仍在延续。虽然不能盖棺定论，但对比吉姆时代与彼得鲍勃时代的业绩是个不错的观察角度。

Yep. And importantly, during periods of peak volatility, like, say, 2020, Medallion continues to shoot the lights out. So from at least the data that we were able to find on Medallion's performance over the past few years, 2020, they were up a 149% gross and 76% net. So the magic is still there. And one way to look at it, which may not be the be all and end all, but I think is a good way to compare Jim's era at Medallion versus Peter and Bob's era.

在吉姆任职期间，从1988年基金成立到2009年他退休时，Medallion的总综合内部收益率（IRR）为年化63.5%的毛收益和40.1%的净收益——当然这包含了许多业绩提成较低的时期，20%对比44%。在后吉姆时代，即2010至2022年彼得与鲍勃主政期间（我们获得的最新数据截止于此），IRR达到77.3%毛收益和40.3%净收益。即便平均费率大幅提高，两项指标反而更优。所以没错，我认为Medallion表现相当出色。

During Jim's tenure, Medallion's total aggregate IRR from 1988 when the fund was formed to 2009 when he retired was 63.5% gross annual returns and 40.1% net annual returns, which, of course, did include many periods of lower carry, 20% versus the 44%. During the post Jim era, the Peter and Bob era from 2010 to 2022 was when we were able to get the latest data. IRRs are 77.3% gross and 40.3% net. So better on both fronts even with much higher average fees. So, yeah, I think Medallion is doing fine.

这太惊人了。

That's amazing.

而我们无法确认。有消息源称他们过去几年从100亿美元规模增长到了150亿美元并保持稳定。若属实，这意味着他们持续在Medallion内部发掘更多盈利策略，才能在更大规模下维持同样惊人的回报率。

And we weren't able to tell. There's some sources that report that they've grown from $10,000,000,000 in the last few years to being comfortable at a $15,000,000,000 fund size. And if so, that just means that they continue to find more profitable strategies within Medallion to keep those same unbelievable returns at larger sizes.

确实。说到底这一切都疯狂至极。就我们所知——本你在节目开头也略有提及——就所有人所知，Medallion拥有迄今为止历史上单一投资工具中最辉煌的投资业绩记录。

Yeah. And at the end of the day, this is all just insane. So as far as we can tell, Ben, you alluded to this a bit at the beginning of the episode, and as far as anybody else can tell, Medallion has by far the best investing track record of any single investment vehicle in history.

把那些净收益数据报给我。

So give me those net numbers.

从1988年到2022年Medallion成立至今的三十四年间，扣除费用后的年化净收益率为40%（三四十年持续保持40%）。费用前收益率为68%，按我们计算，这相当于为整个公司创造了总计600亿美元的业绩提成，贾斯汀·凯里。惊人。这真是巨额财富。

So during the entire lifetime so far of Medallion from 1988 to 2022, that's thirty four years. The total net annual return number is 40%, four zero over thirty four years after fees. It's 68% before fees, which equates to total lifetime carry dollars for the whole firm of $60,000,000,000, Justin Carey, by our calculations. Astonishing. That is a lot of money.

另外大卫·罗森塔尔，这份电子表格做得漂亮。你很久没为节目制作数据表了，我钦佩你这次的工作成果。

Also, David Rosenthal, good spreadsheet work on this. You have not done a spreadsheet for an episode in a while, so I admire your, your work on this one.

是啊，我还会用Excel。勉强算会吧。现在有了Copilot和GPT，这都快成失传的手艺了。

Yeah. I still know how to use Excel. Barely. It's gonna be a dying art now with Copilot and GPTs.

没错。好的。那么总共有600亿的carry。

That's right. Okay. So 60,000,000,000 in total carry.

600亿的carry确实是笔巨款。说到巨额资金，在结束这个故事前我们必须提到，RENTECH的金钱在社会上买到了巨大影响力。鲍勃·默瑟这个名字你们可能听着耳熟——他是Breitbart和剑桥分析的主要资助者，同时也是2016年特朗普竞选和英国脱欧运动的重要金主。不过别以为Rentec的资金只流向某一政治阵营，吉姆·西蒙斯就是民主党的大金主，RENTECH里还有许多其他人士也是如此。

So 60,000,000,000 in total carry is a lot of money. And, well, speaking of a lot of money, we do need to mention before we finish the story here that that RENTECH money has bought a lot of influence in society. So Bob Mercer, that name may have sounded familiar to many of you along the way. Bob was the primary funder of Breitbart and Cambridge Analytica, and one of the major financial backers of both the 2016 Trump campaign and the Brexit campaign in Great Britain. Now lest you think that Rentec dollars are solely being funneled into one side of the political spectrum, Jim Simons is a major Democratic donor as are many other folks at RENTECH.

对，亨利·劳弗等人也是大额捐赠者，金额与右翼的鲍勃·默瑟相当。

Yeah. Henry Laufer and other folks are also huge donors approximately to the same tune as what Bob Mercer is on the right.

没错。Rentec员工和校友们在多个选举周期中向各方捐赠了数千万美元。2016年大选后这确实成了公司的敏感问题——默瑟显然成了公司内外颇具争议的人物。

Yeah. Tens of millions of dollars, many tens of millions of dollars on all sides and through many campaign cycles here from Rentec employees and alumni. This did become a flashpoint for the firm in the wake of the twenty sixteen election. Mercer obviously became a, controversial figure both externally and internally within the firm.

尤其是当人们发现他是串联Breitbart、剑桥分析、特朗普胜选和英国脱欧的关键人物后。确实如此。

Especially once people realized he was the through line through Breitbart, Cambridge Analytica, the Trump election, and Brexit. Yes.

最终，吉姆在2017年要求鲍勃辞去联席CEO职务，他照做了。但他仍以科学家身份留在公司，继续为模型做贡献，只是不再与彼得共同执掌领导层。

Ultimately, Jim asked Bob to step down as co CEO in 2017, which he did. But he did remain a scientist at the firm and a contributor to the models even though he wasn't leading the organization with Peter from a leadership standpoint any longer.

最终，最让我惊讶的是，尽管这些人的政治信仰可能是最对立的，但他们仍然能一起共事。

Ultimately, the thing that surprised me the most is how these people all still work together despite having about the most opposite political beliefs you could possibly have.

是啊，这简直是本世纪最轻描淡写的说法了。

Yeah. Understatement of the century.

而且他们在各自的政治体系中都极具影响力且活跃。是的，鲍勃·默瑟虽然不再是文艺复兴科技公司的CEO或联席CEO，但他仍在公司工作，保持着关联。

And all being extremely influential and active in those political systems. Yes. Bob Mercer is no longer the CEO of Renaissance Technologies or the co CEO. He still works there. He's still associated.

他们仍然对彼此赞誉有加，这出乎意料。

They all still speak highly of each other. It's unexpected.

没错，我觉得用'出人意料'来形容再合适不过。

Yeah. I think unexpected is the best way to put it.

就像文艺复兴公司的一切那样，它的运作方式与外界有些不同。

Like everything with Renaissance, it works a little bit different than the rest of the world.

好的。说到这个，我们转入分析环节吧。我有个有趣的小独白想说说，本，请耐心听我讲完。

Yes. Okay. Speaking of, let's transition to analysis. And I have a fun little monologue I wanna go on if you will. Bear with me, Ben.

我认为这可以称为Rentec的操作手册，但我更愿意将其视为Rentec的织锦。我的灵感来自Costco，因为我们在调研时与许多人交谈，大家都说Rentec就像拼图一样严丝合缝。表面上看，Rentec做的事情与Citadel、D. Shaw、Two Sigma、Jane Street等其他公司无异——他们雇佣世界上最聪明的人，提供最优质的数据和基础设施，然后放手让他们大展拳脚进行盈利交易。

I think this qualifies as the Rentec playbook, but I really kinda think of it as the Rentec tapestry. And I was inspired by Costco here because we were talking to folks in the research and everybody said, you know, Rentec, it just has these puzzle pieces that fit together. On the surface, Rentec does the same things that Citadel, D. Shaw, Two Sigma, Jane Street, others, etcetera do. They hire the smartest people in the world, and they give them the best data and infrastructure in the world to work on, and they say, go to town and make profitable trades.

这两样东西——全球顶尖人才和最优数据基础设施——都是极其昂贵的资源，但终究只是可复制的商品。就像Citadel可以宣称的竞争优势，与沃尔玛和亚马逊声称'我们同样拥有规模化供应商网络来提供低价'如出一辙。但真正的魔法藏在更深层。我认为有三项紧密关联的特质让Rentec独树一帜：第一，他们让全球最聪明的人才协作而非竞争。

Those are very expensive commodities, those two things, the smartest people in the world and the best data and infrastructure, but they are commodities. Like, Citadel can say the exact same things just the same as, like, Walmart and Amazon can say, we too have large scale supplier relationships that we leverage to provide low prices to customers just like Costco. But it's underneath that where I think the magic lies. There are three very interrelated things that make Rendtech unique. So number one, they get the smartest people in the world to collaborate and not compete.

几乎所有其他金融机构内部，员工和团队之间都存在准竞争关系。

Pretty much every other financial firm out there, employees and teams within the firm quasi compete with one another.

是啊，虽然通常是以比较友好的方式，但确实如此。

Yeah. I mean, typically in kind of a friendly way, but yeah.

以风投公司为例，交易由牵头合伙人或交易团队负责，其他合伙人可能稍加协助，但主要精力都在推进自己的项目。这已经是金融界最体现同事情谊的协作方式了。

Let's take like, in a venture firm, you've got your lead partner on a deal or a deal team. They're working that deal, and maybe some of the other partners help a little bit, but mostly, they're off prosecuting their own deals. Yep. And I think that's the most collegial way that this happens in finance. Yeah.

再看多策略对冲基金，不同团队会被刻意安排相互竞争以争夺最终交易模型的权重。但在Rentec，由于采用单一模型架构，所有人都为相同的投资策略和基础设施协同工作。这意味着每个研究团队成员、基础设施团队成员都能看到彼此的工作成果，访问完整模型。

Then you've got multistrategy hedge funds out there where literally firms are being pitted against one another to be weighted in the ultimate trading model for the firm. Yep. At Rentec, though, because of the one model architecture, everyone works together on the same investment strategy and the same investment infrastructure. That means everyone sees everybody else's work. Everybody who works at RENTECH on the research team, on the infrastructure team, they have access to the whole model.

这在其他任何地方都不可能实现。

That's not true anywhere else.

是的，这个观点很好。整个代码库都是完全可见的。

Yeah. That's a good point. The whole code base is completely visible.

这也意味着，因为它只是一个模型、一种策略，当其他人提升该模型的性能时，对你的影响和对他们的影响一样大。这与任何其他对冲基金都截然不同。

And that also means because it's just one model, just one strategy, when somebody else improves that model's performance, that directly impacts you as much as it impacts them. This is really different than any other hedge fund out there.

那么，这与我把自己的一部分薪酬投入到我所工作的多策略对冲基金有什么不同？难道我不喜欢其他团队也创造高绩效吗？

So why is that different than if I roll some of my compensation into a multi strategy hedge fund that I work at? Don't I love other teams creating high performance also?

当然。但你不像喜欢自己团队那样喜欢其他团队，因为无论是薪酬还是职业发展，你更依赖于自己的表现而非他人的表现。

Sure. But you don't love it as much as your team because either compensation or career wise, you are much more dependent on your performance than you are other people's performance.

哦，是的。这是件大事。在大多数情况下，你打算在大多数地方做完这份工作后还有下一份工作。所以你在乎功劳，在乎打破彩罐后去别处，或者建立声誉后去别处。Rendtech的大多数人不会有另一份工作。

Oh, yes. This is a big thing. You intend to have a job after that job at most places most of the time. So you care about credit, and you care about smashing the pinata and then going elsewhere or building reputation and then going elsewhere. Most of the people at Rendtech are not gonna have another job.

你在LinkedIn上发现了什么？至少员工的平均任职时间是，大概十六年？

What did you find on LinkedIn? At least the median tenure of employees is, like, sixteen years?

是的。我刚买了LinkedIn高级会员，可以看到平均任职时间。这太疯狂了。Renaissance只有大概三四百名员工。而平均任职时间，至少根据LinkedIn的报告，大概是十四年。

Yeah. I just got LinkedIn premium, and you can see median tenure. And it's crazy. There's only, like, three, four hundred employees at Renaissance. And the median tenure, at least as reported by LinkedIn, is, like, fourteen years.

是的。好的。这让我想到第二点，你刚才提到，这是一个极其精简的团队。Rentec的员工总数不到400人，其中仅有一半从事研究和工程工作，另一半则是后台或开放式基金机构销售。所以我们估算，大概最多150到200人是直接参与Medallion基金运作的核心人员。

Yes. Okay. This brings me to point number two, which you said, this is an absurdly small team. There are less than 400 employees that work at Rentec, only half of which work in research and engineering, and the other half are either back office or institutional sales for the open funds. So let's call it, I don't know, a 150, 200 people max who are, like, hands on the wheel here for Medallion.

没错。Rentec的所有同业公司——比如Citadel、D.E.Shaw、Two Sigma等，把Jane Street也算上，再加上那些高频交易公司——这些地方最少都有2000到5000名员工。

Yep. Every other peer firm of Rentec, you know, Citadel, D Shaw, Two Sigma, etcetera, all of them, you lump Jane Street, you know, jump the high frequency guys in here. Minimum, two to 5,000 people work at those places.

哇，我之前没想到规模差距这么大。

Wow. I didn't realize it was that big.

其他公司的人力规模比Rentec整整高出一个数量级。

It is an order of magnitude more people who are working at the other firms versus who are working at Rentec.

除非你认为这是资本规模导致的差异——其实不是。机构基金规模已经很大了，巅峰时管理规模超过1000亿，现在除了Medallion基金的100-150亿之外，他们还管理着600-700亿资产。

Unless you think that that's like a capital based thing, no. The institutional funds have gotten big. They peaked at over a 100,000,000,000, but they're currently between 60 and 70,000,000,000 that they manage on top of the 10 or 15 that's in the Medallion fund.

对。所以管理资产规模（AUM）其实和那些大基金相当。这带来诸多优势：首先是类似爱马仕工坊式的协作优势。

Yeah. So AUM is, like, the same Yep. As these big funds. This has all sorts of benefits. Number one, there's, like, the Hermes Atelier workshop benefit.

所有人都能叫出彼此的名字。你了解同事的孩子，熟悉同事的家庭情况。

Everyone knows each other by name. You know your colleague's kids. You know your colleague's families.

是的。他们直接在官网上写明，有90位数学、物理、计算机科学及相关领域的博士。关于页面列出了10条看似随机的要点，这就是其中之一。

Yep. They put right on their website, there are 90 PhDs in mathematics, physics, computer science, and related fields. The about page has these 10 kinda random bullet points, and that's one of them.

没错。还有与此相关的另一面。这家公司位于长岛一个偏僻的地方。你确实会认识同事们的家人和孩子，因为你不会去纽约市和Two Sigma的人喝酒交际。你也不会和别人比较笔记或暗中较劲。

Yes. Then there's the related aspect to all this. The firm is in the middle of nowhere on Long Island. You actually know your colleagues, families, and kids because you're not going out and getting drinks with someone from two Sigma in New York City. You're not comparing notes or measuring parts of your anatomy with someone else.

你更像是，在游泳池边消磨时光。

You're, like, hanging out at the swimming pool.

完全正确。而且由于文艺复兴公司不从金融行业招聘，你不太可能认识其他金融从业者。你来自科学相关领域，现在在东塞托基特工作——那个长岛小镇大概只有一万甚至更少的居民。所以你生活在这个小地方。

Totally. And since Renaissance doesn't recruit from finance jobs, it's kind of unlikely that you know someone else in finance. You came out of a science related field. You now work in East Setauket, Long Island, which has it's like 10,000 people or something or less that live there. So you're in this little town.

你其实不常去纽约市。就算去，也不是为了和其他金融人士喝酒应酬。所以即便没有长达数页的竞业禁止协议和终身保密协议，你也很难进入那些社交圈子。

You're not actually going into the city that often. And if you are, it's, again, not to grab drinks with other finance people. So even if you didn't have a many page noncompete and a lifetime NDA, you're very unlikely to be in the social circles.

你根本接触不到那些人。没错。而且RenTec招聘的是资深科学家和博士，不像Jane Street或桥水那样招收本科应届生。我感觉那里像个没有学生的大学校园。

You're just not getting exposed. Exactly. And Rentec's hiring established scientists and PhDs. They're not hiring kids out of undergrad like Jane Street or Bridgewater is. My sense is that the place is like a college campus without any students.

你看过网上的照片吗？谷歌搜索文艺复兴科技公司，查看园区照片——那是个被树林环绕的小庭院，有蜿蜒的小径和网球场。

Have you seen the pictures online? Yeah. If you look up Renaissance Technologies at Google, and you go and look at the photos on campus, it's a little courtyard and winding walking path and woods all around it, tennis courts.

是的。接下来是小团队要素的最后一部分，就是财务影响的规模，我认为这并不准确。但假设有另一家量化基金创造了与Rendtech相同的绩效回报金额。在Rendtech，这笔钱由几百人分配；而在Citadel，则由五千人分摊。

Yep. So then there's the last piece of the small team element, which is just the magnitude of the financial impact, which I don't think is true. But let's say that there were another quant fund that made the same number of dollars of performance returns that Rendtech does. At Rendtech, you're splitting that a couple 100 ways. At Citadel, you're splitting that 5,000 ways.

去其他地方根本说不通。

It just doesn't make sense to go anywhere else.

我们为这期节目做准备时与人聊天，对方告诉我们，你永远无法与他们竞争，但他们会付给你足够的钱让你不想竞争。

We were chatting with someone to prep for this episode, and they told us, you can't ever compete with them, but they'll pay you enough that you won't want to.

没错。好吧，这就要说到我一直暗示让我超级兴奋的事了。我认为Rentec如此独特且难以复制的第三个关键因素，就是Medallion基金本身的结构——这是一支管理费5%、附带权益44%的有限合伙制基金。

Yes. Okay. So this brings me to what I've been kinda teasing that I'm super excited about. I think the third puzzle piece of what makes Rentec so unique and defensible is Medallion's structure itself. That it is a LPGP fund with 5% management fee and 44% carry.

所以它不像自营交易机构或专属资金池，就是单一资金池。尽管普通合伙人和有限合伙人其实是同一批人，但它确实是标准的GP/LP结构。

So it's not like a prop shop or, like, proprietary it's just one pot of money. It's literally a GP LP even though the GPs and the LPs are the same people.

我是这么理解的——虽然不清楚实际架构，但在研究过程中，这个离谱的44%附带权益始终让我觉得不对劲。因为我不断自问，

So here's my thinking on this. Now I don't know how it is actually structured, but there was something about this whole crazy 44% carry that just wasn't sitting with me right throughout the research. Because I kept asking myself,

为什么？他们已经清退了大部分甚至所有有限合伙人。那为什么还要提高附带权益？

why? Right. They've already kicked out most of the LPs, if not all. So why are they raising the carry? Right.

都是他们自己人。

It's all themselves.

全是内部人士。为什么他们要向自己收取44%的收益分成和5%的管理费？我记得吉姆讨论过这个。他们会说'哦，我和其他人一样支付这些费用'。

It's all insiders. Why do they charge themselves 44% carry and 5% management fees? I think Jim talks about this. They'll oh, I pay the fees just like everybody else.

是啊。这总是个滑稽的论点。就像在问，你这些费用到底付给谁了？

Yes. It's always a funny argument. It's like, who are you paying the fees to?

没错。所以我当时就想，这到底是怎么回事？好吧，这是我的假设：这与设置离谱的绩效费无关。

Right. So I was like, what is happening here? So okay. Here's my hypothesis. This is not about having crazy performance fees.

这也不是为了收取业内最高的收益分成。这是公司内部的一种价值转移机制——从老员工群体向当年负责大奖章基金的现役员工转移。我是这么理解的：新人加入复兴公司时，财富显然远少于老员工，这部分差距既来自每年工作获得的直接回报，也来自你在大奖章基金中的投资份额——顺便说一句，我记得他们好像把纽约州或联邦政府告上法庭，才让401k计划能投资大奖章基金？

This is not about having the highest carry in the industry. This is a value transfer mechanism within the firm from the tenure base to the current people who are working on Medallion in any given year. So here's how I think it works. When people come into Rentec, they obviously have way less wealth than the people who've been there for a long time, both from the direct returns that you're getting every year from working there and just your investment percentage of the Medallion fund, which, by the way, I think they took it was either the state of New York or the federal government to court to be able to have the four zero one k plan at Rentec be the medallion fund?

不可能吧。

No way.

真的。如果你在那里工作，你的401k就是大奖章基金。

Yeah. Yeah. So, like, if you work there, your four zero one k is the medallion fund.

这太疯狂了。所以真的用不了几年，你就能一辈子衣食无忧了。

That's crazy. So it really doesn't take more than a few years before you're set for life.

完全同意。我是说，根据你对'衣食无忧'的定义，我认为这发生得非常非常快。是的。好的。既然如此，你如何避免一群有才华的年轻人分出去创办自己的奖章基金呢？

Totally. I mean, depending on your definition of set for life, I think it happens very, very quickly. Yep. Okay. So given that though, how do you avoid the incentive for a group of talented younger folks to split off and go start their own medallion fund?

没错。尤其是当他们都能接触到整个代码库的时候。整个体系设计得像大学数学系一样，大家不断分享知识，因为我们随时共享所有知识才能产出更好的同行评议研究。你会觉得把全部权限交给每个人风险极高。对吧。

Right. Especially when they all have access to the whole code base. The whole thing is meant to function like a university math department where everyone's constantly knowledge sharing because we're gonna create better peer reviewed research when we all share all the knowledge all the time. You would think that's a super risky thing to give everyone all the keys. Right.

所以我认为是44%的收益分成结构起了作用。因为本质上，你每年有5%的管理费，先扣除这5%，然后是44%的业绩提成。假设奖章基金每年大约翻倍，我们四舍五入相加，每年49%的经济收益归现有团队，51%归资深成员。我当时就想，这相当于什么制度？

So I think it's the 44% carry structure that does it. Because, basically, what you're saying is every year, 5% management fee, so 5% off the top, and then 44 of performance. So let's say Medallion is on the order of, call it, doubling every year. Let's round that up and just add them and say, 49% of the economic returns in any given year go to the current team, and 51% of the economic returns go to the tenure base. I was like, what is the equivalent here?

我觉得有点像学术界的终身教职制度。你在公司资历越深，你的收益分配就越偏向有限合伙人那边。有意思的是，你在公司越年轻，分配就越偏向普通合伙人这边。但归根结底是五十一对四十九的比例。

I think it's kinda like academic tenure kind of thing. The longer tenure you are at the firm, the more your balance shifts to the LP side of things. Interesting. The younger you are at the firm, the more your balance is on the GP side of things. But at the end of the day, it's fifty one forty nine.

所以这个天然的收益转移机制，能让当年工作的员工保持超高积极性。而你待得越久，就相当于在花钱雇佣年轻同事为你工作。

So there's this very natural value transfer mechanism to keep the people that are working in any given year super incentivized. And as you stay there longer, you are paying your younger colleagues to work for you.

没错。有意思。我觉得这个洞察很好——它的结构确实像大学院系的终身教职制度。

Right. It's funny. I think it's a good insight that it's structured like a university department tenure.

嗯，我一直在问自己为什么。为什么？如果没有外部有限合伙人，他们为什么要有这个？这是我能想到的最好解释。实际上我觉得这挺天才的。

Well, I just kept asking myself why. Why? Why do they have this if there's no outside LPs? And this was the best thing I could come up with. And I actually think it's kinda genius.

是啊。这比'全是某个人的钱，每年决定给现有团队发奖金，只给他们足够的钱确保留住人'要优雅多了。

Yeah. It's more elegant than it's all one person's money, and they're deciding to bonus out the current team every year and just give them enough money to make sure you retain them.

对。我认为大多数自营交易公司都是这样运作的。比如Jane Street主要是自营交易，我觉得主要是老板的钱。但那是静态情况。不像...你知道，如果真是那样，吉姆就会永远拥有这东西。

Right. Which is how I think most prop shops work. Like, Jane Street is mostly a prop shop. I think it is mostly the principal's money, But that's a static situation. It's not like, you know, if that were true, then Jim would just own this thing forever.

而我认为在Rentec不是这样。

And I don't think that's true here at Rentec.

没错。所以本质上，大卫，真正的魔力在于他们有一个基金。是常青基金。你刚加入公司时只能拿到分成部分。但随时间推移，你会成为公司的重要投资人，逐渐过渡到那51%的份额。

Yeah. So essentially, David, the real magic is they've got one fund. It's Evergreen. And when you start at the firm, you're only getting sort of paid the carry amount. But over time, you become a meaningful investor in the firm, and you sort of shift to that 51%.

你某种程度上成了有限合伙人。最终你会完全退出，只作为有限合伙人存在。所以你说得对，这是一种价值转移机制，从老将到新将，方式清晰明确，可能还有税收优惠，比'我是老板随便给大家发奖金'强多了。

You're kind of the LP. And then over time, you eventually graduate out entirely, and you're only an LP. And so you're right. It's a value transfer mechanism from the old guard to the new guard in a way that is clear, well understood, probably tax advantaged versus just doing, I'm the owner and I'm giving everyone arbitrary bonuses.

是的。归根结底，我认为这三点构成了RENTECH这幅织锦的核心：所有人共同协作的一个模型；超级精简的团队，彼此熟识，每个人对那个模型的财务贡献都会惠及所有人；第三就是这个LP-GP模型，附带超高绩效分成，为 incoming 的新人才和 outgoing 的老将都创造了正确的激励机制。

Yep. And at the end of the day, I think these three pieces to me are the core of this sort of tapestry of RENTECH. One model that everybody collaborates on together, a super small team where we all know each other and the financial impact that any of us make to that one model is great to all of us. And three, this LPGP model with very high carry performance fees that creates the right set of incentives both for new talent on the way in and old talent on the way out.

是的，我认为没错。好了，听众朋友们，现在正是感谢我们Acquired新合作伙伴Sentry的好时机。Sentry拼写为s-e-n-t-r-y，就像站岗放哨的人。

Yep. I think that's right. Alright, listeners. This is a great time to thank a new partner of ours here at Acquired, Sentry. That's s e n t r y, like someone standing guard.

没错。Sentry帮助开发者调试错误和延迟问题，几乎涵盖所有软件问题，并在用户感到不满前修复它们。正如其官网所言，它被超过400万软件开发者认为‘还不错’。

Yes. Sentry helps developers debug errors and latency issues, pretty much any software problem, and fix them before users get mad. As their homepage puts it, it's considered, quote unquote, not bad by over 4,000,000 software developers.

今天我们要讨论的是Sentry如何与Acquired生态中的另一家公司Anthropic合作。Anthropic原本使用较旧的基础设施监控方案，但在其庞大的规模和复杂度下，他们转而采用Sentry来更快发现和解决问题。

So today, we're talking about the way that Sentry works with another company in the acquired universe, Anthropic. Anthropic used to have some older infrastructure monitoring that was in place, but at their massive scale and complexity, they instead adopted Sentry to help them find and fix issues faster.

确实。在AI领域，崩溃可能造成巨大问题。如果你正在运行像模型训练这样的大型计算任务，一个节点故障可能影响数百甚至数千台服务器。Sentry帮助他们检测故障硬件，从而在引发连锁问题前快速剔除。Sentry让他们能在几小时而非几天内调试重大问题，尽快恢复训练任务。

Yep. Crashes can be a massive problem in AI. If you're running a huge compute job like training a model and one node fails, it can affect hundreds or thousands of servers. Sentry helped them detect bad hardware so they could quickly reject it before causing a cascading problem. Sentry enabled them to debug massive issues in hours instead of days so they could get back to their training runs.

如今Anthropic依赖Sentry实时跟踪异常、分配错误并分析故障，覆盖其研究团队使用的所有主要语言，包括Python、Rust和C++。据Anthropic团队表示，Sentry为开发者提供了调试问题所需全部信息的统一平台。

And today, Anthropic relies on Sentry to track exceptions, assign errors, and analyze failures in real time across all the primary languages used by Anthropic's research teams, including Python, Rust, and c plus plus According to the Anthropic team, Sentry gives our developers one place where they have all the information they need to debug an issue.

Sentry领域的另一个有趣进展是，本月起他们推出了名为SEER的AI调试器。SEER是一个AI代理，它能利用Sentry的问题上下文和代码库，不仅猜测问题根源，还能针对具体应用提出可直接合并的修复方案。

And one other fun update in the world of Sentry is that as of this month, Sentry now has an AI debugger called SEER. SEER is an AI agent that taps into all the issue context from Sentry and your code base to not just guess, but root cause gnarly issues and propose merge ready fixes specific to your application.

我们非常兴奋能与Sentry合作。他们拥有令人惊叹的客户名单，不仅包括Anthropic，还有Cursor、Vercel、Linear等。如果你想像超过13万家组织那样快速修复故障代码——从独立开发者到世界顶级企业——可以访问sentry.io/acquired了解更多。他们为所有Acquired听众提供两个月免费试用，只需告诉他们是本和大卫推荐来的。

We are pumped to be working with Sentry. They've got an incredible customer list, including not only Anthropic, but Cursor, Vercel, Linear, and more. If you wanna fix broken code like the over 130,000 organizations using Sentry from indie hobbyists to some of the biggest companies in the world to find and fix broken code fast. You can check out sentry.i0/acquired to learn more, and they are offering two free months to all Acquired listeners. That's Sentry, sentry,.i0/acquired, and just tell them that Ben and David sent you.

好的。这个故事还有几个部分我们之前跳过了，因为没有合适的地方插入，但这些确实是值得了解且客观上有趣的历史事件。第一个叫做篮子期权。时间是2002年2月。RENTECH公司十三年来都清楚自己基本上拥有一台印钞机。

Okay. There's a few other parts of the story that we skipped along the way because there was no real good place to put them in, but these are objectively fascinating historical events that are totally worth knowing about. And the first one is called basket options. So the year is 02/2002. RENTECH has thirteen years of knowing that they basically have a machine that prints money.

那么当你拥有一台印钞机时该怎么做？加杠杆。当然这类公司会受到各种限制，规定它们能承担多少杠杆。你不能直接说我要借入相当于股本100倍的资金。所以需要巧妙地从银行或其他贷款方借入大量资金来放大收益。

So what should you do when you have a machine that prints money? Leverage. Now there are all sorts of restrictions around firms like this and how much leverage they can take on. You can't just go and say, I'm gonna borrow, you know, a $100 for every dollar of equity capital that I have in here. So you need to sort of get clever to borrow a whole bunch of money from banks or from any lender to basically juice your returns.

如果你拥有一台可靠的印钞机——大多数人都没有，大多数人也不应该加杠杆，因为他们搞砸的概率和成功的概率一样高。关于篮子期权，我要直接引用那位破解市场的人——格雷格·扎克曼的完美描述：'篮子期权是一种金融工具，其价值与特定股票组合的表现挂钩'。

If, again, you have a money printing machine that's reliable, most people don't. Most people probably shouldn't take leverage because they're just as likely to blow the whole thing up as they are to be successful. So basket options. I am gonna read directly from the man who solved the market because Greg Zuckerman just put it perfectly. Basket options are financial instruments whose values are pegged to the performance of a specific basket of stocks.

虽然大多数期权基于个股或金融工具，但篮子期权与一组股票挂钩。如果这些标的股票上涨，期权价值就会上升。这就像实际持有股票却无需真正持有。事实上，提供资金的银行才是篮子里股票的法定所有者，但从实际效果看，这些股票就是奖章基金的财产。

While most options are based on an individual stock or a financial instrument, basket options are linked to a group of shares. If these underlying stocks rise, the value of the option goes up. It's like owning the shares without actually doing so. Indeed, the banks who, of course, loaned the money, who put the money in the basket option, were legal owners of the shares in the basket. But for all intents and purposes, they were Medallion's property.

这招非常聪明。奖章基金表示：'我们加杠杆的方式是通过一个股票篮子。我们拥有购买这个篮子的期权。篮子里大部分资金实际来自银行，但银行雇佣我们交易篮子里的期权。一年后当长期资本利得税生效时，我们有权购买这个篮子。'

So this is very clever. Medallion's saying, well, the way we're gonna lever up is there's a basket. We have an option to purchase that basket. Most of the capital in that basket is actually the bank's capital, but the bank has hired us to trade the options in the basket. And then after a year, when long term capital gains tax kicks in, we have the option to buy that basket.

总之，奖章基金的电脑全天候向银行发送自动化指令，有时每分钟甚至每秒都在操作。这些期权使其能借入远超正常允许范围的资金。竞争对手通常每1美元现金对应7美元金融工具，而奖章基金的期权策略让每1美元现金能控制12.5美元的金融工具，更容易碾压对手——前提是他们能持续找到盈利交易。当发现特别诱人的机会时，杠杆率可提升至每1美元现金对应近20美元资产。

So, anyway, all day, Medallion's computer sent automated instructions to the banks, sometimes in order a minute or even a second. The options gave Medallion the ability to borrow significantly more than it otherwise would be allowed to. Competitors generally had about $7 of financial instruments for every dollar of cash. By contrast, Medallion's option strategy allowed it to have $12.50 worth of financial instruments for every dollar of cash, making it easier to trounce rivals assuming they could keep finding profitable trades. When Medallion spied an especially juicy opportunity, it could boost leverage holding close to $20 of asset for every dollar of cash.

2002年2月，奖章基金管理着50多亿美元，却控制着超过600亿美元的投资头寸。大卫，这揭示了我们本期节目尚未提及的一点：他们不仅能找到50亿美元规模的盈利交易，更想把这50亿美元杠杆放大到极致，从而操作600亿美元的盈利交易。而篮子期权为他们提供了一种合法途径，让他们能以自认安全的方式获得惊人的杠杆率。

In 02/2002, Medallion managed over 5,000,000,000, but it controlled over $60,000,000,000 of investment positions. David, this exposes something we haven't shared yet on the episode, which is it's not just that they could find $5,000,000,000 worth of profitable trades. It's that they wanted to lever the crap out of $5,000,000,000 and find $60,000,000,000 of profitable trades to make. And basket options gave them a legal way to have an incredible amount of leverage in a way that they felt safe about.

是的。如果采用这种策略，无杠杆回报会低得多。

Yeah. The unlevered returns, if you were running this strategy, would be much lower.

没错。所以我们这个策略手册中没谈到的一个重要部分就是杠杆。但每个量化基金都在使用杠杆，而文艺复兴公司只是比其他人更聪明。

Yep. So a big piece of this playbook that we didn't talk about is leverage. But every quant fund does leverage, and so Renaissance was just more clever than everyone else.

是的。不过这一点很重要。在我们收购的每10家公司中，有9家与杠杆完全无关。对吧。对我们这些来自科技和风投领域的人来说，杠杆就像个忌讳词。

Yep. It's an important point, though. Nine out of every 10 companies that we cover on acquired, leverage is zero part of the story. Right. And for us coming from the world we come from in tech and venture capital, leverage is like a dirty word.

比如，我就很害怕它。

Like, I'm scared of it.

对。你可以想象，假设他们不是50.25%的时候正确，而是50.001%的时候正确。他们就需要进行大量交易才能产生足够利润。所以这就是为什么需要600亿美元现金来执行这个策略，以实现他们期望的回报。对，基于50亿美元的股本。

Right. I mean, you could imagine let's say it wasn't they were right 50.25% of the time, but they were right 50 o o o 1% of the time. They would need to do a ton of trades in order to generate enough profits. So that's why you need, you know, $60,000,000,000 of cash to actually execute the strategy to produce the returns that they were looking for Yep. On $5,000,000,000 of equity.

总之，这还有第二部分——他们通过这种方式获得大量杠杆固然很好，但这只是一部分。另一部分是他们认为这是非常节税的载体。他们报税的方式是说：'当然，篮子里有这些东西。'

Anyway, there's a second chapter to this, which is it's all well and good that this is how they get a bunch of leverage. That's one piece of it. The other piece is they thought this was a remarkably tax efficient vehicle. The way that they were filing their taxes said, oh, sure. There's stuff in that basket.

但实际上我们拥有的是购买或出售那个篮子的期权。我们大约每13个月才行使一次。具体数字我不清楚，但大概是一年多。因此，我们买入后持有大约一年时间。

But the thing that we actually own is an option to buy that basket or sell that basket. And we only exercise that once every thirteen months or so. I don't know the exact number, but something like that over a year. And so, therefore, we're buying something. We're holding it for a year.

我们正在出售它。当然，篮子里有数百万笔交易在进行，但我们并不拥有那个篮子。银行才是所有者，我们只是提供咨询。你大概能明白其中的逻辑。

We're selling it. Oh, of course, there's millions and millions of trades going on inside the basket, but we don't own that basket. The banks do. We're just advising them. You can kinda see the logic here.

随着时间的推移，最终在2021年，美国国税局表示：不行。是你们进行了所有这些交易，那并非完全独立的实体。因此你们欠缴了68亿美元的税款，现在需要连本带利加上罚款一并补交。

Over time, eventually, in 2021, the IRS said, no. You made all those trades. That was not a completely separate entity. And so you guys owed $6,800,000,000 in taxes that you didn't pay. You're gonna need to pay that with interest, with penalties.

顺便说一句，吉姆·西蒙斯，我们希望你和另外几位合伙人真正承担这笔债务的重担，而他们确实承担了。仅西蒙斯一人就向国税局补缴了6.7亿美元的税款，因为这个篮子期权策略最终未被认定为长期资本利得。好了，现在说说今天的业务数据，然后我们将深入探讨权力与策略手册。

And by the way, Jim Simons, we're gonna want you and the other few partners to really bear the load of that, and they did. So for Simons alone, he paid $670,000,000 to the IRS in back taxes for this basket option strategy that turned out not to be a long term capital gain. Yep. Alright. So numbers on the business today, and then we will dive into power and playbook.

今天我们讨论过奖章基金，规模在100到150亿美元之间（不同来源说法不一），历史上更接近50到100亿美元。机构基金规模约600到700亿美元，曾一度达到1000亿美元。大卫你提到总收益是600亿美元。据《福布斯》估计，仅吉姆·西蒙斯目前身价约300亿美元，这与历年其他统计数据基本吻合——他持有文艺复兴约半数股份。

So today, we've talked about Medallion, 10 or 15,000,000,000, depending on who you ask. Historically, it was more like 5 or 10,000,000,000. The institutional fund is about 60 to 70,000,000,000, and that's one point was a 100,000,000,000. The total carry generated, David, you said is $60,000,000,000. Forbes estimates that Jim Simons alone is worth about $30,000,000,000 today, which kinda pencils with a bunch of other stats over the years that he owned about half of Renaissance.

收益率方面，奖章基金从1988到2020年扣除费用后年化约66%，净收益约39%——这相当惊人。我做了个假设性测算：你认为文艺复兴公司每年营收多少？机构基金按600亿美元资产的10%计算，管理费收入6亿美元，绩效费6亿美元，合计12亿美元年收入。

The returns, obviously, the Medallion fund generated approximately 66% annualized from 1988 to 2020 after those fees was about 39% wild. So an interesting thing to understand, I ran a hypothetical scenario of how much money do you think Renaissance the business makes a year in revenue? And so the institutional fund, let's call it 10% on 60,000,000,000 of assets. So that's 600,000,000 from fees and 600,000,000 from performance. So 1,200,000,000.0 a year in revenue to the firm from the institutional side of the business.

因为我总在思考：这真的重要吗？他们费尽心思建立机构业务部门。但谁在乎呢？假设奖章基金150亿美元规模实现平均66%总收益，管理费7.5亿美元，绩效费43亿美元。

Because I always ask myself the question, does that actually matter? They did all this work to stand up the institutional side. Who cares? Well, let's say Medallion does their average 66% gross on 15,000,000,000. That is 750,000,000 in fees and 4,300,000,000.0 on performance.

因此奖章基金贡献50亿美元，机构业务贡献12亿美元。当然，员工本身就是奖章基金的投资人，可以说这样分割没有意义。但无论如何，这是个年营收708到900亿美元的业务。

So a total of 5,000,000,000 from Medallion, and 1,200,000,000.0 from the institutional side of the business. Now, of course, the employees are the investors in Medallion, so you could just argue it's actually silly to cut them up. But I don't know. It's a $7.08, $9,000,000,000 revenue business.

没错。因为这还没有算上Medallion基金的LP回报部分。

Right. Because that's not including the LP return on Medallion.

百分百同意。确实没包括。

A 100%. It's not.

就像我们之前讨论很久的，归根结底这都是同一回事。

Which, again, as we spend a long time talking about, it's all the same thing.

是的。但和其他公司对比会很有趣，你可以把这个数字记在脑子里——这可是年收入780亿美元的生意。

Yes. But it's kind of interesting just to compare it against other companies to have this in the back of your head. This is a $78,000,000,000 a year revenue business.

我认为他们在基础设施方面有很多支出。

Now I think there are a lot of expenses on the infrastructure side.

完全正确。这正是我想讨论的另一点。就拿Medallion基金来说，他们收取7.5亿美元管理费，但年支出肯定远不到7.5亿，天知道他们运行的是什么基础设施，某种超级计算集群吧。

Totally. That was another thing I wanted to talk about. The fact that they do, let's say, Medallion alone. So they have $750,000,000 in fees. I don't think they come close to $750,000,000 a year in expenses, but they are running who knows what infrastructure, some kind of supercomputing cluster.

运营一个亚马逊数据中心要花多少钱？我觉得规模应该小得多。

What does it cost to run one Amazon data center? I mean, it's, I think, much smaller scale.

我不知道。我是说，你这里提到的数据量太大了。

I don't know. I mean, you're talking about a lot of data here.

没错。他们官网上明确写着，拥有5万个计算机核心，150吉比特每秒的全球连接速度，研究数据库每天增长超过40太字节。这数据量确实惊人。

Yeah. It says right on their website, they have 50,000 computer cores with a 150 gigabits per second of global connectivity and a research database that grows by more than 40 terabytes a day. That's a lot of data.

对。那一年是7.5亿吗？我不确定，但肯定不是零。

Right. Is that 750,000,000 a year? I don't know, but it's not zero.

我不这么认为。他们肯定没在手续费上亏钱，但这业务确实有实实在在的硬成本。没错。

I don't think so. They're certainly not losing money on the fees, but there are actual hard costs to this business. Right.

我也在想，Medallion的手续费部分是否基本覆盖了现有团队的基本工资。

I wonder too if the fee element of Medallion basically pays the base salaries for the current team.

听起来挺合理。如果你曾参与过数据中心建设，或者有办法推算出Medallion在计算、数据和网络方面的运营成本，我们很乐意听取你的意见。欢迎联系Acquired.fm。好的。

That feels like it's right. If you're, someone who has done a data center build out before or has any way to sort of back into what the costs of Medallion's operating expenses are on the compute and data and network side, we would love to hear from you. Hello at Acquired. Fm. Okay.

电力？

Power?

权力。这个话题很有趣。

Power. This is a fun one.

没错。新听众们注意，这是汉密尔顿·赫尔默在《七种权力》一书中提出的框架。是什么让企业能持续获得差异化回报，长期比最接近的竞争对手更盈利？这七种权力分别是：反定位、规模经济、转换成本、网络效应、流程优势、品牌效应和垄断资源。大卫，我想在这个环节首先问你关于Rentec终身竞业禁止协议的问题。

Yeah. So listeners who are new to the show, this is Hamilton Helmer's framework from the book seven powers. What is it that enables a business to achieve persistent differential returns to be more profitable than their closest competitor on a sustainable basis? And the seven are counter positioning, scale economies, switching costs, network economies, process power, branding, and cornered resource. And, David, my question to you to open this section is specifically about Rentec's lifelong non competes.

这似乎是他们保持竞争优势的重要原因。我很好奇你是否同意这个观点，你会把它归为哪一类？

That feels like a big reason that they maintain their competitive advantage. And I'm curious if you agree with that, what would you put that under?

嗯，我认为是终身保密协议和竞业禁止条款——只要纽约州法律允许的范围，但并非真正终身。我听过不同说法，六年、五年左右。

Well, I think it's lifelong NDAs and noncompetes as long as the state of New York legally allows for, but that is not lifetime. I've heard various figures, six years, five years, something like that.

是的。

Yep.

说到底，竞业禁止更像是看哪一方更愿意对簿公堂。

I mean, at the end of the day, non competes are more like, what is one side willing to go to court over?

对。

Right.

但现实是，人们不会离开。人们就是不会离开。尤其是不会离开去创办自己的公司。是的。我半夜想到这个，我认为RENTECH的有效竞业禁止有三个层面。

But the reality is people don't leave. People don't leave, period. And people especially don't leave and start their own firms. Yep. I was thinking about this in the middle of the night, and I think there's three layers to the effective noncompete that happens with RENTECH.

首先是法律层面，也就是你提到的基础层面。就像你签署的那些协议。然后是经济层面，我们在Tapestry讨论了很久，离开简直是愚蠢的。留在那个小团队里比去人更多的Sigma对你更有利。是的。

There's the legal layer, the base layer that you're talking about. It's like the agreements you sign. Then there's the economic layer of what we spent a long time talking about in Tapestry of it would just be dumb to leave. You are better off staying there as part of that team with a smaller number of people than going to Sigma with a lot more people. Yep.

我认为那是下一层面，而最高层面可能就是社交层面。你和世界上最聪明的人在一起，在一种学院式的氛围中，你们都在努力做一些对你有直接影响的事情。没错。那是你的社区。那是你的社区。

I think that's the next level of and then I think the highest level is just probably the social layer. You're there with the smartest people in the world in a collegial atmosphere where you're all working hard on something that has direct impact on you. Right. It's your community. It's your community.

完全正确。你不在纽约市。你不在汉普顿。你不在硅谷。你是主动选择了那里。

Totally. You're not in New York City. You're not in the Hamptons. You're not in Silicon Valley. You are selecting into that.

我想如果这就是你想要的，他们会说，世界上还有比这更好的地方吗？

And I think if that's what you want, they're like, what better place in the world?

好的。那么归类一下。这属于哪种权力？

Alright. So classify it. What power does that fall under?

嗯，我是说，具体的人你可能会归入垄断资源，但我不确定这完全能概括这里的情况。我更多想到的是流程权力，因为我认为这是人员、模式和激励结构的结合。

Well, I mean, I think the people specifically you would put into cornered resource, but I'm not actually sure that fully captures it here. I was thinking more process power, because I think it is the combination of the people and the model and the incentive structures.

是的，我认为没错。我最大的优势也是流程处理能力。你确实可以深入了解系统运作方式，并围绕其构建难以复制的流程。我认为这些系统也是随时间层层累积的，近五年加入公司的人都不清楚其完整运作过程。

Yep. I think that's right. I also had my biggest one being process power. You actually can develop intricate knowledge of how a system works and then build processes around that that are hard to replicate elsewhere. I think these systems have been layered over time also, where anyone who's come into the firm in the last five years doesn't know how it works start to finish.

我没让人核实过，但代码量超过1000万行。系统执行交易时的复杂度、交易内容、原因及所需速度，我认为没人能在脑中完整掌握整个模型。正因积累了三十多年的复杂度，才形成了这种流程处理能力。

I didn't ask anyone to verify that, but it's over 10,000,000 lines of code. And the level of complexity of the system of when it's putting on trades, what trades it's putting on, why, the speed at which they need to happen, I actually don't think anyone holds the whole model in their head. And so I think there's process power just because it's thirty plus years of complexity that's been built up.

没错，我完全同意，尤其是关于模型本身。或许可以说这个模型是垄断性资源。

Yep. I totally agree with that, particularly in the model itself. I mean, maybe you could argue the model is a cornered resource.

我要说的是数据

I am going to argue that the data

哦，好吧。

Oh, okay.

才是垄断性资源。我不确定模型是否算，也许吧。某种程度上，了解这1000万行代码的功能就等于掌握了模型。

Is a cornered resource. I don't know for sure about the model. Maybe. I mean, I guess that's the same thing as saying the knowledge of what the 10,000,000 lines of code does. That's the model.

但我认为真正关键是他们拥有干净的数据，并建立了相关系统——让全球顶尖的博士们研究数据清洗。这工作不吸引人，但他们拥有的历史市场数据宝库，其格式之优无人能及。这才是真正的垄断资源。

But I actually think the fact that they have clean data and they've been creating systems like, have the best PhDs in the world thinking about data cleaning. That's not a sexy job. And yet, they have probably the treasure trove of historical market data in the best format that nobody else has. That's an actual cornered resource.

我对这个问题有几点细微的看法。首先，我认为他们确实拥有比其他公司更好的数据，这要归功于桑多尔·斯特劳斯以及他在八十年代就开始的工作，那时其他人还没涉足这一领域。是的，他们拥有这些优势，而其他公司没有。不过，其他量化公司肯定也在投入大量资源追赶。

I have a couple nuances on this. So one, I think it probably is true that they have better data than any other firm, thanks to Sandor Strauss and the work that he started doing in the eighties before anybody else was really doing this. Yep. So they have that, and other firms don't. That said, certainly, all the other quant firms are throwing untold resources at all this too.

没错。他们想做这件事，资金不是问题。

Right. They wanna do this, and money is not the issue.

在与几位朋友讨论这一集内容时，不止一个人告诉我，RENTECH的运作方式有两种可能。一种版本是他们在二十多年前发现了一个永恒的秘诀，并一直基于此进行交易。

So in chatting with a few folks about this episode, I had more than one person say to me, there's two ways that RENTECH could work. And one version of how it works is they discovered something twenty plus years ago that is a timeless secret, and they've been trading on that for twenty plus years.

对。他们只是利用了某类股票之间的特定关系，而这种关系除了他们没人能破解。

Right. There's one particular relationship between types of equities that they've just been exploiting, and no one can figure out except them.

确实。这种可能性完全存在。

Right. And that may entirely be possible.

这难道不疯狂吗？

Isn't that crazy?

没错。现在，Rentec的人都会说这绝对不是他们的运作方式。完全不是这样。即使事实如此，他们当然也会否认，因为他们不想让任何人知道

Right. Now, Rentec will say they will all say that is a 100% not the way that it works. It's not that at all. If that were the way that it works, they would, of course, still say that because they don't want anybody to

对，别去研究大豆期货与通用汽车之间的关系。千万别这么做。

Right. Don't look at the relationship between soybean futures and GM. Just don't do it.

没错。我们先承认这种可能性或许存在。但更可能的是Rentec所说的才是事实——根本不存在什么圣杯。我们这里所做的就是每两年彻底重构整个系统。

Right. So let's accept that there is a possibility that that might be true. More likely, though, is that what Rentec does say is true, which is no. There is no holy grail. What we do here is we completely reinvent the whole system continuously on a two year cycle.

我听说的周期大概是两年。模型每两年会完全重构一次。并不是说固定在某个日期，而是每天都在进行重构。但整体来看大约是两年一个周期。

Two years is kinda what I heard. The the model is fully restructured every two years. It's not like on a date every two years. It's being restructured every day. But collectively, it's about a two year cycle.

那么这就说明这些人确实可以做到。如果有五个人离职，他们很可能能重建系统，唯一需要的就是数据。

So that would be an argument then that the people actually could. If five people left, they probably could go recreate it, and all they would need is the data.

这也说明无论是模型本身还是数据，这里都不存在真正垄断性的资源。

It's also an argument that there is no actual cornered resource here in terms of either the model itself and maybe not the data either.

但我打赌数据确实是。假设你在那里工作了十年，你也不知道1955年的大豆期货数据是怎么进入数据库的。即使你习惯使用这些数据并且能在别处重建模型，你也不知道它最初是如何被录入的。

I bet the data is, though. Let's say you've been working there for ten years. You don't know how the 1955 soybean futures data ended up in the database. Even if you're used to using that data and you're able to go recreate the model elsewhere, you don't know how it originally found its way in.

这个观点很公允。我认为关于那些历史数据确实有帮助，但随着市场发展，其价值会随时间递减。确实如此。我想强调的更宏观观点是：其他所有主要量化公司也都在这些方面投入了数亿甚至数十亿美元。

I think that's fair. I think there might also be some argument to the data that that older data is helpful, but its value decays over time as markets evolve. Definitely. The broader point I wanna make here is just that every other major quant firm out there is also spending hundreds of millions, if not billions, on this stuff too.

人们正在四处寻找替代数据。像桥水基金这样的机构，正为那些你根本想不到会影响股市的数据支付巨额资金，他们为此投入数百万、数千万甚至数亿美元。

And people are looking for alt data everywhere. The Bridgewaters of the world are paying gobs of money for things that you would never dream could possibly have an effect on the stock market, and yet they're paying millions or tens of millions or hundreds of millions of dollars for it.

没错。所以我认为我们完全可以排除规模经济的影响。相反，这里存在的是反规模经济效应。

Yep. So I think we can rule out scale economies for sure. If anything, there are anti scale economies here.

哦，是的。这里完全存在规模不经济。当资产管理规模过大时，你们的策略就会失效。

Oh, yes. There's totally there's diseconomies of scale. Your strategies stop working when you get too much AUM.

对。会出现滑点问题。我不认为这里存在任何网络经济效应。我的意思是，他们基本上不跟任何人交流。

Yep. You get slippage. I don't think there's any network economies here. I mean, they literally don't talk to anybody.

不过，他们确实与电子经纪商和交易执行链中的不同参与者建立了非常稳固的关系。我认为他们的交易执行能力非常出色，市场数据获取速度极快。他们从市场中提取数据的能力质量极高。

Although well, they do have some very well established relationships with electronic brokerages and different players in the trade execution chain. I think they have very good trade execution and very fast market data. Their ability to pull data out of the market is very high quality.

但你觉得这真的比他们的竞争对手强吗？

Do you think it's actually better than their competitors, though?

我不确定。这可能不是他们的制胜法宝。

I don't know. That's probably not the secret sauce.

是的，我不这么认为。这是基本门槛。转换成本不适用。品牌效应或许在他们为机构基金筹集资金的能力上有所体现，但那不是业务的主要部分。

Yeah. I don't think so. It's the table stakes. Switching costs, don't think apply. Branding maybe applies in their ability to raise money for the institutional funds, but that's not a big part of the business.

机构基金的费用流可能完全归功于品牌效应。确实如此。但我认为有很多上市股权投资公司和大量对冲基金拥有强大的品牌影响力，平均能获得市场回报率和不错的夏普比率，并且能够筹集资金正是因为建立了品牌。没错，风险投资公司也是如此。

The fee stream on the institutional fund may entirely belong to branding. Yes. But I think there's a lot of public equity firms and a lot of hedge funds that have a lot of branding power that have, on average, market returns with decent sharp ratios and are able to raise because they've built a brand. Yep. Venture firms the same way.

完全同意。

Totally.

所以对我来说，这就剩下反向定位了。我确实认为这里存在一些反向定位，而且我们即将连续两期讨论大规模的反向定位策略。

So for me, this kinda leaves counter positioning. I actually think there's some counter positioning here, and I think we're gonna have two episodes in a row of counter positioning at scale.

说说你的反向定位。谁在被反向定位？以什么方式？

Tell me about your counter positioning. Who is being counter positioned in in what way?

他们在市场上的直接竞争对手，其他量化公司。我说的直接竞争对手，显然不是指有限合伙人资金方面，而是指同类型的交易活动。

Their direct competitors in the market, the other quant firms. And when I say direct competitors, I obviously don't mean for LP dollars. I mean for, like, the same type of trading activity.

就像他们在交易中的对手方？

Like, their counterparties in trades?

我不认为他们是交易对手。我认为他们都在试图利用类似的交易类型。我认为真正的交易对手是那些被他们利用的牙医之类的人。

I don't think they are counterparties. I think they are all seeking to exploit similar types of trades. I think the counterparties are the people there, the dentists that they're taking advantage of.

不过，量化基金之间经常互为交易对手。

Well, but quant funds are often counterparties to each other.

确实如此。但我认为，在寻找同类交易时他们是对手。就RENTECH或具体到Medallion基金的反向定位而言，其一，我认为单一模型策略与多数机构采用的多模型多策略相比确实具有优势，就像我之前讨论挂毯理论时说的。但更重要的是，RENTECH的所有激励机制都完全围绕着为业绩优化基金规模而设计，这几乎是其他机构无法比拟的。我认为他们比任何人都更有动力真正实现业绩最大化。

That's true. But I think, yes, adversaries in finding the similar types of trades. And I think the counter positioning for RENTECH or for Medallion specifically is one, I do think, the single model approach versus the multi model, multi strategy approach that most others have does have benefits like I was talking about in the tapestries. But I think also and maybe bigger is every incentive at Rend Tech is fully aligned to optimize fund size for performance in a way that is not true just about everywhere else. I think they have the most incentive of anybody to truly maximize performance we're able to achieve.

没错。尽管管理规模扩大会带来更多管理费收入，但他们独特的激励机制使得他们不愿为追求规模而牺牲业绩。

Right. Even though the dollars would continue to rise because they get fee dollars from more money in the door, they are incentivized in a unique way that makes it so they're not willing to trade the dampener on performance to get those dollars.

是的。特别是因为GP和LP两边其实是同一批人。

Yes. Particularly because it's all the same people on the GP and LP side.

唉，我们又绕回这个老问题了。我勉强接受这个反向定位的说法。但我觉得答案其实简单得令人恼火——他们就是比所有人都更擅长这类数学和机器学习，而且钻研时间更长。所以他们能一直碾压你。

Oh, we keep going round and round that axle. I loosely buy the counter positioning thing. I just think the answer is disgustingly simple and kind of annoying here, which is they're just better than everyone else at this particular type of math and machine learning, and they've been doing it for longer. So they're just gonna keep beating you.

哦，这倒是另一个我听过的观点——从某种意义上说，RENTECH本质上是个数学系，这是其他任何公司都无法比拟的。

Oh, that's another argument I heard from people, in that RENTECH basically is a math department in a way that none of these other firms are.

可能是文化因素。

It could be culture.

是啊，可能是文化因素。

Yeah. Could be culture.

说真的，可能只是因为文化氛围的构建方式能持续吸引合适的人，并以一种虚假的利他主义方式激励他们。比如，这里只是让我愉快工作的地方。而结果就是变得非常富有，但我不会去城堡公司工作。

I mean, honest to God, it could just be that the culture is set up in a way that continues to attract the right people and incentivize them in a sort of fake altruistic way. Like, this is just a fun place to do my work. And, the outcome is getting really rich, but I wouldn't go work at Citadel.

对，我觉得有可能。所以这或许与流程效能有关。对，好的。

Yep. I think that could be. So maybe that feeds into process power. Yep. Okay.

对我来说，是流程效能与反定位的结合，我认为与其他效能无关。

For me, it is some combination of process power and counter positioning, and I don't think it's any of the other powers.

对我来说，是流程效能和垄断资源。

For me, it is process power and cornered resource.

对，好的。我认同这个观点。

Yep. Okay. I buy that.

七种力量框架中未涵盖的是战术层面，比如执行力。七种力量的核心在于战略与战术的区别。我认为，RENTECH可能确实持续在执行力上超越了竞争对手。某种程度上，就是因为他们比你更聪明。

And a thing that's not captured in seven powers is tactical, like execution. The whole point of seven powers is strategy is different than tactics. And I think, legitimately, RENTECH may just have persistently been able to out execute their competitors. There's part of it that's just like they're smarter than you.

是的。如果你认同整个体系每两年就会被彻底重构一次，那么

Yeah. Well, if you buy the the whole thing gets reinvented continuously every two years, then

没错。而且还有遗留知识。比如，如果你从2019年（或1964年）就开始构建机器学习系统，那么你现在肯定非常擅长机器学习。那些过去十五年与你共事、吸收你历史知识、在你系统里工作的人，也会比那些受当下热潮影响才开始学习机器学习的人更精通。所以我的答案是：知识会复利增长。

yes. And there's remnant knowledge. Like, if you started building a machine learning system in '19, whatever it was, '64, you're gonna be really good at machine learning today. And the people that you've been spending time with for the last fifteen years, learning all of your historical knowledge and working in your systems, are also gonna be better at machine learning than probably the other people who are out in the world learning it from people that just got inspired to start learning machine learning based on the new hotness. So learnings compound is my answer.

很棒。

Great.

好的。战术手册。除了你编织的David Rosenthal三重奏挂毯之外

Okay. Playbook. So in addition to the three part David Rosenthal tapestry that you have woven

我没什么要补充的了。

I have nothing more to add.

我认为有几点值得强调。第一，信号处理就是信号处理就是信号处理。他们不关心底层资产，实际上不基于基本面交易——除了在机构基金里会稍微参考基本面。他们在机构基金里使用市盈率这类指标，这有点讽刺，因为这完全是另一套技能。但如果你只看奖章基金，那全是抽象数字。

There are a handful of things that I think are worth hitting. So the first one is signal processing is signal processing is signal processing. They by not caring about the underlying assets, they literally don't trade on fundamentals, except in the institutional fund when they trade on fundamentals a little bit. They use price to earnings ratios and stuff like that in the institutional fund, which is kinda funny because that's a completely different skill set. But if you just look at medallion, it's all just abstract numbers.

你其实不必关心这些数字背后的含义。你只需要看它是线性回归还是他们做的那些更花哨的东西，就是数据之间的关系。一旦你简化到这种程度，就会发现这太精妙了——他们可以从任何领域招人。一个人过去做过多么复杂的信号处理并不重要，不管他是天文学家试图对一颗极其遥远的恒星照片进行'降噪'，还是尝试做过自然语言处理之类的工作。说到底都是信号。

You don't actually have to care about what underlies those numbers. You just have to look for whether it's linear regression or any of the fancier stuff that they do, just relationships between data. And once you reduce it to that, it is so brilliant that they can just recruit from any field. It's not relevant how someone has done sophisticated signal processing in the past, whether it's being an astronomer and trying to denoise a, quote, unquote, photo of a star super far away or whether they've tried to do, like, natural language processing. It's just signal.

吉姆、彼得和其他人被问及为什么只从学术界招聘而不从华尔街之类的地方招人时，他们有个特别搞笑的说法。他们会说：'我们发现教聪明人投资业务，比教投资人变聪明要容易得多。'

There's this really funny line that Jim and Peter and others will say when asked about why they only hire academics enough from, you know, Wall Street and whatnot. And they're like, well, we found it's easier to teach smart people the investing business than teach investing people

如何变聪明。

how to be smart.

没错。这太荒谬了。他们根本没教任何人任何投资知识。他们只是在做信号处理。我打赌RENTECH研究部门至少一半的人都看不懂资产负债表。

Right. That's ridiculous. They don't teach anybody anything about investing. They're just doing signal processing. I bet at least half the people at RENTECH on the research side could not read a balance sheet.

太好笑了。投资行业里一大群人，没一个是真正的投资者。

It's so funny. It's a whole bunch of people who are in the investment business, none of which are investors.

确实。

Yes.

还有个例子，你可以自行判断是否贴切。我最近一直在思考复杂适应系统——自从几年前NZS Capital那帮人来做客，读了他们的著作和圣塔菲研究所的相关研究后，这个概念就一直萦绕在我脑海。在复杂适应系统中，要真正理解一件事如何影响其他所有事物非常困难，因为这些关系在组合上过于复杂，你无法确定性地认定这件事就是那件事的原因。就像蝴蝶扇动翅膀那样。

Another one that you can decide if this fits or not. I was thinking a lot about complex adaptive systems. It's always been on my mind since we had the NZS Capital guys on a few years ago and read their work and the Santa Fe Institute's work on this. In a complex adaptive system, it's really difficult to actually understand how one thing affects everything else because the idea is the relationships are so combinatorially complex that you can't deterministically nail down this one thing as the cause of that other thing. It's the butterfly flapping its wings.

但实体之间存在一些你无法从表面理解或看到的关系。还记得我们之前做第二期NVIDIA节目时，我开场提到小时候总盯着火焰想：如果你真知道木头里原子的组成，真知道风的走向，真知道所有条件，你能否模拟出火焰？小时候你总觉得不可能。但实际上答案是肯定的。这就是一个已知现象——未来三小时这根木柴燃烧会发生什么，你甚至能精确看到火焰形态。

But there are relationships between entities that you can't understand or see on the surface. Do you remember way back when we did our second NVIDIA episode, I opened with the idea that when I was a kid, I always used to look at fire and think, like, if you actually knew the composition of the atoms in the wood, and you actually knew the way the wind was blowing, and you actually knew that like, all the could you actually model the fire? And when I was a kid and you always just assumed no. But, actually, the answer is yes. This is a known thing of what will happen when you light this log on fire for the next three hours, and can you see exactly the flames.

我认为RENTECH本质上还没为市场解决这个问题。他们无法预测未来。但如果他们有50%或1%的正确概率，就能对复杂适应系统说：我们不在乎它多复杂，我们的模型已足够理解这些实体间的关系，只需多次运行模拟就行。通过无数次微小优势积累的利润，他们已成为最接近预测市场这个复杂适应系统运作的机构。不过我认为他们无法逆向推导原理。

I think RENTECH has basically they haven't figured that out for the market. They can't predict the future. But if they have a 50 o 1% chance of being correct, then they can sort of take a complex adaptive system and say, we don't really care that it's a complex adaptive system. Our models understand enough about the relationships between all these entities that we're just gonna run the simulation a bunch of times, and we're gonna be profitable enough from all the little pennies that we're collecting on all the little coin flips where we have a slight edge over and over and over and over again that they're sort of the closest in the world to being able to actually predict how the complex adaptive system of the market will work. Now I don't think they can back out to it.

没人能解释清楚，但我认为他们的计算机可以。

No person could explain it, but I think their computers can.

是的。我听RENTECH的人讨论时，他们都会说：模型其实不懂市场，但能预测。我们对它预测市场行为的信心足以让我们依赖它。懂不懂其实不重要，就像它无法解释原因。

Yes. And I think when I've heard people from RENTECH talk about this, they will all say, the model does not actually understand the market, but it can predict, and we can be so confident in its predictions about what the market will do that we rely on it. Whether it understands or doesn't understand doesn't actually matter. Like, it can't tell you why. Right.

但这没关系。

But that's okay.

但它确实知道自己有微小优势，所以应该据此交易，尽管无法解释原因。说到模型，我一直想确定这个问题的答案：你认为RENTECH是机器学习的发源地吗？

But it does know it has a slight edge, and so it should trade on it even though it can't explain why. Yes. Well, speaking of models, I've been trying to nail down an answer to this question. Do you think RENTECH was the birthplace of machine learning?

这真是个难题。我们其实发邮件问了些顶尖AI研究者和历史学家朋友，得到的答案不出所料：'我们不知道，因为他们从不公开任何信息。'

This is such a tough answer to tell. We actually emailed some friends who are very prominent AI researchers and AI historians and sort of asked this question. And the answer we got back is unsurprising. They said, we don't know because they don't share anything.

没错。这些原理确实源自孕育了机器学习的同一个数学界。但Rentech过去二十年在谷歌Gemini模型和ChatGee中取得的成果并非如此。他们并未反哺任何研究成果。实际情况可能是，Rentech确实抢先所有人一步，他们拥有强大AI或某种比当今世上所有AI都精妙得多的技术。

Right. It's like the principles certainly came out of the same math community that spawned machine learning. But is what Rentech has figured out over the last couple decades in Google's Gemini model and in ChatGee no. It's not, because they don't contribute any research back. It may be the case that, actually, Rentech has beat everyone else to the punch, and they have a strong AI or something that is actually much more sophisticated than all the AI we have out in the world today.

他们只是选择将其封锁独占来赚大钱。我的意思是，可能文艺复兴公司只是尽可能多地吸收非结构化数据，他们比其他人早十年二十年意识到：拥有大量无结构、无标签的数据时——对于大语言模型能生成看似正确的表述，对于交易策略则能获得超过50%的胜率。

And they've just chosen that they'd rather keep it locked up and captive and make a bunch of money. I mean, it could just be the case that Renaissance is just taking in as much unstructured data as it possibly can, and they sort of were just a decade or two ahead of everyone else in realizing that you can have unstructured, unlabeled data. And if you have enough of it, you can make it, in the case of an LLM, say things that sound right or sound true, or in the case of these trades, be right more than 50% of the time.

没错。做出看似正确的交易。

Right. Make trades that sound right.

是的。在去年AI爆发之前，他们早就比其他所有人更早掌握了这项无监督学习技术。

Right. They figured out this big unsupervised learning thing before anybody else all the way up until last year when AI moment happened.

若真如此，我们对鲍尔斯就该有截然不同的答案了。

If that were the case, we should have very different answer to Powers.

说明这一点很有趣——彼得·布朗的学术导师正是杰弗里·辛顿。

To illustrate this point, it's quite interesting. Peter Brown's academic adviser was Jeffrey Hinton.

没错！哦，真高兴我们提到这个。是的，Rentec和AI确实源自完全相同的人才储备圈层，来自同一个学术团体和社交圈。

Yes. Oh, I'm so glad we brought this up. Yeah. It was the exact same stew and the exact same cohort of people in social group and academic groups that Rentec came out of, that AI came out of.

对于有些人可能会问‘你为什么要这么说’的情况，为了让事情更加明确，另一位导师是杰弗里·辛顿的人是伊利亚·苏茨凯弗，他是OpenAI的联合创始人。我是说，那是很多年后的事了，但依然如此。

The other person, just for people who are like, why are you saying that? To make it super explicit, the other person whose academic adviser was Jeffrey Hinton is Ilya Setskiever, who is the cofounder of OpenAI. I mean, many years later, but still.

是的。就像我们之前讨论马尔可夫模型和隐马尔可夫模型那样。那是Rentec的基础，也是当今人工智能和生成式AI的基石之一。

Yeah. I mean, it's like we were talking about with Markov models and hidden Markov models. That is the foundation of Rentec. That is one of the foundations of AI and generative AI today.

没错。好的。另一个重要概念是，你应该基于别人没有发现的秘密进行交易。表面上看这似乎很明显。当然，我应该想出一些别人没有采用的交易策略。

Yep. Okay. Another big one is this concept that you should trade on a secret that others are not trading on. So on the face of it, it seems obvious. Of course, I should come up with some strategy to trade on that other people aren't trading on.

但我刚才用了几个词——‘当然，我应该想出’——这里就存在谬误。我认为大多数投资公司都试图从人那里获取想法，然后进行极其严格的数据分析来决定是否执行这些交易。我可能错了，但我不认为现代RENTECH是这样做的。我认为他们所有的投资理念都来自数据和信号处理。因此，你会执行那些在直觉上毫无意义的交易。

But I said a couple of words there, which is, of course, I should come up with, and therein lies the fallacy. I think most investment firms try to get their ideas out of people and then do an incredibly rigorous amount of data analysis to figure out if they should put those trades on or not. I could be wrong, but I do not think modern RENTECH does that. I think all of their investment ideas come from data and come from signal processing. And so, therefore, you are going to put trades on that make no intuitive sense.

所以当你执行那些有利可图但在直觉上毫无意义的交易时，你就不会有竞争对手。如果你发现两个事物之间存在人类永远无法想到或梦想到的关联——而且我们说的可能是10个、20个甚至100个事物，在不同的时间尺度上以不同的权重组合——这就是利用无人知晓的秘密、在市场上击败他人的绝佳配方。

And so when you're putting trades on that are profitable and make no intuitive sense, you aren't going to have competitors. If you find a relationship between two things that a human could never come up with or dream of those relationships and and we're saying too, it end things, you know, 10 things, 20 things, a 100 things, and in various different weights at various different time scales, that is a killer recipe to exploit a secret that no one else knows and be able to beat other people in the market.

说得太对了。许多（如果不是大多数）其他量化公司都没有这样做。也许有些公司会，但我认为大多数情况下是模型提出建议，然后由一位或多位主投资组合分配者决定是否执行。

Such a good point. And many, if not most of the other quant firms are not doing that. Some of them maybe, but I think most of them are the model is suggesting things, and there is a person or persons who are the master portfolio allocators that pull the trigger or don't pull the trigger.

是的。为了更形象地说明，因为人们自然倾向于认为‘哦，我能理解为什么这两件事有关联’——这种关联可能并非你所想。例如，可能有两样东西总是同步变动，比如特斯拉股票和小麦期货。作为天生爱编故事的人类，你可能会在脑子里编造出它们为何会同步变动的原因。

Yes. And to be super illustrative, because I think your natural tendency is like, oh, I can understand why these two things would be related. The relationship may not be what you figure. For example, there could be two things that always move together, Tesla stock and wheat futures. And you might try to because humans are storytellers, concoct some story in your head of why those move together.

如果你相信这一点，那么你可能会认为它们应该在某个日期停止同步波动。但很可能只是某家大型对冲基金同时持有这两样东西。当它们调整仓位时，就会导致这些资产同步波动。但你永远想不到这点。你会以为这两者之间存在直接关联，而不会想到只是因为某个持有它们的机构同时在市场上提供了流动性。

And if you believe it, then you might decide there's some date where they should stop moving together. Well, it could very well be that some other big hedge fund just owns both of those things. And when they rebalance, it causes those assets to move together. But you would never think of that. You would think these things have a direct relationship with each other, not just that there's liquidity in the market from both of them at the same time because someone else owns both of them.

所以我认为Rentec某种程度上承认的是：我们根本不知道事物之间为何存在关联，但这并不重要。没错，完全如此。

So I think what Rentec sort of admitted is, we have no idea why anything is actually connected, but it doesn't matter. Yep. Totally.

这个研究发现让我很惊讶。我原本以为整个量化行业都是这样运作的，但惊讶地发现其实并非如此——基本上只有RENTECH和少数几家机构采用这种方式。

And that was surprising for me in the research. Like, I sort of assumed that was the whole quant industry, and it was very surprising to me to discover that I believe, no. It is pretty much only RENTECH and maybe a couple other people.

好的。下个观点来自节目好友Brett Harrison，他在量化交易行业深耕多年。他提出了一个二维矩阵理论：横轴代表交易执行速度的快慢，纵轴代表策略的聪明程度与明显程度。

Okay. My next one is brought to you by a friend of the show, Brett Harrison, who has worked in the quant trading industry for a long time and shared an idea that he has with us, which is that there's basically this two by two matrix. You have, on the one axis, fast and slow in terms of trade execution. And on the y axis, you have smart versus obvious.

是的。他对我们的表述是'聪明vs愚蠢'，但这里的愚蠢并非真指愚笨。

Yeah. The way he phrased it to us was smart versus dumb, but dumb doesn't mean dumb.

没错，指的是显而易见的交易。关键点是：并非所有量化基金都是高频交易公司，反之亦然。作为行业外行，这个认知盲区现在让我豁然开朗——我原以为它们是一回事。

Right. It's the obvious trades. And the high level point is all quant funds are not high frequency trading firms and vice versa. And this is something that I didn't know, not coming from this industry, and now makes total sense to me. I think I thought they were the same thing.

而'快速+明显'就是典型的高频交易者。他们进行交易前置，把服务器架设在交易所附近的数据中心——就像《高频交易员》描写的场景。或者在纽泽西和芝加哥之间架设微波线路，试图套利两地市场价差。这需要世界上最快的连接速度才能实现。

But fast and obvious is your classic high frequency trader. They're front running trades. They're locating in a data center that's really near the you know, this is Flash Boys. Or they've got a microwave line between New Jersey and Chicago, and they're trying to arb the difference between two markets. You need to have the fastest connectivity in the world to pull this off.

是的，这里是简街。

Yep. This is Jane Street.

没错。有快速和聪明两种路线，你其实不需要同时具备。你不需要世界上最快的连接速度，也不需要设计最精妙的交易策略。所以人们往往会选择一条路，要么做高频交易员，要么尝试进行最聪明、最不显而易见的交易。这自然就引出了奖章基金，它属于慢而聪明的象限。

Yes. There's fast and smart, which you kind of don't need to be both. You don't need the fastest connectivity in the world and the most clever trades to put on. So people kinda tend to pick a lane that they're either a high frequency trader or they're trying to make the smartest, you know, most non obvious trades possible. And that, of course, leads us to medallion, which is in the slow and smart quadrant.

所有机器学习系统都在数据中发现了关联性，因此需要巨大的计算量。

All the machine learning system discovered the relationships in the data, so there's a huge amount of compute.

那些不显而易见的交易。

The non obvious trades.

正是如此。这些计算用于发现不显而易见的交易，但实际执行时却相对缓慢。虽然仍需在几秒或几分钟内完成，但优势不在于像《闪电小子》里描述的那种高频交易。

Exactly. That goes into finding the non obvious trades, but then they're actually made reasonably slowly. They still have to happen within seconds or minutes, but the advantage isn't that they're high frequency the way that all the Flash Boys stuff is.

我的感觉是Rentec不是高频交易公司。他们不搞抢先交易，明白吗？他们不是闪电小子。虽然相比你我，他们的操作速度仍然快得惊人，但更注重智慧而非速度。

My sense is Rentec is not a high frequency trading shop. They're not front running things. You know? They are not Flash Boys. Compared to you and me, they still operate incredibly fast, but it's more about the smartness and less about the fastness.

格雷格在他的书中有句名言：他们随时持有数千个多头和空头头寸，持仓周期从一两天到一两周不等。每天进行15万到30万笔交易，但大部分交易是通过小批量买卖来避免影响市场价格，而非通过抢在其他投资者前面获利。

Greg has a quote in his book. They hold thousands of long and short positions at any given time, and their holding period ranges from one to two days or one to two weeks. They make between a 150,300 trades a day, but much of that activity entailed buying or selling in small chunks to avoid impacting market prices rather than profiting by stepping in front of other investors.

哦，这是我们听说的另一件事。文艺复兴科技公司在掩饰交易方面是世界一流的。

Oh, this is another thing that we heard. RenTech is world class at disguising their trades.

是的。他们能做到不惊动市场，你不知道是谁在操作或何时操作。这是因为早期他们并不擅长这个，人们基本上能截获他们的交易并抢先交易，他们不得不适应并开发这些聪明的系统，让你不知道是谁在买，不知道买了多少，也不知道他们是否会继续买。没错。

Yeah. They can make it so that they don't move the market, and you don't know who is acting or when. And this is because in the early days, they weren't good at this. And people basically intercepted the trades that they were making and were front running them, and they had to adapt and develop these clever systems to make it so you don't know who's buying, and you don't know in what quantities, and you don't know if they're gonna keep buying. Yep.

在我们讨论价值创造和价值捕获之前，我要说的最后一点是，这是一个令人恐惧的行业。你需要的控制措施、风险模型和紧急停止机制都至关重要。如果软件有漏洞怎么办？有没有可能在几分钟内做出一大堆无利可图的交易并亏光一切？要知道，在以前打电话给经纪人的时代，这是不可能发生的。

My last one before we get into value creation, value capture, is that this is a terrifying business to be in. The amount of controls and risk models that you need and kill switches are just so important. What if the software has a bug? Is it possible to make a ton of unprofitable trades in a matter of minutes and lose it all? You know, that wasn't possible in the old world where you're calling your broker.

在这里完全有可能发生。而且确实发生过。是的。虽然文艺复兴科技从未遇到过，但2012年有一家叫骑士资本的公司一天就损失了4.6亿美元。他们部署新代码的过程中出现了一个漏洞。

That totally is possible here. And it happened. Yeah. And while it's never happened to Rendtech, there was a company called Knight Capital in 2012 that lost 460,000,000 in a single day. There was a bug in their process to deploy the new code.

简单来说，发生的是一个简单的标志位错误——一个从0到1的位设置被误解了，导致无限循环运行。当某个交易发生时，本应翻转某个位，结果翻转了另一个位。系统没有在内存中查看同一个位的位置，所以它以为这个位从未被翻转。这个无限循环在45分钟内执行了400万笔交易，而系统没有内置适当的紧急停止机制。他们基本上眼睁睁看着资金流失，却无能为力。

And, basically, what happened, it was a simple flag error, a misinterpretation of setting a bit from zero to one that caused this infinite loop to run, where once a certain trade happened, it was supposed to flip the bit. It flipped a different bit. The systems were not looking at the same location and memory for the same bit, And so it basically thought it was never flipped. This infinite loop ran 4,000,000 trade executions in forty five minutes, and there wasn't the appropriate kill switches built in. And they basically watched it all to just drain out, and there was nothing they could do.

是啊，就像整个投资组合都没了，对吧？

Yeah. So, like, the whole portfolio gone. Right?

是的。嗯，我不知道是不是整个投资组合，但他们损失了大量有限合伙人的资金。而且他们是上市公司。一夜之间，他们的股价下跌了75%，然后有人介入收购了他们。

Yes. Well, I don't know if it's the whole portfolio, but it was enough that they lost a huge amount of the LP capital. And then they were a publicly traded firm. Overnight, their equity traded down 75%, and then someone stepped in and bought them.

嗯，他们很可能被所有交易对手点名批评了。

Well, they probably got marched and called by all their counterparties.

所以，负责RENTECH财务控制和安全系统的人，在这个行业里可是个重任。完全同意。好了，为了开启价值创造和价值捕获的话题，我要发表一个挑衅性的观点：大卫，文艺复兴科技公司其实并不从事投资业务，他们做的是赌博生意。

So whoever is in charge of the financial controls and safety systems at RENTECH, that's a huge job for someone in this industry. Totally. Alright. To kick off value creation, value capture, I have a provocative statement, which is, David, Renaissance Technologies is actually not in the investment business. They are in the gambling business.

具体来说，他们就是庄家。

And in particular, they're the house.

好吧，我本想顺着你的思路说下去，结果发现完全赞同。他们确实不在投资行业，根本不懂投资——是他们的模型在操作。

Well, I would tell I thought where you thought you were going with this, I was like, yes. I would totally agree. They're not in the investment business. They have no idea how to invest. The model does.

我这么说吧：他们不是投资者，也不做投资业务。虽然他们交易的市场里处处都是投资行为，但他们在那些市场里不是以投资者身份存在，而是像凯撒皇宫那样开店营业，让所有人都来和他们交易，同时保持微弱的优势。

I'll say this. They're not investors, and they're not in the investment business. There is investment going on all around them in the markets that they trade in. But the fact that they're in those markets, they're not there as investors. They're there setting up shop as Caesar's Palace, letting everyone come in and do business with them while they have a slight edge.

他们有时会输，但大多数时候都能小赚一笔。假设他们正确的概率是50.1%，这些年来他们只是向所有愿意与之交易的人收取佣金。大规模运作下确实奏效——吉姆·西蒙斯从所有交易对象口袋里掏走了300亿美元。

And they'll lose sometimes, but most of the time, they're gonna come out slightly ahead. And I think, let's say they do have a 50 o 1% chance of being right. They're just there to collect their vig on everyone who is willing to trade with them over all these years. And at scale, it really worked. Jim Simons managed to drain $30,000,000,000 into his own pocket out of everybody that he ever traded with.

现在我明白你的意思了，这确实类似于凯撒皇宫或赌场的模式。他们不从事投资业务，而是在提供一种服务。

Now I think where you're going with this is perhaps similarly along the lines to Caesar's Palace or a casino. They are not in the investment business, but they are providing a service.

当然。

Sure.

这就是你要表达的意思吗？

Is this where you're going with this?

嗯，我是说，投资业务其实取决于你如何定义投资者。如果你想装腔作势的话——就像这个示例中我正在做的那样——那么投资者就是为商业提供资本的人，也就是风险资本，目的是让企业在未来以某种方式创造价值；或者你把钱借给某些具有内在价值的资产，使其能够利用这些资本产生生产力，并为作为投资者的你带来回报。当然，有很多被称为投资的行为其实并非如此。如果我投入资金后能获得更多回报，却根本不关心钱是怎么赚来的，而且这实际上是零和游戏——我只是把钱从别人那里吸走——这还能算投资吗？

Well, I mean, the investment business, it sort of depends how you define investor. If you wanna be, like, all hoity toity about it, which I'm you know, in this illustrative example, I'm kind of being one, and saying an investor is someone who provides capital, you know, risk capital to a business for that business to create value in some way in the future, or you lend money to some intrinsic underlying asset so that it can be productive with that capital and produce a return for you as an investor. And, of course, lots of things are called investing that are not that. Is it investment if I put money to work and then I get more money back later, and I don't actually care how the money got made, and it's actually zero sum? I'm just vacuuming it out of

对，对。没错。这些钱根本没有被投入任何生产性领域。

Right. Right. Yeah. The money is not being invested in anything to produce.

正确。但这本质上和赌场的商业模式一模一样。你拥有微小的优势，然后让大量顾客进来，在你的微小优势下输钱给你。

Correct. But it's literally the same business model as a casino. You have a slight edge, and you let a whole bunch of patrons come in and lose money to you in your slight edge.

其实我想说的是，赌场也是服务提供商。它们为顾客提供娱乐服务。所有人都知道赌场设计的游戏规则对自己有利。同理，我认为可以论证——而且这个观点很可能非常准确——像Rentec这样的量化公司其实是在为市场提供服务，因为它们让人们想进行的交易能以更快的速度和更低的价差达成。

Well, where I was going with the service provider, I think casinos are service providers. They are providing entertainment to their customers. Everybody knows that the games are stacked in the casino's favor. Similarly, I think you could make an argument, and I think this is probably quite accurate, that Rentec and all other quant firms like them are providing a service to the market in that they are allowing trades that people want to make to happen faster and at much lower spreads.

完全正确。这是不可否认的——是的，量化基金确实为世界创造了价值。虽然很容易说量化基金没有创造价值，因为看起来像是零和游戏。它们实际上并没有为企业提供运营资本，纯粹是在进行套利或我们这期节目讨论过的各种策略。但你说得对，市场流动性确实有其价值所在。

Absolutely. That is the undeniable, yes, quant funds create value in the world thing, which I think is very easy to say quant funds provide no value because it's like it's zero sum. They're not actually providing the capital to businesses to do something with. They're purely looking to do an arbitrage or any of the strategies we've talked about this episode. But you're totally right that there is a value to market liquidity.

增加市场深度意味着，如果我们回到文艺复兴时期，零售投资者绝不可能像今天这样运作——零交易费用且几乎实时投资于所有这些不同公司。

Creating more depth to a market makes it so that if we go back to the era that Renaissance was started, there's no chance that retail is able to function like it does today with zero transaction fees and people able to invest in all these different companies at near real time.

我们任何一个人几乎都能在任何市场、一天中的任何时间即时购买证券，并获得非常非常非常精确的价格。是的，这些在过去都不可行。

And any single one of us can go buy a security in just about any market at just about any time of day pretty much instantaneously and get a very, very, very granular price on it. Yep. None of which used to be true.

没错。有大量量化基金、对冲基金愿意成为任何交易者的对手方，这本身就是一种服务。你说得对。它们也不都是奖章基金。实际上并非所有基金都有优势，尽管它们可能声称有。

Nope. The fact that there is a whole bunch of quant funds, hedge funds out there that are ready to be willing counterparties to anyone who wants to trade, that is a service. You're right. They're also not all medallion. They actually don't all have an edge even though they might purport to.

很多基金都会把钱输给你。

Lots of them are gonna lose money to you.

对。很多基金都在亏钱。听众们，你们也可能跑赢市场。这不是投资建议。请别尝试。

Right. Lots of them lose money. You too, listeners, could beat the market. Not investment advice. Please don't try.

是的。平均而言，奖章基金不会让你亏钱，但要知道市场上还有很多其他对冲基金、高频交易公司和对手方可供你选择。只是它们不是吉姆·西蒙斯。

Right. On average, Medallion will not lose money to you, but, you know, there are plenty of other hedge funds out there and high frequency shops and counterparties for you where you could take them. It's just not Jim Simons.

格雷格的书末尾有个精彩片段。那是什么时候？大概是在2010年代中期市场抛售期间，吉姆打电话给他的家族办公室负责人——要知道那时他已从文艺复兴科技公司退休多年——问道：'面对市场抛售我们该怎么办？' 而对方回应：'你可是吉姆·西蒙斯啊。'

There's this great, great vignette at the end of Greg's book. When was it? It was during one of the, like, sell offs in the mid twenty teens in the market where Jim calls the head of his family office. He's, you know, long retired from Rentec at this point, calls the head of his family office and says, what should we do with all the sell off in the market? And it's like, you're Jim Simons.

对。你是

Right. You're

吉姆·西蒙斯。我们该怎么办？

Jim Simons. What should we do?

我们该怎么办？是啊。

What should we do? Yeah.

是啊。人无完人。

Yeah. All humans are fallible.

完全同意。

Totally.

还有几个值得怀疑的地方。价值创造确实存在。人们很容易指责所有这些聪明人都投身金融业，希望他们能为世界做些更有建设性的事。但归根结底，人们总会去做那些有激励的事。除非出现一个能极大激励人们的全球性议题——比如二战时期，人们的爱国热情和拯救世界于邪恶的愿望，就曾成为移山填海的巨大动力。

A couple of other are squintable. The value creation exists. It's easy to knock that all these smart people are going into finance, and you wish they were doing something more productive for the world. At the end of the day, humans are going to do what they are incented to do. And so absent a larger global concern that is incredibly motivating to people I mean, you look at World War two, people's level of patriotism and wanting to go save the world from evil was a huge, unbelievable motivating factor to move mountains.

当这种动力缺失，或当人们感觉缺乏某种关乎存亡的信念时，他们就会去做对自己和家人最有利的事。如果是帝国建造者，就去建立帝国；如果是狂热的资本家，就去赚大钱。现行制度就是这样运作的。所以，你可以对此感到愤怒。

When that is absent or when people feel that there's some existential thing that is absent, they're gonna go do what's best for them and their family. And if they're an empire builder, go build empires. And if they're fierce capitalists, go make a bunch of money. And so the system is set up the way that it is. So, like, you can be mad about that.

既然如此，好吧，人们会投身量化金融这一高薪职业。幸运的是，这能带来许多有价值的东西。我认为常被忽视的是，这些高利润的职业和行业往往能产生具有广泛价值的研发成果。比如我们刚做的英伟达系列专题——在大型语言模型出现之前，你觉得Mellanox是做什么用的？

Given that, okay, people are gonna go engage in quantitative finance as a lucrative profession. Fortunately, there's a bunch of valuable stuff that comes out of that. And I think that is often missed, is that these really lucrative professions and businesses can often produce r and d that becomes valuable elsewhere. For example, we just did this big NVIDIA series. What do you think Mellanox was used for before large language models?

哦没错，在价值创造和价值捕获方面，这个观点真是令人震撼。继续讲，展开说说。

Oh, yes. This is such a really mind blowing point here in value creation, value capture. Go for it. Take it away.

其实很简单，就是大量InfiniBand技术被高频交易公司使用。虽然不确定，但我猜Mellanox是靠量化金融起家的。对，这只是众多案例之一。不过现在，你知道的，这有其局限性。

Well, there's not much to it other than a huge amount of InfiniBand was used by high frequency trading firms. And I don't know for sure, but I kinda think Mellanox built their business on Quant Finance. Yes. That's one of many examples. But now, you know, that has limits.

但我认为这里的技术创新经常被忽视。

But I think it goes overlooked that there's a lot of technology innovation here.

没错，这些都是很好的观点。研究中也提到了这些。我完全同意所有看法。在我看来，说量化金融没有为世界创造价值是错误的。

Yep. These are all great points. They all came up in the research. I totally agree with all of them. It is, in my opinion, false to say that quantitative finance does not create value for the world.

在我看来，它绝对创造了价值。

It definitely does, in my opinion.

但它创造的价值能与其捕获的价值相提并论吗？

But does it create anywhere near as much as it captures?

话虽如此，他们在价值捕获方面确实非常、非常在行。没错。这里可不是维基百科。这简直是光谱另一端的存在。

That said, they're really, really good at value capture. Yes. This is not Wikipedia here. This is about as far away on the spectrum as you can get.

《费城永远阳光灿烂》里有集特别棒，弗兰克（丹尼·德维托饰演）重操旧业——管他八十年代创立的是什么公司——又穿起细条纹西装什么的。他重新掌权后带着查理一起。查理就问他：'弗兰克，这公司到底是干嘛的？我们在这儿做什么业务？公司生产什么？'丹尼·德维托看着我说：'你什么意思？'

There's a great Always Sunny in Philadelphia where Frank, Danny DeVito, sort of goes back to his whatever business he founded in the eighties, and he's, like, dressing in his pinstripes and stuff again. And he's taken back over, he brings Charlie with him. And Charlie, you know, he's like, so Frank, what is the business, what do we do here? What does the business make? And Danny DaVita looks at me and goes, what do you mean?

‘我们赚钱啊。’查理说：‘不，不是。我是问你们制造什么？’他回答：‘我们制造财富。’

We make money. He's like, no. No. Like, what do you build? He goes, we build wealth.

我觉得这挺能概括这里发生的事。

I think that's a pretty good meme for kinda what's going on here.

没错。完全同意。在价值捕获方面也超级、超级擅长。确实。

Yeah. Totally. Very, very good at value capture too. Yes.

好了。熊牛之争。这个环节我们保留了很久，但上期节目没放进去——天啊，听众意见可大了。所以感谢大家的热心反馈。熊牛大战没死透，它又回来了。

Okay. Bear bull. So this was a section that we had for a long time that we did not put in the last episode, and boy, did we hear about it. So listeners, thank you so much for expressing your concern. Bearer versus Bull is unkilled, and it is back.

像凤凰一样浴火重生了。

Resurrected like a phoenix.

复活了。不过，选在这一集复活实在是最糟糕的时机。RENTECH的看涨理由是什么？

Resurrected. However, this is about the lamest episode to resurrect it on. What's the bull case for RENTECH?

过往表现是未来成功的风向标。

Past performance is an indicator of future success.

对。比如，他们能持续吸引全球最聪明的人才，保持其独特的企业文化，不会让机构基金业务成为主导业务。说白了，'保持现状'就是看涨的核心论点。

Right. Like, they're gonna keep attracting all the smartest people in the world. They're gonna have the ability to keep their incredibly unique culture. They're not gonna get tempted to let the business of institutional funds become the dominant business. You know, keep on keeping on is basically the bull case.

或许是因为他们确实仍处于领先地位。

Maybe, that they're actually still ahead.

对Medallion基金GP和LP利益相关者的看涨理由——虽然全球可能只有500人能参与，我们其他人都无缘接触。

The bull case for the GP and LP stakeholders in Medallion, which is, I don't know, 500 people in the world, And none of the rest of us can get any exposure to it.

没错。看跌的理由是形势在变。我认为任何维度的变化对他们都是利空——比如竞争对手正在赶超，又或者科技行业掌握了大型语言模型技术，这可能降低与RENTECH竞争的难度。

Yeah. The bear case is things are changing. And I think things are changing basically on any axis is the bear case for them. So things are changing where competitors are catching up, Maybe. Maybe the fact that the tech industry has figured out these large language models, maybe that trickles into making it easier to compete with RENTECH.

界限很模糊，但确实有可能。好比RENTECH比同行早入场十年，现在大家都来参加派对了。他们的文化也在变化——西蒙斯离开已久，鲍勃·默瑟也不再担任联席CEO。

It's a blurry line, but it is plausible. Like, maybe RENTECH actually was here a decade before everyone else, and now everyone else has arrived to the party. There's things that are changing maybe about their culture. Like, Jim Simons has been gone for a long time. Bob Mercer is no longer a co CEO.

彼得·布朗是联席CEO之一，他们刚刚宣布，原机构基金负责人大卫·利皮也将成为联席CEO。这可能是个利空因素——来自机构业务部门的人出任现任联席CEO，甚至可能最终成为CEO，如果你认为奖章基金才是核心业务而机构基金是瑕疵的话。用大卫的话说，这些机构基金就像是爱马仕苹果表带。这或许是个看跌信号，也可能意味着他们的人才正变得和其他公司没什么两样。

Peter Brown is a co CEO, and they just announced that they're making the guy who was in charge of the institutional funds, David Lippy, he is becoming a co CEO as well. So maybe there's a bear case around that, that someone from the institutional side of the house is becoming the current co CEO and maybe eventually CEO if you believe the medallion is the special thing and the institutional funds are sort of a blemish on the business. You know, they're the, Hermes Apple Watch strap in David's parlance. Maybe that's a bear case. Maybe there's a bear case that their talent is becoming kind of the same as everyone else's talent.

我在领英上看到很多RENTECH初级员工之前就职的公司都很眼熟。过去这里应该都是刚从大学研究机构出来的人。如果他们的人才渠道真的开始和其他公司趋同，那就值得担忧了。这些都是可以编造的故事，根本无法验证真伪。

When you look on LinkedIn, I recognize a lot of the companies that people worked at who are more junior at RENTECH. And in the past, I think it would have been all people just out of university research shops. So I think if it's true that they're starting to see the same talent flow as everyone else, that would be concerning. These things are all sort of narratives you can concoct and really no way to know if they're true or not.

没错。我们无从得知这些，因为这些根本无从考证。

Right. There's no way for us to know any of this because there's no way to know any of this.

是啊，都是秘密。

Right. It's all the secret.

好的。我们新的结尾环节——'心头刺'，也就是关键启示

Yep. Okay. Our new ending section, the splinter in our minds, the takeaway

那个让你无法停止思考的问题。

The one thing you can't stop thinking about.

过去一个月研究RENTECH的工作中，我们各自最难忘的一点是什么？对我来说——可能从我之前关于织锦的牢骚中就能看出来——我认为这是个关于激励机制力量的绝佳例证，如何正确制定和建立激励机制。还有企业文化，这点也不容忽视。

What is the one thing for each of us personally from doing this work over the past month on RENTech that sticks with us? For me, perhaps this is obvious from my little diatribe on the tapestry. I just think this is such a powerful example of the power of incentives and getting them right and setting them up right. And culture too. I don't wanna shortchange that.

我认为管理学术环境的文化应该像实验室那样运作，但又不至于陷入吉姆·西蒙所设立的那种轻浮的实验室氛围。

I think the culture of managing an academic environment in a fashion like a lab, but without letting it spin into the frivolity of a lab that Jim Simon set up.

对。换句话说，就是早期的谷歌。

Right. In other words, early Google.

没错。这就像早期的谷歌。完全正确。根据我们的研究，历史上从未有过，就目前所知，RENTECH也没有任何轻浮的行为。他们都非常专注，这再次让我想到激励的力量。

Yeah. This is like early Google. Exactly. There historically has not from our research, and as best as we can tell currently, is not anything going on at RENTECH that is frivolous. They are all very focused, which again to me then speaks back to the power of incentives.

当你身处一个不到400人的团队，在研究和工程方面不到200人。而那些与你共事的同事，是你所做一切的唯一提供者、监督者和受益者。这太强大了。是的。我想不出世界上还有哪里是这样的。

When you're there with less than 400 people, and on the research and engineering side, less than 200 people. And those colleagues who you work with are the sole purveyors, supervisors, and beneficiaries of all of this that you're doing. Like, that is so powerful. Yep. I can't think of anywhere else like that in the world.

我是说，也许一些风投基金或其他投资公司有类似情况，但没有像这样每天都有如此高流动性回报的。这是独一无二的。

I mean, maybe some venture funds or other investment firms, but not on a day to day fully liquid with returns like this. There's nothing like it.

没错。就像直接往血管里注射纯汽油。

Nope. Pure gasoline right into the veins.

是的。但这并不是说我就一定想在那里工作。我想我完全...不会。是的。

Yeah. Which is not to say I would necessarily wanna work there. I think I would Totally. Not. Yeah.

但它确实独一无二。

But it is truly unique.

是的。我一直在思考的就是之前提到的复杂适应系统这个概念。从外部观察来看，文艺复兴公司确实建立了一个大型计算机系统，能够发现世界上不同实体之间的关系——股票、大宗商品、债券价格等。无论它能否解释这些关系，大多数时候它都是正确的。可能只是略微多数，但只需要多数就够了，这样就能运作赌场生意。

Yep. The one thing I can't stop thinking about is the idea of the complex adaptive system that I was talking about earlier. I think from what everything we can tell from the outside, Renaissance actually has built a large scale computer system that discovers relationships between different entities in the world, stocks, commodities, bond prices. And whether it can explain them or not, it is correct most of the time. And it might be a small most, but all you need is most, and then you can operate a casino business.

我的理解是：他们就是庄家，拥有优势。这种优势建立在所有实体间关系图谱之上——在我们看来只是噪音的数据中，他们能识别出信号。

That is my takeaway, is that they are the house, and they have an edge. And that edge is predicated on a graph of all the relationships between these entities that we think are just noise, and they know the signal.

这确实让人好奇，你之前提到的科技行业近年来所谓的'追赶'。考虑到现在开源技术和市售技术，要构建这样的系统到底有多难？

It does make you wonder to what you were talking about with the tech industry catching up, quote, unquote, in recent years. How hard is it to build this now given the technology open source and otherwise that's available for sale out there?

这就是看空论点了。我也不知道。

That's the bear case. I don't know.

是啊。接下来会发生什么？既然这是个复杂适应系统，如果你现在能购买或构建它，那么收益就会被套利摊薄。

Yeah. And then what's gonna happen? By nature, given that it's a complex adaptive system, if you can now buy and build this, well, the returns will get arbitrage down.

没错。好吧，要不要来点有趣的特别环节？找点乐子吧。太棒了。

Yep. Alright. Should we have some fun carve outs? Let's have some fun. Sweet.

我有一部电视剧推荐，其实和《Acquired》有关，名叫《新风尚》，在Apple TV+上播放。

So I have one TV show, and it is actually Acquired related. It is called the new look on Apple TV plus.

哦，是的。但克里斯汀，这真是全新的风貌。

Oh, yes. But Christian, it is such a new look.

没错。听过LVMH那期节目的听众会记得，我们讨论过克里斯汀·迪奥的突破性设计——他的‘新风尚’系列，那是二战后震撼登场的一场时尚革命。

Exactly. So for anyone who listened to the LVMH episode, remember we were talking about the groundbreaking thing that Christian Dior did was his collection, the new look, that was a post World War two explosion onto the scene.

对生命的礼赞。

Celebration of life.

是的。军装式方正剪裁的时代过去了，现在我们迎来了这些性感撩人的设计，甚至可以说是——

Yes. Gone are the days of the militaristic boxy clothing, and now we are in with these seductive and, dare I say

奢华的面料。战时配给制结束了。正是如此。

Sumptuous materials. War rationing is over. Exactly.

没错。那些挑衅性的裙装。这部Apple剧集以震撼的闪回手法，展现了克里斯汀·迪奥、巴伦西亚加、可可·香奈儿在战时的痛苦经历，以及他们命运交织的传奇故事。

Yes. Provocative dresses. The Apple TV show is this incredible drama of kind of flashbacks to the wartime experiences, harrowing wartime experiences of Christian Dior, of Balenciaga, of Coco Chanel, and everything they went through and how all their paths crossed.

哦，可可也在里面？是的。哇，他们是怎么处理这个的？

Oh, Coco's in it? Yes. Oh, wow. How do they treat that?

如果很多人看这个节目，看看是否会影响香奈儿的产品销量，那会非常有趣。我也很好奇观众的看法，欢迎在Slack上留言讨论。你觉得她是个令人同情的人物吗？还是觉得她是个反派角色？我很好奇你们对她剧中形象与现实对比的看法。

It will be very interesting if a lot of people watch this show to see if that affects product sales of Chanel. I'm also very curious for people who are watching, feel free to put a thing in the Slack and carve outs. Do you think she's a sympathetic figure? Do you think she's a villainous figure? I'm curious how you think of her portrayal versus reality.

嗯，香奈儿有段疯狂的历史，公司最终被香奈儿香水部门收购，那是纽约的两位犹太兄弟——没错，就是韦特海默家族。

So Well, there's the whole crazy thing with Chanel where the company ends up getting bought by Chanel, the perfume division, which is the two Jewish brothers in New York. The Wertheimers, indeed.

天啊，我们迟早得做一期香奈儿的专题。但Apple TV+的《新风尚》，我保证无论你是否对时尚奢侈品感兴趣，这都是个既美丽又揪心的故事。

Oh, god. We gotta do a Chanel episode at some point. But the new look on Apple TV plus, I promise you whether or not fashion luxury is your thing, it's a beautiful and harrowing story.

哦，你和听众都知道，我平时不看电视，但这个太对我的胃口了。

Oh, you and listeners know, I'm not a TV guy, but this is so up my alley.

整个故事发生在战时的巴黎。

The whole thing, takes place in wartime Paris.

唉，好吧。我得去看看。

Ugh. Alright. I gotta watch it.

你得看看这个。好吧，大卫，说说你的专长领域。

You gotta watch it. Alright, David, your carve outs.

我的专长以一种非常独特的方式与新潮流相关，既涉及视频消费，也涵盖时尚、奢侈品和风格。这是棕榈滩Instagram和TikTok账号的典范。简直太棒了。

My carve out is related to the new look in a very different way, but both video consumption and fashion and luxury and style. It is the class of Palm Beach Instagram and TikTok account. This is so great.

大卫，我和你去棕榈滩待了两天，你就迷上了

David, you and I go to Palm Beach for two days, and you get hooked on

这太不可思议了。最近我和本去棕榈滩参加了一场演讲活动，待了几天，体验非常棒。我之前从没去过棕榈滩。

This is amazing. So Ben and I went to Palm Beach for a couple days for a speaking event recently, which was amazing. I've never been to Palm Beach before.

哦，那里确实不错。

Oh, it is nice.

太棒了。我们没特意留意到Rentec的人在场，但可能有。

So great. We didn't knowingly spot any Rentec people there, but we may have.

不过我们倒是特意注意到了几只铂金包。

We did knowingly spot some Birkin bags, though.

是的。我们刚录制完关于爱马仕的那期节目，就在棕榈滩。天啊，我当时特别开心能去那里。结果回到家，我妻子珍妮就说：你难道不知道棕榈滩的档次吗？

Yes. The style in Palm Beach we had just recorded the Hermes episode. And, oh, man, I was so pleased to be there. And then I got home, and Jenny, my wife, was like, do you not know the class of Palm Beach

TikTok账号？然后大卫说，我一千岁了。我完全听不懂你在说什么，珍妮。

TikTok account? And David's like, I'm a thousand. I have no idea what you're talking about, Jenny.

对。没错。没错。我就是个与世隔绝的老古董。

Yeah. Right. Right. Right. I live under a rock.

我是个老爸嘛。她给我看了这个——有个住在棕榈滩的女人，整天在Instagram和TikTok上发视频，随机采访路人穿什么牌子、什么风格的服装。简直精彩绝伦。

I'm a dad. And she showed it to me. This is a woman who lives in Palm Beach, and she goes around. She posts on Instagram and on TikTok, and she just interviews people on the street about what they're wearing, what brands they're wearing, their style. It is magnificent.

我最爱的一个片段——等会儿看看能不能找到链接放在节目备注里——有个被采访的女士，她的铂金包里还装着个迷你凯莉包。

My favorite is we'll see if we can find it and link to it in the show notes. There's a video of one woman who's being interviewed who has a mini Kelly inside her Birkin.

呃。太浮夸了。真是奢靡无度。

Ugh. Excess. Truly excess.

那一刻我就上瘾了。心想：这绝对是我看过最棒的东西，完全着迷了。

And that's when I was hooked. I was just like, this is the greatest thing I have ever watched. I'm obsessed.

好吧，如果我玩抖音的话，我会订阅的。

Alright. If I used TikTok, I would subscribe.

不用，你在Instagram上也能看到。哦，好吧。其实我已经在Instagram上关注了被收购的棕榈滩班级账号。

No. You can get it on Instagram too. Oh, alright. Good. I actually subscribed the acquired account on Instagram to class of Palm Beach.

我不知道我们关注了多少人。不多，但我们确实关注了棕榈滩班级。

I don't know how many people we're following. It's not many, but we are following class of Palm Beach.

看大卫打开了我们的Instagram账号。你真显年轻。哦不，大卫，我知道你收到了一些和你交谈过的人的感谢，其中几个是我们一起做的。是的。

Look at David opening up our Instagram account. You're so youthful. Oh, no. David, I know you've got some thank yous from folks you talked with, and a few of them we did together. Yes.

首先要特别感谢本期节目中那些慷慨贡献时间和见解的受访者。首先要大大感谢格雷格·祖克曼，《解构市场之人》的作者，这本关于文艺复兴科技公司和吉姆·西蒙斯的权威著作。没错。格雷格非常慷慨，花时间与我们交谈、邮件往来，确保我们理解正确。他和他的书也是大奖章基金投资回报数据的权威来源。

For sources for this episode who were so generous with their time and thoughts. First, huge thank you to Greg Zuckerman, author of The Man Who Solved the Market, the canonical book out there about Rentec and Jim Simons. Yep. Greg was super generous, spending time talking to us, emailing with us, making sure we're getting things right. He also he and the book is the canonical source of Medallion's investment returns.

我知道他费了很大功夫整理那份表格，现在网上到处都是这个数据，这也是应该的。

And I know he worked so hard to get that table together that is now all over the Internet as it should be.

太疯狂了。到处都在引用那个66%的数字，这都来自格雷格的分析。

It is crazy. Everywhere you hear that 66% number quoted, and that is from Greg's analysis.

是的。他进行那项研究并获取那些回报，确实为我们以及全球的企业历史学家和金融历史学家提供了宝贵的服务。

Yes. Truly a service to us and to corporate historians and financial historians everywhere that he did that research and got those returns.

还有几份其他的一手资料。实际上资料不多，我们可以在这里全部列出来。有彼得·布朗关于篮子期权事件的国会证词，还有彼得·布朗在高盛交易所接受的一次采访——其中许多问题直接出自格雷格的书和讲述的故事。

And there's a few other primary sources. There's really not much, so we can actually list all of them here. There's a congressional testimony of Peter Brown about the basket options thing. There's Peter Brown doing an interview at GS exchanges, which, again, many of the questions were straight out of Greg's book and the stories told.

没错。那是个有趣的时刻，彼得当时说：'你们从哪儿找来这些问题？怎么知道所有这些事的？'我心里想：拜托，他们读过那本书啦。

Yeah. It's a funny moment where Peter's like, where are you getting these questions? How do you know all this stuff? And I'm like Come on. They read the book.

很明显。是的。

Clearly. Yeah.

有本很棒的书叫《量化交易员》，出版时间稍早，大概是2011年，所以不如《操纵市场的男人》更新。书里只有几章讲到RENTECH，但内容不错。另外还有一篇2016年彭博社的好文章，我们会附上链接。我认为在那之前，《量化交易员》算是第一本真正提及RENTECH的出版物。

There's a great book called the Quants, which is a little bit earlier. I think it's 2011, so it's not as updated as the man who sold the market. And there's only sort of a couple chapters about RENTECH, but some good stuff in there. And then there's a good Bloomberg piece from 2016 that we'll link to that. I think between that and the quant, it was sort of the first time there was really anything at all that was published about RENTECH.

这些都会放在节目注释里。还要感谢其他人，大卫。

So all those will be in the show notes. Other people to thank, David.

还要感谢霍华德·摩根，我们采访了他，听他讲述第一轮融资的历史非常有趣。当然还有Rentec的创立、与吉姆的合作、互相投资彼此的基金等等，都很有意思。本，你提到的布雷特·哈里森现在正在创建Architect——我超爱这个项目。

Other people to thank Howard Morgan, who we spoke to, which was so fun to get a bunch of the first round history from him. And then, of course, the, you know, founding of Rentec and partnering with Jim and investing in each other's funds and all that. So fun. Brett Harrison, who you mentioned, Ben. Brett is now building architect, which I love this.

这个世界太需要这样的产品了。这是二十一世纪的盈透证券。用过盈透的人都知道其中的机遇。布雷特，谢谢你。还有我之前交谈过的马修·格雷纳德。

This is so needed in the world. It's the Interactive Brokers for the twenty first century. Well, anybody who uses Interactive Brokers knows exactly the opportunity there. So thank you, Brett. And then Matthew Grenade, who I spoke with.

马特是Domino Data Lab的联合创始人，这家由红杉资本等多家机构支持的企业级AI运维平台非常出色。它帮助模型驱动的企业和产品加速研究、增强协作、快速交付新机器学习模型，正是我们与Rentec讨论的这类技术。马特在创立Domino前来自量化领域，曾在Point72和桥水基金任职——虽然桥水自成一派，但他在两家机构都担任过长期高管。他对行业格局及Rentec的定位提供了极具洞见的分析。

Matt is the cofounder of Domino Data Lab, which is a great enterprise AI ops platform backed by Sequoia and many others. It allows model driven businesses and products to accelerate research, increase collaboration, rapidly deliver new machine learning models, all of the sorts of things that we were talking about here with Rentec. Matt, before starting Domino Data, came out of the quant world. He was at point seventy two and Bridgewater, which isn't really quant sort of its own thing, but he was a longtime senior employee at both of those firms. And he gave us great, great perspective on the landscape of everybody out there and where Rentec fits in.

太棒了。如果喜欢本期节目，可以去听听我们几年前做的伯克希尔·哈撒韦系列，那是完全不同的投资风格。订阅节目邮件请访问acquire.fm/email，我们会分享每期节目发布后的新发现，包括听众指正。在任意播客平台搜索'Acquired'即可订阅，最新一期我们采访了利拉鲁肽（后发展为司美格鲁肽，即诺和泰、维戈维等药物）研发团队的领军人物。

Awesome. Well, if you liked this episode, you should check out our Berkshire Hathaway episodes from a few years ago for a very different style to investing. You can sign up for new episode emails at acquire.fm/email. We'll be including little tidbits that we learn after releasing each episode, including listener corrections. You can listen to a c q two, search and subscribe in any podcast player, and listen for our most recent episode with the, well, really creator or person who led the team that created Lyri Glutide, which went on to become Semaglutide, which of course is Ozempic, Wegovy, etcetera.

没错，所有现代GLP-1类药物都源于此。

Yep. All modern GLP ones.

诺和诺德的拉塔·比尔努森，能请到她真是太好了。听完节目后，欢迎来acquired.fm/slack与Acquired社区的聪明人们讨论。想要周边商品？acquired.fm/store有售。

Lata Biernudsen from Novo Nordisk. Was awesome to have her on the show. And after you finish this episode, come talk about it with other smart members of the Acquired community at acquired.fm/slack. If you want some merch, we've got some. Acquired.fm/store.

听众朋友们，我们下次见。

And with that listeners, we'll see you next time.

下次见。

We'll see you next time.