Right When Everyone Else Is Wrong

In servicing these billion-dollar fundamental hedge funds, what are the use cases that they see for the LLMs?

在服务这些十亿美元级的基本面对冲基金时，他们觉得 LLM 有哪些用例？

You can scrape discussion boards, you can scrape Discord channels, you can scrape Reddit, you can scrape all sorts of podcast transcripts, and summarize all that information.

你可以爬论坛，可以爬 Discord 频道，可以爬 Reddit，可以爬各种播客的文字稿，然后把这些信息全都总结出来。

What do you think about the intellectual laziness that can come up from using these tools?

用这些工具可能带来的思维懒惰，你怎么看？

I do think there are differences in the work ethic of fundamental analysts.

我确实觉得，基本面分析师之间的勤勉程度是有差别的。

Those investors who are very curious, this is their biggest dream come true.

好奇心极强的那批投资者，这简直是他们最大的梦想成真。

The good ones are just going to keep coming up with more and more and more questions.

厉害的那批人只会不停地冒出越来越多的问题。

The downside of hanging out with billionaires is you kind of want to become one, too.

跟亿万富翁混在一起的坏处是，你多少也会想成为其中一员。

Matei, thanks for coming on the pod.

Matei，谢谢你来上这档播客。

Hey Ethan, yeah, it's great to be here.

嘿 Ethan，很高兴来到这里。

Thanks for having me.

谢谢你邀请我。

What is the single biggest constraint

最大的那个瓶颈是什么——

that fundamental hedge funds have with their data science layer?

基本面对冲基金在数据科学层上遇到的？

I think it boils down to talents.

我觉得归根结底是人才。

The space is evolving so quickly, especially in the world of AI, that it is hard to keep up with all the changes.

这个领域演进得太快，尤其是 AI 这一块，快到很难跟上所有变化。

So, finding the right people to be able to understand the technology, and in particular how the technology applies to fundamental investing, is it was always hard, but it's gotten even harder given the the pace of all this all this change.

所以，要找到合适的人——能看懂这项技术、尤其能看懂它怎么用在基本面投资上的人——本来就一直很难，而在眼下这种变化节奏下，变得更难了。

And so, give me some context.

那给我讲点背景。

Let's say I'm sitting in the seat of a of a fundamental hedge fund, right?

假设我坐在一家基本面对冲基金的位置上，对吧？

Why can't I just hire an internal, you know, a team of like two guys with Claude Code, spin up a couple projects, you know, what is the need for Where is the need coming from for the product you guys offer?

为什么我不能就在内部招个团队，比如两个人配上 Claude Code，起几个项目——你们提供的这个产品，需求到底从哪来？

I'm talking to one of the founders of one of the one of our clients, and he is extremely bullish on all the AI developments out there.

我最近在跟我们一个客户的创始人聊，他对现在 AI 的各种进展极度看好。

And he was telling me how he was using Cohere to and to ask a question.

他跟我讲他怎么用 Cohere 去问一个问题。

The question was about some market share of certain certain companies.

问题是关于某几家公司的市场份额。

And they give he types in the answer, he gets a response, and it looks phenomenal.

他把问题打进去，拿到一个回答，看上去棒极了。

Until he looks at the actual charts that it was showing.

直到他去看它给出的那些实际图表。

And the market share was exactly equal across all the companies in that in that space.

那个领域里所有公司的市场份额完全相等。

And he asked the follow question, like, wait, why are all the market shares equal in this in this chart?

于是他追问：等等，为什么这张图里所有的市场份额都是相等的？

And then the the GPT's response was, oh, because I didn't have any data, so I just made them all equal.

然后 GPT 的回答是：哦，因为我没有任何数据，所以我就把它们都设成相等了。

So, it was it worked perfectly until you realized that it just gave you a [ __ ] answer.

所以一切完美无缺——直到你发现，它就是给了你一个 [ __ ] 答案。

So, one of the biggest problems in like right now with with AI as it relates to fundamental investing is that it's obviously hallucinations have have always been been an issue, but in our world, accuracy is paramount.

所以眼下 AI 用在基本面投资上最大的问题之一是，幻觉显然一直都是个问题，但在我们这行，准确性压倒一切。

And in particular, the way I I think about what we do and why what we're doing is is ultimately helpful is that to make money in the the investing world is you need to be right, but not only do you need to be right, is you also need to be right when everybody else is wrong.

尤其是，我这么看我们做的事、以及它最终为什么有用：在投资这行赚钱，你得判断对，但光判断对还不够，你还得在其他所有人都错的时候判断对。

If you're right, but so is everyone else, there's no alpha.

如果你对了，但其他所有人也对，那就没有 alpha。

Like everyone knows the same information, so you're not going to make any money.

大家掌握的是同样的信息，所以你赚不到钱。

The only time you can actually make money is if you're right and everyone else is missing is missing something.

你真正能赚到钱的唯一时刻，是你对了，而其他所有人都漏掉了什么。

In other words, in the the in the context of of math, you were looking for outliers.

换句话说，放到数学的语境里，你要找的是异类。

There's an outlier in some data set that sparks your attention.

某个数据集里有一个异类，引起了你的注意。

And if you're if you if you notice that outlier, and if you're right that it is a real outlier, and everyone else is is missing it, then you make a lot of money.

如果你注意到了那个异类，而且你判断对了它是一个真实的异类，其他所有人都漏掉了它，那你就能赚很多钱。

But the problem with outliers is that they're often wrong.

但异类的问题在于，它们往往是错的。

There's so many issues in the data that can cause that outlier to happen.

数据里有太多毛病，都可能造出这么一个异类。

And so, when you need the data the most is when you're noticing an outlier, but that's also when it's least reliable.

所以，你最需要数据的时候，正是你注意到一个异类的时候，而那也正是数据最不可靠的时候。

Like you're you you have the most concern about the the accuracy of that of that data.

就是说，你对那份数据准不准最不放心。

And so, if you now bring this back to the the AI world, you almost like by by by definition, like if the AI is telling you something interesting, something novel, then okay, like I wasn't expecting that, that could be actionable.

所以，把这一点带回 AI 的世界，几乎是按定义如此：如果 AI 告诉你一件有意思的、新鲜的事，那好，我没料到这个，这可能就是能下手的机会。

But that's also when you should have the most concern about it, like whether this was real or whether it's a hallucination.

但那也正是你该最不放心的时候：这到底是真的，还是一个幻觉。

And at the end of the day, it's only by getting comfortable with these outliers, and noticing that they're real when other people think that they're they're mistakes or other people might not have noticed it at all, that you can actually make money.

说到底，只有当你能对这些异类感到踏实，能在别人以为那是错误、或者别人根本没注意到的时候看出它们是真的，你才真的能赚到钱。

So, that brings us back to the the point about having actual like human expertise in going in and knowing the right follow-up questions to ask, knowing how to ultimately get to the the point about improving the the data to the extent that you can until you reach enough confidence given the amount of time and resources that you have, cuz by the way, we don't have infinite time or money to uh to understand each question fully.

所以这就把我们带回到那一点：得有真正的人类专业判断，进去之后知道该追问哪些对的问题，知道最终怎么把数据改进到你力所能及的程度，直到在你手上的时间和资源之内攒够信心——顺便说一句，我们并没有无限的时间和金钱去把每个问题都彻底搞懂。

So, at some point you just have to call it.

所以到某个点上，你就只能拍板。

And in relying on that decision to figure out what's what's happening.

然后靠那个判断去弄清楚到底在发生什么。

So, that's why it's really hard to to use AI in this investment context.

所以这就是为什么在投资这个语境下用 AI 真的很难。

I want to touch on AI more broadly before we go into what you've said, because one of the things that you touched on was that in order to make money in the hedge fund world or in the markets, you need to be right when everyone else is wrong, right?

在深入你刚才说的之前，我想先更宽泛地聊聊 AI，因为你提到的一点是：要在对冲基金的世界里、或者在市场上赚钱，你得在其他所有人都错的时候判断对，对吧？

That's alpha, per se.

这本身就是 alpha。

Um just on generative AI for a second,

就生成式 AI 说一下，

doesn't everyone using these models lead to a lack of a differentiated view?

所有人都在用这些模型，这难道不会导致缺乏差异化的观点吗？

Or can you boil down how these LLMs can be used to enhance or or capture a differentiated view that one might have?

或者你能不能提炼一下，这些 LLM 怎么用，才能强化、或者抓住你本来可能就有的那个差异化观点？

Yeah, I mean, so if you look at how the LLMs work, you have you're trained on sequences of tokens, and it predicts the next token in in a sequence.

是这样，如果你看 LLM 是怎么运作的，它是在 token 序列上训练的，然后预测序列里的下一个 token。

But it also responds to your context window.

但它同时也会响应你的上下文窗口。

So, the what kind of input what kind of prompt you gave it, as well as increasingly all the other history of prompts that that it has accumulated on you.

也就是你给它什么样的输入、什么样的 prompt，以及越来越多地，它在你身上积累下来的全部历史 prompt。

And so, what you'll what you'll notice, and this is how these models get better, right?

所以你会注意到——这也正是这些模型变好的方式，对吧？

They they know you now, and so it knows the types of they knows about my company, for example, it knows about System2.

它们现在了解你了，所以它知道——比如它知道我的公司，它知道 System2。

So, when I just ask something about work, it has the context to realize that it's talking about about System2.

所以我随口问一个跟工作有关的问题时，它有上下文，知道说的是 System2。

What is up happening is actually much more customization than than you'd expect, whereas previous like, you know, the technologies relied on that you even if you if you look at how recommendation engines work, it was more about what other people are doing, and it kind of was the the law of averages that got people into the the same spot, like the most watched shows on Netflix, like all these things that rely on on people seeing the the the same thing.

实际发生的事情是，定制化的程度远超你的预期；而以前的技术依赖的是——哪怕你去看推荐引擎是怎么运作的，它更多是看别人在做什么，是某种平均法则把所有人带到了同一个位置，比如 Netflix 上观看最多的剧集，所有这些东西都依赖于人们看到同样的内容。

Whereas with these models, your context creates a bubble that is very unique to you.

而在这些模型这儿，你的上下文会造出一个非常专属于你的气泡。

So, I actually think that the the AI is much more differentiated than previous technologies.

所以我其实认为，AI 比以前的技术要差异化得多。

So, it's not going to result in the same consensus answer.

所以它不会导向同一个共识答案。

Like you and I can have the same prompt in there, and we will end up with different results depending on the the context and queries of our entire history with that with that product.

你和我可以输入同样的 prompt，最后拿到的结果却不一样，取决于我们各自跟那个产品的全部历史所积累的上下文和查询。

So, that like, you know, this it's a feature, it's not a bug.

所以这是特性，不是 bug。

But do you want this?

但你想要这个吗？

You want the the model to really understand you and to respond to you the way you like to be to be spoken to.

你希望模型真正理解你，用你喜欢的说话方式回应你。

But at the same time, that can lead to a false sense of precision.

但与此同时，这也可能带来一种虚假的精确感。

Going back to the outlier point, because it can say something that like aligns with what you already believe, and because, you know, you there are different philosophies of investments.

回到异类那个点：它可能说出正好跟你原有信念相符的东西，而投资本来就有不同的哲学。

We have a bunch of investors who love to be contrarian, there are investors who like growth companies.

我们有一批投资者热衷于做逆向，也有投资者喜欢成长型公司。

Whatever your your your thesis, your favorite thesis is, you you like to go and find those kinds of stories, and these models will reveal some of those stories to you, and they'll be like, oh, holy [ __ ] this really fits all the criteria in my investment playbook of what makes an investment interesting.

不管你的投资逻辑、你最偏爱的那套逻辑是什么，你都喜欢去找那一类故事，而这些模型会把其中一些故事揭示给你，然后你会想：哦，我 [ __ ]，这完全符合我投资打法里对“什么才算有意思的标的”的所有标准。

But it might not be real, cuz it just feeds you the things that it knows that you like.

但它可能不是真的，因为它只是把它知道你喜欢的东西喂给你。

And so, you will have this false sense of confidence in a result that's not necessarily tied to data.

于是你会对一个未必与数据挂钩的结果，生出这种虚假的信心。

And like we can talk about like how to actually try to ensure that the results are based on real data and and real analyses that you have done instead of this pattern matching that is getting better better at understanding what you like, that doesn't necessarily result in in the in an accurate response.

我们可以聊聊怎么真正确保结果是建立在真实数据、和你自己做过的真实分析之上，而不是建立在这种模式匹配之上——它越来越懂你喜欢什么，却未必能给出准确的回答。

And by the way, this isn't just unique to to the models.

顺便说一句，这并不是模型独有的问题。

I was talking to a former partner at a very large hedge fund who's who's become a mentor to me, and he was saying that one of the problems that you face in the executive teams of these hedge funds is that as the founders become more and more successful, they start being treated like gods by the rest of the investment team.

我跟一家非常大的对冲基金的前合伙人聊过，他后来成了我的导师，他说这些对冲基金的高管团队里会遇到的一个问题是：随着创始人越来越成功，投资团队的其他人开始把他们当神一样供着。

And it turns into an echo chamber where people are afraid to say no to them.

于是就变成一个回音室，没人敢对他们说不。

So, you're surrounded, like if you're the founder of one of these places, and you've achieved success, right?

所以你被这样围着——假设你是这类机构的创始人，而且已经做出了成绩，对吧？

So, you're you already think you're you're great cuz you've done well yourself.

你本来就觉得自己很厉害，因为你自己干得确实好。

And now everybody is telling you things exactly the way that you want to hear them.

而现在所有人对你说的话，都完全是你想听的样子。

It leads to that issue, because people can't actually speak truth to power.

这就导致了那个问题，因为没人真的敢向权力讲真话。

And this is again, this is coming from um this yeah, like a a partner or CEO of of a fund that has been in those in those meetings, and has seen that first hand, is what it's like to to be in this kind of echo chamber.

再说一次，这话是出自一位基金的合伙人或者说 CEO，他坐在过那些会议里，亲眼见过身处这种回音室是什么感觉。

And so, it's not just in like the context windows of LLMs that produce these echo chambers.

所以，产生这些回音室的不只是 LLM 的上下文窗口。

Like this is a phenomenon that happens in real human organizations, and it is problematic.

这是真实的人类组织里就会发生的现象，而且很成问题。

You have to actively try to combat it and in in to continue to to succeed, and not let your previous success now start hampering it.

你必须主动去对抗它，才能继续成功，别让过去的成功反过来成了阻碍。

In servicing these billion-dollar fundamental hedge funds, what are the use cases that they see for the LLMs?

在服务这些十亿美元级的基本面对冲基金时，他们觉得 LLM 有哪些用例？

What's at the forefront of their mind?

他们最挂心的是什么？

It's a lot.

确实不少。

Yeah, a lot of it is I mean it's it's really varied.

对，其中很多东西真的五花八门。

So, we are working with clients to for example to do a better job synthesizing qualitative information.

比如说，我们跟客户合作，帮他们把定性信息综合得更好。

So, before the alternative data revolution that started maybe I don't know 10 15 years ago, uh the well, the primary research methodologies was talking to experts.

在大概 10 到 15 年前那场另类数据革命之前，主要的研究方法就是找专家聊。

And having these hour-long conversations with various experts and then you form an investment opinion.

跟各路专家做一场场一小时的对话，然后形成投资观点。

And by the way, that is still very much alive and it's a it's a big business within fundamental investing.

顺便说一句，这套做法今天依然很活跃，在基本面投资里是一门大生意。

Well, enter AI.

然后，AI 登场。

Now, you can do so many so much more using the same conversation with with an with an expert that can leverage LLMs on a scale that hasn't been been able to to be done before.

现在，还是跟专家的同一场对话，你能做的事情多得多——因为可以借助 LLM，规模是以前做不到的。

So, for example, you can do things like Well, one of the challenges is speaking to one of our one of my my friends who was who has a consumer fund, his point was that I would do a lot more expert interviews.

比如说，你可以做这些事——我跟一个管消费基金的朋友聊，他讲了一个挑战：我其实愿意做多得多的专家访谈。

The constraint isn't money.

约束不是钱。

The constraint is I I just don't have time to talk to more people because I'm working on so many other things.

约束在于我实在没时间跟更多人聊，因为手上还有太多别的事。

So, maybe I'll do five 10 conversations with with experts on a particular project.

所以一个具体项目上，我可能只做 5 场、10 场专家对话。

I would love to do more.

我很想多做一些。

I just can't clone myself.

但我没法克隆自己。

Enter AI.

AI 登场。

And what if you had an LLM that was trained on your interview techniques, that understands the the main goal, the thesis, the types of questions that you want to answer for that particular project, and then conducts these interviews on your behalf with with a bunch of experts, and then it can also synthesize that information after the fact?

那么，如果你有一个 LLM，拿你的访谈技巧训练过，懂得主要目标、投资论点、你在那个项目上想回答的问题类型，然后代替你去跟一批专家做访谈，事后还能把这些信息综合起来呢？

And what if you can then say, "Okay, great. I got I did my my first, you know, couple of calls to get a primer on a new industry or a new type of company. But now I understand it, but I have a couple of specific a very specific follow-up questions based on some of those conversations."

而如果你接着能说：“好，很好。我打了头几通电话，对一个新行业、一类新公司算是入了门。现在我懂了，但基于其中几场对话，我又有了几个非常具体的追问。”

Can we do that at scale?

这件事能不能规模化地做？

Instead of just spending an hour on on a call, can I talk to somebody for 15 minutes and just drill in on that one specific follow-up question?

与其在一通电话上耗掉整整一小时，我能不能只跟人聊 15 分钟，就盯着那一个具体的追问往深里钻？

It's almost like creating the the mosaic in this like fog of war.

这几乎就是在战争迷雾里拼出那幅拼图（mosaic）。

You're trying to to understand something.

你想搞懂一件事。

All of a sudden, okay, you got a clearer picture, but there's like spots across the map that are unclear to you.

忽然之间，好，画面清楚了一些，但地图上还有几块你看不清。

It's like, "What if I just focus on that? And I want to expand that a little bit. And I want to expand that a little bit."

于是就是：“那我干脆只盯着那一块？把这块再往外推一点，那块也再往外推一点。”

Obviously, overlay all this with with data, but these these changes in both the ability to have an agent conduct an interview on your behalf, as well as changes that we're now seeing in the recruitment of experts, the payment of experts.

当然，这一切还要叠上数据。但这些变化——一边是 agent 能代你做访谈，一边是我们现在看到的专家招募、专家付费方式的变化。

You can now it becomes a lot easier to pay somebody or some of these technologies that you can cobble together.

现在你可以……给一个人付钱变得容易多了，也有一些技术你可以自己拼起来。

Instead of relying on these very large expert networks, which are still like I think they're doing a great job.

而不用去依赖那些体量很大的专家网络——虽然我觉得它们做得挺好。

We have many friends there.

那些机构里我们有不少朋友。

I don't want to uh you know, dunk on their business model, but it is it is becoming increasingly easier to decompose that process, to have companies that just focus on on sourcing people, then having a payment overlay on top of it, having the AI interviewer on top of that, having those conversations, you know, recording, transcribed, summarized, and ultimately integrated into the the broader research process.

我不想去踩人家的商业模式，但把这个流程拆开确实越来越容易了：有的公司只专注找人，上面叠一层支付，再往上叠 AI 访谈员，把这些对话录音、转写、总结，最后整合进更大的研究流程。

So, just like one example of one area of fundamental research that's been going on for decades that is now at the precipice of being revolutionized by some of these language models.

所以这只是一个例子——基本面研究里一个做了几十年的环节，如今正站在被这些语言模型彻底改造的临界点上。

Um and there are many many more examples.

还有非常非常多别的例子。

I mean, we're working on projects where you want to stay on top of it thing within this this the the context of of language and qualitative data just from from language.

我们在做的一些项目，是要你持续盯住某件事——都在语言、以及纯粹来自语言的定性数据这个范畴里。

You can scrape discussion boards.

你可以爬讨论区。

You can scrape Discord channels.

你可以爬 Discord 频道。

You can scrape Reddit.

你可以爬 Reddit。

You can scrape all sorts of like podcast transcripts and and summarize all that information and have it synthesized into a system that then correlates that to the companies I'm invested in or the themes I'm interested in and have that all be part of this broader piece of of information that you're collecting from from ultimately from from people on the internet or directly on on the phone.

你可以爬各种播客转写稿，把这些信息都总结出来，综合进一个系统，再由它关联到我持仓的公司、我关心的主题上，让这一切成为你收集的那一大块信息的一部分——这些信息最终来自互联网上的人，或者电话那头的人。

I want to push back or um I guess they've, you know, explore further what you said about the expert calls example.

我想反驳一下——或者说，就你讲的专家电话那个例子再往下挖一挖。

Um cuz I can 100% see the value, you know, you can absolutely get more volume in.

因为价值我百分之百看得到，量确实能做上去。

But what do you think about the say intellectual laziness that can come up from using these tools?

但用这些工具可能带来的、姑且叫智识上的懒惰，你怎么看？

And in my head, I'm thinking more of junior talent, right?

我脑子里想的更多是初级人才，对吧？

Cuz if you haven't been trained the analog way of putting in the reps, you know, uh like, you know, 10 hours of calls on a Tuesday, right?

因为如果你没经历过那种老派的训练方式、没有实打实地把量练够——比如一个周二打 10 个小时电话，对吧？

If you haven't been trained that way, how would you how would you get the context to use these LLMs?

如果没受过那样的训练，你要从哪儿拿到用这些 LLM 所需要的上下文？

And I guess my real question is, yeah, where do you see the use for these tools when they can easily kind of destroy one's skill set?

我真正想问的是：这些工具很容易毁掉一个人的功底，那你觉得它们的用武之地在哪？

Yeah.

对。

Yeah, so going back to how these models are trained.

对，那我们回到这些模型是怎么训练的。

This is actually is interesting to to spend a little bit of time understanding the the way the model works.

花一点时间搞懂模型怎么运作，其实挺有意思的。

So, that the first step of training an LLM is you're taking a corpus of all the text on the internet.

训练一个 LLM 的第一步，是拿到互联网上所有文本的语料。

And you're essentially predicting you have a long string of of tokens.

然后你本质上是在做预测——手上有一长串 token。

A token is essentially a word, although it's not really a word.

token 本质上就是一个词，虽然它其实并不真是一个词。

It could be like a couple of words, whatever.

它可能是几个词，或者别的什么。

But you're predicting the next token in that in that sequence.

但你要做的是预测这个序列里的下一个 token。

And so, if you and I are having this conversation and I stop right here, the LLM will fill in some sentence that will look very similar to the sentence that I just uttered.

所以，如果我们俩正在聊天，我说到这儿停住，LLM 会补上一句跟我刚说完那句非常像的话。

Great.

很好。

So, now you have that.

好，这一步有了。

But how do you go from from that token prediction into the format that we know, which is a question and answer format, where you're giving it a prompt and then it spits out a question?

但你怎么从这种 token 预测，走到我们熟悉的那种形式——问答形式，你给它一个 prompt，它吐出一个问题？

Because that's not that's not what what the model does.

因为模型本身干的并不是这件事。

Again, if if if it were to to do that, it would continue with this like question and answer format, where like it would it's trained on a bunch of interviews before and so it says, "Okay, like yeah, you asked me a question, I give a response."

再说一次，如果它真那么干，它会顺着这种问答形式一路接下去——它之前在一堆访谈上训练过，所以它会说：“好，你问我一个问题，我给一个回答。”

And it won't stop now.

而它到这儿是不会停的。

It will then fill in some question for you.

它接着会替你补上一个问题。

And so, it continues on and on.

于是就这样一直往下滚。

So, there's a post-processing step in these in these LLMs that it's the thing that the first step is to predict the next tokens in in the sequence.

所以这些 LLM 里有一个后处理步骤：第一步是预测序列里的下一批 token。

But then you can train it on question and answer interviews to say, "Okay, this is a question. Now, an an answer came from it."

但接着你可以用问答式访谈来训练它，告诉它：“好，这是一个问题。接下来，由它引出一个回答。”

And this is this was a um this was done using data that was created by by humans, where you have like people putting in prompts, people answering those prompts, and it was basically like fed in a a corpus of questions and answers to then learn what it's like to respond to a question.

而这……这一步用的是人写出来的数据：有人去写 prompt，有人去回答这些 prompt，基本上就是喂进一份问答语料，让它学会该怎么回应一个问题。

And this data set, by the way, is order or magnitude like smaller than the initial training data set.

顺便说一句，这个数据集比最初的训练数据集要小一个数量级。

So, the initial training data set creates these linkages in the brain, and then you're teaching it, "Okay, now you know all these different concepts. I'm now going to teach you what it's like to have a question and answer format."

所以最初的训练数据集在大脑里建立起这些连接，然后你再教它：“好，现在这些概念你都懂了。我接下来教你问答形式是怎么回事。”

And that's how the the models ultimately like respond to what we're doing here, where you're putting in a prompt and it's it's bringing in um a response.

这就是这些模型最终能像我们现在这样应答的原因——你输入一个 prompt，它给出一个回应。

Because it's like it's you can think of it as a corpus like a training set that has a bunch of text, but the text is question answer, question answer, question answer.

因为你可以把它想成一个语料、一个训练集，里面有一堆文本，但这些文本是问、答，问、答，问、答。

Whereas the internet is like one long string of all sorts of different things.

而互联网是一长串什么都有的东西。

And the model gets really good from just a small set of questions and answers to realizing, "Okay, this is what these these guys want. They want an answer every time there's something that ends in a in a question mark."

而模型只靠一小批问答，就能很快明白过来：“好，这些人要的是这个。每当有东西以问号结尾，他们就想要一个回答。”

So, going back to the the point that you're you're making about you will you'll leverage some of the insights that the or the techniques that these these funds have, well, it's very similar to this like post-processing step in an LLM training model, where you can feed in your own transcript library of like in the a lot of these libraries a lot of these transcripts get recorded.

回到你刚才提的那个点——你会借用这些基金已有的洞见或者说技巧——这其实跟 LLM 训练里的后处理步骤很像：你可以把自己内部的转写库喂进去，这类库里很多对话都是有记录的。

So, you have internally, let's say like, you know, Matei's hedge fund has a history of conversations that my analysts have had with experts over the years, and I can feed that in.

所以内部就有——比方说，Matei 的对冲基金存着我的分析师这些年跟专家做过的所有对话，我可以把这些喂进去。

But now, instead of training it on a question and it spits out a response, I now would train it vice versa, where I give it a response from an expert and train it on what's the next follow-up question that should be asked.

但现在，不是给它一个问题、让它吐出一个回答，我要反过来训练：给它一段专家的回答，训练它给出下一个该问的追问。

So, it's almost like the the reverse of these LLMs.

所以这几乎就是这些 LLM 的反向操作。

I want to generate the next question.

我想生成的是下一个问题。

I don't want to generate the next response.

我不想生成下一个回答。

But I have a training data set that I can use in order to to to do that.

而我手上正好有能拿来做这件事的训练数据集。

And so, it's again very right up the alley of how these models are are trained, and it they don't need a large corpus of information to do this like post-processing step.

所以这又完全走的是这些模型训练的路子，而且做这个后处理步骤并不需要庞大的语料。

Um but it yeah, it requires a little bit of work, and that's some of the work that we're helping our our clients with is going in, using their internal data sets to like fine-tune models in in ultimately learning how to ask a great next question, just like their internal analysts have been doing for for years.

嗯，但确实要花点功夫，我们帮客户做的一部分工作就是这个：进去用他们的内部数据集微调模型，最终学会把下一个问题问得漂亮，就像他们的分析师这些年一直在做的那样。

And each client is going to have a different fine-tuned model.

而且每个客户的微调模型都不一样。

Each Each client has a different style of ask asking these kinds of of questions.

每个客户问这类问题的风格都不同。

But the point is that you can actually get get better.

但关键是，你确实能变得更强。

You can use these these LLMs and and train them and get them to be phenomenal question uh interviewers of of experts.

你可以拿这些 LLM 来训练，把它们练成极出色的专家访谈者。

That's one way that they're applied at fundamental hedge funds.

这就是它们在基本面对冲基金里的一种用法。

Uh yeah, I'd call that a part of the research process, right?

嗯对，我会把那算作研究流程的一部分，对吧？

What would you say are the other layers in which they can be applied?

那你觉得还有哪些层面能用上？

Is it every single layer um on the portfolio management side, you know, the top guy looking down at the portfolio?

是每一层都能用吗——比如组合管理这一侧，那个居高临下盯着整个组合的老大？

Um get Can it be used there?

那儿能用上吗？

I'm curious to hear it from your perspective.

我很想听听你的看法。

Yeah.

对。

So, I mean, there's there's so much innovation happening.

现在正在发生的创新实在太多了。

I was um I was just catching up with an old friend who I would spend like, I don't know, two decades at um at D.E. Shaw.

我刚跟一个老朋友叙旧，我们俩在 D.E. Shaw 大概共事了……我也说不好，二十年吧。

And he's building some really interesting fundamental overlays on top of of this of of data.

他在这些数据之上做了一些很有意思的基本面叠加层。

And and part of it, so when

而其中一部分，所以当

you think you were talking about what is the job of a fundamental portfolio manager?

你想想，你刚才谈的是：基本面 PM 的工作到底是什么？

When you really boil it down, it is to come up with a a thesis about how not how how the world works, but how this company works in the context of the the broader world.

说到底，就是要形成一个论点——不是关于世界如何运转，而是这家公司放在更大的世界里如何运转。

And you're looking at If you If you think about this in terms of I don't know, like it could be argued in the context of of a graph where you have different nodes in a graph.

而你在看的是——你可以，怎么说呢，把这件事想成一张图谱，图谱上有一个个不同的节点。

And you have the the companies that are related to products, related to competitors, to markets.

图谱上有公司，公司跟产品相关、跟竞争对手相关、跟市场相关。

And you have these connections between all these different entities and and concepts.

而所有这些不同的实体和概念之间，都有连接。

And you're trying to to figure out how everything fits together as part of this I give people talk about the the mosaic.

你要弄明白的是，这一切怎么拼成一个整体——也就是人们常说的 mosaic（拼图式研究）。

How does the the mosaic actually come together to explain what's happening here?

这块 mosaic 到底怎么拼起来，才解释得了这里正在发生的事？

And it does so in a causal way, by the way.

而且顺带说一句，它是靠因果关系做到这一点的。

It's not just like you're looking at correlations in in data, which is how a quant would would look at investing.

这不只是在数据里看相关性——量化的人才是那样看投资的。

You're looking at trying to find causal relationships between different concepts and and different different entities.

你要找的是不同概念之间、不同实体之间的因果关系。

And most fundamental PMs I just do that in their heads.

而大多数基本面 PM 就是在脑子里做这件事。

They're like, "Okay, I'm I'm interested in some some stock and has exposure to it's I don't know, it's like a consumer name. And it's it's related to you have like tariffs that might be impacting the demand for for those products."

他们会想：「好，我对某只股票有兴趣，它有敞口，比如它是个消费类标的。而它又牵扯到——比如关税可能正在影响这些产品的需求。」

So, that's one angle that you're you're exploring.

所以这是你在探索的一个角度。

But if you think about it in the context of like you have something like a database, you have an entity of a company, you have like this potential shock of a of a tariff, and what does that do to the company?

但你换个角度想，把它想成一个数据库：有「公司」这么一个实体，有关税这么一个潜在冲击，那它对这家公司会造成什么？

We can expand that in a more systematic way to try to identify like all influences on all companies.

我们可以用更系统的方式把它扩展开，试着把所有公司身上的所有影响都识别出来。

And and then leveraging AI to to fill in some of these some of these gaps.

然后再借助 AI 去填补其中的一些空白。

But it's almost like you're you're layering your own framework.

但这几乎就是在叠加你自己的一套框架。

A fundamental investor layers their own framework onto the world.

基本面投资者把自己的框架叠加到这个世界之上。

And then they And then they look for for data to to figure out how those different parts of the the graph communicate with each other.

然后他们再去找数据，弄清楚图谱里那些不同的部分之间是怎么互相传导的。

So, there's a lot of like overlay like structure that you can put on top of the data to to help you facilitate your your investment thesis.

所以，有很多结构是可以叠在数据之上的，帮你把投资论点撑起来。

Um but then there's also So, once you have that, then you spend a lot of time trying to figure out the data behind each one of those particular nodes.

但还有一点——有了这个之后，你会花大量时间去搞清楚每一个节点背后的数据。

And it becomes really interesting when you're you're looking at not just like first-order effects.

而真正有意思的地方在于，你看的不只是一阶效应。

Like all right, like a tariff impacting a company that imports a lot of goods, that's easy.

比如关税影响一家大量进口商品的公司，这很容易。

But what are like second-order effects that you might not even realize exist, but they but they do?

但那些你甚至意识不到它存在、却确实存在的二阶效应呢？

And so, like I'm one of the classic examples here is let's say you're going back to the the same consumer example.

这里有个经典的例子——还是回到刚才那个消费的例子。

There might be a second-order effect between the price of oil and the sales of clothing.

油价和服装销售之间，可能存在一个二阶效应。

Like, what does that have to do with with each other?

这两者能有什么关系？

It's like, "Well, it could be that this particular clothing company has a big presence in Texas.

答案可能是：「也许这家服装公司在得州布局很重。

And Texas has a lot of people that work in the oil industry.

而得州有很多人在石油行业工作。

And so, if the oil industry is not doing well, those people are not going to have money to go spend on some whatever like high-end apparel brand that happens to be over-indexed to Texas."

所以石油行业不景气，这些人就没钱去买某个刚好在得州过度集中的高端服饰品牌了。」

So, if you have a data set of all the store locations of all the the companies in your universe, and then you have a macro data set or like a labor data set of what's happening in those different geographies, you might be able to notice these like second- or third-order effects of say, yeah, like there's some macro shock that's causing unemployment or it's causing bonuses to go down in certain parts of the the country that then are over-indexed to certain types of stores or to be certain types of restaurants.

所以，如果你有一份数据集，涵盖你股票池里所有公司的所有门店位置，再有一份宏观数据集、或者说劳动力数据集，告诉你这些不同地区正在发生什么，你就可能注意到这类二阶、三阶效应——比如某个宏观冲击导致失业，或者导致这个国家某些地区的奖金下降，而这些地区又过度集中着某类门店、某类餐厅。

And so, if you have these these kinds of And some some of the the you know, the the great investors can think about this.

所以，如果你有这类——当然，有些顶尖投资者是自己就能想到这一层的。

And they'll just sort of be like, "Oh, yeah, I think this is a potential risk.

他们大概会这么说：「哦对，我觉得这是个潜在风险。

Like I'm seeing this thing happening."

我看到这件事正在发生。」

And they they connect the dots between what's happening in the the world and their and their the portfolio company.

他们能把世界上正在发生的事和自己的持仓公司连起来。

But you can also do this at scale using using these these types of data systems to infer connections between all sorts of different entities.

但你也可以用这类数据系统大规模地做这件事，去推断各种不同实体之间的连接。

And then if you have a forecast on say, yeah, like employment in the oil industry, to then see how that scenario can like cascade to what the impact is on your entire portfolio of companies that have nothing to do with oil.

然后，如果你对石油行业的就业有一个预测，就能看到这个情景怎么层层传导，最后对你整个组合里那些跟石油毫无关系的公司造成什么影响。

And I guess going back to I think an earlier question I had, what do you think about the intellectual laziness that can come from this?

回到我前面提过的一个问题：你怎么看这里面可能出现的智识懒惰？

So, I mean, I'm I'll give an example.

我举个例子。

I'm there and I'm thinking, "Okay, oil and gas industry, clothing industry, right?

我坐在那儿想：「好，油气行业、服装行业，对吧？

I didn't even think of that link.

这层关联我自己根本没想到。

I was just prompting an LLM to kind of figure that out.

我只是提示 LLM，让它把这层关系找出来。

Like what are the potential, you know, give me 200 potential links, right?

比如：给我 200 个潜在的关联，对吧？

And it's trained on your all your past research reports, everything, all your all your you know, the data that's that's been purchased by your hedge fund, alternative data, right?

而它是在你所有过往的研究报告、所有东西、你的对冲基金买过的所有数据、另类数据上训练出来的，对吧？

And so, it can map out 100, 1,000 different links and and synthesize all that different information.

所以它能梳理出 100 个、1,000 个不同的关联，把所有这些信息综合起来。

What's stopping an analyst from going, "Wait Wait, hold on a second.

那有什么能拦住一个分析师这么想：「等等，等一下。

I can just spend 90 minutes a day prompting this LLM, generate some reports, right?

我每天花 90 分钟提示这个 LLM，生成几份报告，不就行了？

System2 is our data science layer, so all the information's going to be accurate, right?"

System2 是我们的数据科学层，所以所有信息都会是准确的，对吧？」

Um and then he goes to his boss, he says, "Yo, look at this.

然后他去找老板说：「喂，你看这个。

I have all these different reports on what could happen."

我这儿有一大堆报告，讲可能会发生什么。」

I guess what's stop What's preventing something like that from happening in your eyes?

我想问的是，在你看来，什么能阻止这种事发生？

I I like I think that this is we are moving more in that direction.

我觉得我们确实正在往那个方向走。

And the the point that you to your question about laziness, I do think there are there are definitely differences in the work ethic of different fundamental analysts.

至于你问的懒惰这一点，我确实认为，不同基本面分析师在勤奋程度上差别很大。

It's kind of cool to be in our position as well because we work with a lot of different fundamental funds, and we can see differences in in you know, like the investment processes.

我们所处的位置也挺有意思，因为我们跟很多不同的基本面基金合作，能看到它们投资流程上的差异。

The But the the cool the exciting part is that those investors who are very

但真正让人兴奋的是，那些极其

curious are living like this is their biggest dream come true.

好奇的投资者，现在活得就像最大的梦想成真了。

They have way more questions than they can ever get answered.

他们的问题永远多过他们能得到解答的。

The the the really good ones.

真正厉害的那批人。

And you know, they come to us, they ask us, "Wait, before having access to us, they were limited in okay, I I can talk to a couple of experts.

他们来找我们、问我们——等一下，在接触到我们之前，他们能做的很有限：好吧，我可以找几个专家聊聊。

I can read up about this on the internet.

我可以上网查查这方面的资料。

And now you bring in like System2, it's like, 'Oh, I can ask like what is a magnitude more questions that now have answers because of all this granular data that we can get our hands on.'

而现在你把 System2 这样的东西引进来，就变成：「哦，我能问的问题多了一个数量级，而且现在都有答案，因为我们能拿到这么多颗粒度很细的数据。」

So, we work really well for those investors who are extremely curious.

所以我们对那些极度好奇的投资者特别管用。

And And even with us, their current business model that has again increased the number of questions they can get answered by orders of magnitude, they still are hungry for more.

而且即便有了我们——我们现在这套模式已经把他们能得到解答的问题数量又提升了好几个数量级——他们还是渴求更多。

And so, if you have an AI layer that can answer some of these questions faster, the good ones are just going to keep coming up with more and more and more questions.

所以，如果你有一个 AI 层能更快地回答其中一些问题，那些厉害的人只会不停地提出越来越多的问题。

And we're already seeing that.

我们已经看到这一点了。

Like we we've built AI tools on top of them.

比如我们在这之上搭了 AI 工具。

We would create a custom portal for each one of our clients with the analyses that we're developing for them.

我们会给每一个客户做一个定制门户，把我们为他们做的那些分析放进去。

And we're we're now putting in a um basically like a the chat box on top of it, so you can start interacting with certain components like the charts or the data sets that that we have there.

现在我们正在上面加一个，嗯，基本上就是个聊天框，这样你可以直接跟里面的某些组件互动，比如图表，或者我们放在那儿的数据集。

the the initial response we're just starting to like roll this out to some some early um like beta testers within our our client base is awesome.

最初的反馈——我们刚开始把它推给客户里的一小批早期 beta 测试者——非常好。

Like it's it's again orders of magnitude more questions that you can you can answer than you could before.

就是说，你能回答的问题又比以前多了好几个数量级。

And so, the what I what I think is going to end up happening is that those investors who are just insatiable in their curiosity will now have the tools to ask and get answers to so many more questions.

所以我觉得最后会变成这样：那些好奇心永远填不满的投资者，现在有了工具，能问出多得多的问题，而且拿得到答案。

So, you can know so much more about the company that you're invested in.

所以你能对自己投的那家公司了解得多得多。

And the only limiting factor is like what are you curious about?

而唯一的限制就是：你对什么好奇？

What kind of connections can you make between all the the different things that that you're seeing?

你看到的这么多东西之间，你能建立起什么样的连接？

And the more questions you ask and get answered, and we see this in our business model, like you answer one question and three more pop up.

而且你问出并得到解答的问题越多——我们在自己的业务里也看到这一点——你回答掉一个问题，就会冒出三个新的。

Because like, "Oh, I didn't realize that was happening, too.

因为会变成：「哦，我没意识到那件事也在发生。

Like what's what's going on there?"

那边到底怎么回事？」

And so, you can end up knowing so much more about these companies than you ever could before.

所以你最后能对这些公司了解到前所未有的程度。

And then we see what would happen in in in the past when like access to information like this is is improved by orders of magnitude.

然后我们再看看过去发生过什么——当这类信息的可获取性提升了好几个数量级的时候。

Like I used to have to go to a library like flip through an encyclopedia to to get an answer to something.

比如我以前得跑去图书馆翻百科全书，才能查到一个答案。

And then it's like, "Okay, now I can go on Wikipedia and I can get it instantly."

后来就变成：「好，现在我可以上维基百科，马上就拿到。」

Or I can go on Google and I can get instant answers to tons of my questions.

或者我可以上 Google，一大堆问题都能立刻得到答案。

And it's like essentially free.

而且基本上是免费的。

But it doesn't mean like everyone's using, you know, Google Wikipedia the same.

但这并不意味着每个人用 Google、维基百科的方式都一样。

And it's the same here the same same here.

这里也一样，完全一样。

It's like these fundamental investors who are like, "Okay, like I've gotten my initial question done.

就是说，那些基本面投资者里，有人是这样的：「好，我最初那个问题已经搞定了。

I'm I'm I'm fine moving on to to something else."

那我就转去看别的了。」

That I think will be at a disadvantage instead of the ones that be like, "Oh, I'm going to do a bunch of like expert calls.

我认为这种人会处于劣势；另一种人则会说：「哦，我要去做一堆专家访谈。

I'm going to like look at all these different data sets.

我要把这些不同的数据集全看一遍。

I'm going to play around with with a bunch of this the this data and information until I know more about the companies than those companies know about themselves."

我要把这些数据和信息反复摆弄，直到我比这些公司自己还了解它们。」

Which I love it when that happens.

每次发生这种事，我都特别喜欢。

Like when our our clients tell us that they just got off the the phone with, you know, investor relations at the company to to tell them something that they're seeing in the data, and the company's like, "How are you guys How do you know this?"

比如我们的客户告诉我们，他们刚跟一家公司的投资者关系部门通完电话，去告诉对方他们在数据里看到的某个现象，公司那边的反应是：「你们是怎么——你们怎么会知道这个？」

And they themselves don't have access to to that kind of information.

而他们自己都拿不到那种信息。

It's it's a crazy world, right?

这世界挺疯狂的，对吧？

Cuz knows about themselves versus what they're reporting to investors once a quarter.

因为——公司自己知道的，跟它每季度一次报给投资者的，完全是两回事。

But if you can get to a point where the investors or anybody who's curious enough and has enough like money to be able to spend on on these tools can know more about the company than people inside the company, then you end up the idea is like you end up having like much better forecasts of what's going to happen, which companies are going to be successful or not, and you end up making much better investments as a result.

但如果能做到这一步：投资者，或者任何足够好奇、又有钱买得起这些工具的人，比公司内部的人更了解这家公司，那你最终——这个设想就是，你最终会对将要发生什么、哪些公司会成功哪些不会，有好得多的预测，结果就是你能做出好得多的投资。

First off, when are you rolling out that tool to the general public?

先问一句，你们那个工具什么时候向公众开放？

Cuz I need to get access to it.

因为我得用上它。

Yeah, I know.

是啊，我知道。

I mean, I think our business model unfortunately is not for the general public.

很遗憾，我们的业务模式并不面向公众。

We work with a relatively limited number of of hedge funds.

我们合作的对冲基金数量相对有限。

And uh and even that like I don't know if we're necessarily going to roll it out to to everybody yet.

而且即便是在这个范围内，我也不确定现在是不是一定会向所有人开放。

We're starting with some of our um Yeah, I mean, there there there is there's there's a ton of demand for for something like this.

我们会先从一部分——嗯，是啊，这种东西的需求非常大。

And uh

而且……

Oh yeah, as you said that, my like light bulb just shot off and I was like, how do I try this to a lot?

哦对，你一说这个，我脑子里的灯泡「叮」地一下就亮了，心想：我怎么才能试上这个？

Yeah.

是啊。

Yeah.

对。

No, I mean, it's yeah, like it's it's yeah, it's it's awesome.

不不，我是说，这确实挺棒的。

I mean, I think it's yeah, if you're a curious person and you want to know a lot about these different companies, like having access to these kinds of tools is going to make it yeah, that much better.

我觉得，如果你是个好奇的人，又想深入了解这些不同的公司，那能用上这类工具，确实会让一切好得多。

Now, the follow-up that I also had was you're sitting in a very unique vantage point where you're able to see and interact with these huge fundamental hedge funds.

我接下来还有个追问：你所处的位置非常独特，你能看到、也能直接接触这些巨型的基本面对冲基金。

And you're seeing the process you're seeing all of their processes, right?

你看得到它们的流程——所有这些基金的流程，对吧？

And so my question is with these new tools, right?

所以我的问题是，有了这些新工具之后，对吧？

Where do you think the competitive advantage of funds or or rather where are you seeing enhancements in some funds' competitive advantages because of

你觉得基金的竞争优势会在哪里——或者说，你在哪些基金身上看到了竞争优势的增强，因为

Yeah, that's a great question.

嗯，这是个好问题。

So, let me take a step back.

那我先退一步讲。

What's been happening in the fundamental investment world is that you've seen a roll-up of single managers into the multi-manager platforms like Citadel, Millennium, Baupost, and so forth.

基本面投资这块正在发生的事是：你看到单一管理人被整合（roll-up）进 Citadel、Millennium、Baupost 这类多策略平台。

The There are many reasons for this.

这背后原因很多。

Primarily, the returns have been phenomenal in in those kinds of places relative to relative to the returns of single managers.

最主要的一条是，这类平台的回报，相对单一管理人的回报，一直好得惊人。

And there's also if you go a little bit further back in time in fact like, you know, before 2008, there was a preponderance like of a fund of funds where the idea was you're going to diversify across a number of managers because diversification makes sense on, you know, in in theory.

而且，把时间再往回推一点，其实在 2008 年之前，占主导的是 fund-of-funds（外部母基金）：思路是在若干个管理人之间做分散，因为分散在理论上是说得通的。

What happened post uh like 2008 is that it turned out that those funds that were supposed to have done a good job doing due diligence on their their managers didn't.

2008 年之后的情况是，那些本该把管理人尽调做好的基金，其实并没做好。

There was the Madoff scandal.

出了 Madoff 丑闻。

Long story short, fund of funds fell out of favor.

长话短说，fund-of-funds 从此失宠。

But you then saw the rise of these uh these multi-managers.

但接着你就看到这些多策略平台崛起了。

They're essentially fund of funds, right?

它们本质上就是 fund-of-funds，对吧？

They're bringing in these individual pods to to work within them.

它们把一个个独立的 pod（多策略平台下的投资小组）引进来，在自己内部运作。

And and they're achieving like they're doing a much much better job of on the risk management portion of it than the old-school fund of funds did because they have very sophisticated models and they have very sophisticated centralized teams that can offer all sorts of services to those to those managers.

而且在风险管理这块，它们比老派的 fund-of-funds 做得好太多，因为它们有非常精密的模型，也有非常成熟的集中化团队，能给这些管理人提供各种服务。

Not just saying, "Okay, I invested in a bunch of them and and that's that."

而不是只说一句「好，我投了一批人，就这样了」。

Now, there are still a lot of folks who want to manage their own funds.

不过，仍然有很多人想自己管自己的基金。

I mean, there are definitely a ton of of drawbacks to working for one of these like multi-managers, primarily being that if you one of the primary risk management tools is that if you lose money, you get fired.

在这些多策略平台里干，弊端确实一大堆，最主要的一条是：它最核心的风险管理手段之一就是——你一亏钱，就被开掉。

And so it works, but it also creates a very stressful environment for many many of the people who work as portfolio managers in these in these pods.

这招是管用，但也让很多在这些 pod 里做 PM 的人处在极大的压力之下。

Like I have many friends who have gone through that that process.

我有很多朋友都经历过这个过程。

And so, they do it because it kind of works and it's easy to get capital and all you have access to all these centralized resources, but if they're being honest, they would much rather operate independently.

他们之所以还做，是因为这套东西确实管用，拿钱容易，还能用上所有这些集中化的资源；但说实话，他们其实更愿意独立干。

And many of you also see some of the like these trends within the the large guys as well where they're now essentially still like seeding um seeding different pods and allowing them to operate independently with their own brand and but still getting access to some of these resources.

而且在这些大机构内部，你也能看到类似的趋势：它们现在实际上还是在出资孵化（seeding）不同的 pod，让这些 pod 用自己的品牌独立运作，同时仍然用得上其中一部分资源。

There's There's always going to be this tension between the individual managers like want to be left alone.

这中间永远存在一种张力：独立的管理人希望别人别来插手。

Nobody wants to be held to like you can't lose 5% off of your high watermark and get fired.

没人愿意被这么管着：从高水位线回撤 5% 就得走人。

I mean, that's a very stressful position to to be in.

那处境压力非常大。

Um and not only that, it also changes the the type of investment structure.

而且不止如此，它还改变了投资结构的形态。

You're going to have to be like market neutral.

你必须做市场中性。

You can't take uh directional concentrated bets in stocks, which is what many of our clients do.

你不能在股票上下方向性的、集中的赌注，而那恰恰是我们很多客户在做的事。

And there are reasons like to to to believe that yeah, like if you can't take concentrated directional bets, you will become very rich.

而且有理由相信，如果你不能下集中的方向性赌注，你就发不了大财。

And that's honestly one of the one of the only ways to to become very very rich is to be pretty concentrated and take long-term bets.

说实话，能变得非常非常有钱的少数几条路之一，就是持仓相当集中，并且下长期的赌注。

It's also a very easy way to lose all your money, but it's it is one of the one of the few ways of making it making it rich.

这同时也是很容易把钱全亏光的路子，但它确实是能致富的少数几条路之一。

And so, if you're okay with that kind of that kind of risk reward profile, these single managers can um can can be quite quite appealing.

所以，如果你能接受这种风险收益特征，这些单一管理人可以相当有吸引力。

Now, going back to your question is like what what can this all these additional tools do in in this in this world?

回到你的问题：所有这些新增的工具，在这个世界里能干什么？

Well, I think if you can help bring some of the advantages of the multi-manager platforms to individual single managers, you they might be able to get the best of both worlds.

我觉得，如果你能把多策略平台的一部分优势带给独立的单一管理人，他们也许就能两全其美。

And that is in in some ways what System2 does.

某种意义上，这就是 System2 在做的事。

I mean, they have these some of these larger funds have in-house data teams.

我是说，其中一些规模更大的基金，是有内部数据团队的。

Um but we have a bigger team than most of them.

但我们的团队比它们中的大多数都大。

And we provide that to a you know, to broad a range of independent managers.

而且我们把这份能力提供给很广的一批独立管理人。

And and we're incentivized And to be honest, I mean, I mean, I've been talking about this a lot, but like I being an independent company that has, you know, needs to to make payroll and and pay the bills and and grow, we have a very different incentive structure than if you're an in-house department of an organization in which you're a back office.

另外我们的激励是对的——说实话，这个我已经讲过很多遍了，但作为一家独立公司，要发工资、要付账单、要增长，我们的激励结构，跟你在一个组织里当内部部门、当后台，是完全不一样的。

You're not even the the thing that generates revenue.

你甚至都不是创造收入的那一环。

And so, the amount of like resources, capital, and innovation that you can achieve externally, even though you might have a lot less money than they do, in the long run, I think we're we're just going to continue to to crush them.

所以在外部，你能调动的资源、资本和创新，即便钱比它们少得多，长期看，我认为我们会持续碾压它们。

Um and and so, there are other companies like us out there obviously providing services to to single to single managers.

当然，外面还有其他像我们这样的公司，在给单一管理人提供服务。

And so, there is a world where you can envision a an ecosystem of independent companies that are providing services these single managers that are not just on par with what these like Citadel or Millennium can do, but actually better.

所以你可以设想这样一个世界：有一个由独立公司组成的生态，给这些单一管理人提供服务，而且不只是跟 Citadel 或 Millennium 能做的持平，是真的更好。

And leveraging AI, like being much faster at deploying.

再借助 AI，部署速度快得多。

And I've I've talked to many people that are coming in at or still at these these large funds.

我跟很多刚进这些大基金、或者还待在里面的人聊过。

It takes forever.

什么事都慢得没完没了。

It's like, you know, I was talking to a friend and works at a very large asset manager, and they just enabled people to get access to ChatGPT in at work.

比如我跟一个在超大型资产管理公司的朋友聊，他们才刚刚放开，让员工在工作里能用 ChatGPT。

And it's been like 3 years, you know, since these things have have rolled out.

而这些东西推出到现在都快 3 年了。

And due to compliance reasons like you have to make sure the data privacy is set up that they have access.

出于合规原因，你得先把数据隐私那一套配好，他们才能用。

Like we And there are like there are legit reasons.

这确实有正当理由。

But at the same time, they're they're operating with a handicap when technology is evolving so quickly that if you can do anything very quickly, you can you can end up you can end up winning.

但与此同时，他们等于是带着劣势在跑；而技术演进这么快的时候，你要是什么事都能做得非常快，最后就可能赢。

And so, that is the the exciting thing about what we're seeing with our client base is that they are extremely receptive to innovation.

所以，我们在自己客户群里看到的、让人兴奋的地方就是：他们对创新的接受度极高。

And as companies like ours are getting bigger and better, we are able to provide a better service than if any of these any of these folks were doing it in-house.

而随着我们这样的公司越做越大、越做越好，我们提供的服务会比这些人自己在内部做更好。

And right now, I mean, touched on incentives, you're actually constructing an even better incentive structure in the form of a fund.

那现在——你刚提到激励——你其实正在以一只基金的形式，搭一个更好的激励结构。

Can you tell me a little bit about that?

能跟我讲讲这个吗？

Tell us about

给我们讲讲

that.

那件事。

Yeah, so the I mean, it's it's interesting to to be in this in this space where our biggest competitor for for talent is my my dear friends at Citadel who have have poached people from us over the the years.

是的，身处这样一个领域挺有意思：我们争夺人才最大的对手，是我在 Citadel 的好朋友们，这些年他们一直在从我们这儿挖人。

And we're at the end they're like, you know, a startup, a tech startup.

而我们说到底就是家创业公司，一家科技创业公司。

It's very hard for a company of our size to to compete with a fund like that when it comes to the the paying and recruiting people.

我们这种体量的公司，要在开价和招人上跟那样一只基金竞争，非常难。

And yet we've been doing it and we've been doing it successfully.

但我们一直在这么干，而且干得还挺成功。

And then by the way, I do think that in some ways it is a compliment that they want to to steal our folks.

顺带一提，他们想挖走我们的人，我确实觉得某种程度上是种恭维。

Um and so, as part of that, we we've been thinking about how I tell you we're competing, right?

所以在这个背景下，我们一直在琢磨——我跟你说，我们是在竞争，对吧？

And we're competing and it's it is kind of cool.

我们在竞争，这其实还挺酷的。

It is an honor to be competing with some like the best companies in the world.

能跟世界上最好的一批公司竞争，是种荣幸。

And like we're nobody.

而我们不过是无名小卒。

Like they don't know like Ken Griffin doesn't know I exist.

他们不认识我们，Ken Griffin 根本不知道我这号人存在。

And yet like he's he's my competitor, right?

但他就是我的竞争对手，对吧？

And it is kind of cool to be in that in that environment to say, "Okay, like what else can I do that he's able to do?"

身处这种环境挺酷的，你会问自己：「好，他能做到的事，我还能做点什么？」

And then in this particular context that you're referring to, which is um yeah, like the employee retention.

回到你问的这件具体的事，也就是员工留存。

And um yeah, so one of the things I used to to work at a a large hedge fund myself, and one of the things that is offered to employees once they reach a certain level is the ability to invest in the fund.

对，说到这个——我自己以前也在一家大型对冲基金干过——员工到了某个级别，会拿到一项待遇：可以投资这只基金。

And these funds typically have like my my previous fund had like a $50 million investment minimum.

这些基金通常都有门槛——像我上一家基金，投资门槛是 $50 million。

Like many funds today, even like smaller ones, have like $10 million investment minimums.

今天很多基金，哪怕比较小的，投资门槛也有 $10 million。

Some are completely closed to outside investment.

有些则完全不对外部资金开放。

And so, if you're just some employee who you're not playing like 50 grand into into the fund or 100 grand to the fund, like no one's going to pay attention to you.

所以如果你只是个普通员工，往基金里投的不过 5 万或者 10 万美元，没人会搭理你。

Except if you actually work there.

除非你真的在那儿上班。

And they do offer that as a as a perk.

他们确实把这个当一项福利提供。

And so, we were thinking about that in the context of yeah, like these funds offering folks the ability to invest in a Citadel.

所以我们就顺着这个想：这些基金能让员工投进 Citadel 这样的机构。

And say, why can't we try to do something similar?

那我们为什么不能试着做点类似的事？

Why can't I compete on that part of it as well?

我为什么不能在这一块也参与竞争？

And so, what we've been been doing as an employee like retention and incentive alignment tool is going to our hedge fund clients and asking them if they would kindly let us come in as LPs with much lower minimums, obviously.

所以，作为员工留存和激励对齐的工具，我们一直在做的是：去找我们的对冲基金客户，问他们能不能行个方便，让我们以 LP 的身份进来，门槛当然要低得多。

And and so far, probably all of them have said yes, which is awesome.

到目前为止，大概所有人都答应了，这非常棒。

One of I remember one of the guys in particular said it would be an honor to have us, which was like one of the sweetest things to to hear, especially when you know already you're just going to play in a couple hundred grand into into each of these these things.

我记得有一位特别说，能有我们加入是他的荣幸，这是我听过最动听的话之一，尤其是当你心里清楚，自己往每一只里也就投个二十来万美元。

But, the cool part about it is that we have access to these investors that again, you will not be able to access otherwise.

但酷的地方在于，我们能接触到这些投资人，换别的路子你是接触不到的。

And some of our funds are awesome, like, you know, like invested in cool companies like OpenAI, Waymo, and tons of interesting things on private credit, like CLOs, like all sorts of instruments that you will not be able to access as your average average, you know, investor.

而且我们有些基金非常厉害，投了 OpenAI、Waymo 这样很酷的公司，还有一大堆有意思的私募信贷，比如 CLO，各种你作为普通投资者根本碰不到的工具。

So, we as we're we're starting a fund of funds that is going to be initially for employees to be able to put some of their some of their bonus, some of their their savings back into our clients.

所以我们正在启动一只 fund-of-funds（内部母基金），初期是让员工能把一部分奖金、一部分储蓄投回到我们的客户里。

And it already has started to to pay dividends on the incentive alignment piece, where I was talking to one of our client leads, and and she was like, she was like, wait, so my bonus would be invested in this in in my my client.

而在激励对齐这块，它已经开始见效了：我跟我们一位客户负责人聊，她说，等一下，所以我的奖金会被投进我负责的这个客户里？

I was like, yeah, that's right.

我说，对，没错。

She was like, well, what if I don't like some of the things they're they're working on?

她说，那如果我不喜欢他们正在做的某些事呢？

I was like, you should say something today if you feel like something's like you have a differentiated view or like you don't agree with with something, like you should speak up.

我说，如果你觉得自己有差异化的观点，或者不认同某件事，那今天就该讲出来，你应该开口。

And yet, when it's your money, like the this thing flips, you know?

但当这是你自己的钱时，整件事就翻转了。

Like yes, we obviously do, and like we care and want our clients to make money, but you care so much more when it's your bonus.

当然我们本来就在乎，也希望客户赚钱，但那是你的奖金时，你会在乎得多得多。

Even if like you're putting in whatever, like 50 grand, it's 50 grand of your money.

哪怕你投的就 5 万美元，那也是你自己的 5 万美元。

And you could ask me what that fund is, nothing.

你可以问我这笔钱对那只基金算什么，什么都不算。

I was like, a drop in the bucket of what the fund manages.

我说，相对这只基金管理的规模，就是沧海一粟。

And yet, that 50 grand is going to make us that much better at doing our job because our incentives are going to be aligned with it.

但正是这 5 万美元，会让我们的活儿干得好得多，因为我们的激励跟它对齐了。

And now you have people are excited about this.

现在大家对这件事都很兴奋。

They're like they're asking, like, can I invest in my client?

他们会问：我能投我自己负责的那个客户吗？

Because we're going like making the rounds trying to talk to our our different clients to ask them to allow us to do this.

因为我们正挨个去找不同的客户，请他们允许我们这么做。

And people want Yeah, people want in.

大家都想——对，大家都想进来。

So, it's and we do have like a really cool We're in a really cool position given the level of like access we have to their investment process to be able to identify interesting funds to put our our money into.

所以，鉴于我们能介入他们投资流程的深度，我们的位置其实非常好，能挑出值得把钱投进去的有意思的基金。

So, it's Yeah, that's been one of the cool things about our job is the the set of opportunities that are presenting themselves, and every once in a while we try to execute on one of them.

所以，这也是我们这份工作挺酷的一点：机会不断冒出来，我们时不时就抓一个去落地。

I mean, on a fund's end, it's a no-brainer to let you guys invest cuz it's going to make their data science layer that much better, right?

我是说，站在基金那边，让你们投进来是想都不用想的事，因为这会让他们的数据科学层好上不少，对吧？

Exactly.

正是。

And I think yeah, I think that that one manager's response kind of sums it up when he said it would be an honor to have us, right?

我觉得，那位管理人的回应挺能概括这件事，他说能有我们加入是他的荣幸，对吧？

That is the kind of attitude you want from a partnership.

这就是你希望在一段合作关系里看到的态度。

And and it goes both ways.

而且这是双向的。

Like yeah, he's he's helping us out.

对，他在帮我们。

We're obviously helping him.

我们显然也在帮他。

Like we're we are giving them money.

我们是在给他们钱。

It doesn't matter like we feel like it's it's nothing, and they're billionaires.

虽然我们觉得这点钱什么都不算，而且他们都是亿万富翁，但那不重要。

Like why would they care about 50 grand?

他们凭什么会在意 5 万美元？

But they are like that is a mutually partnership that is in the long run, hopefully going to be great for for all of us.

但这确实是一种互利的合作关系，长期看，希望对我们所有人都好。

We have seen our clients over the years that we've worked with some of our our clients for, I don't know, like almost 10 years now.

我们看着客户这些年一路走过来，有些客户我们已经合作了，我不知道，差不多快 10 年了。

And and yeah, like it would've been awesome to put in some money back then to given the amount of growth that that they've seen.

说真的，看他们这些年的增长，当年要是投点钱进去就太好了。

So, it's yeah, it's it's an exciting exciting position to be in.

所以，能站在这个位置上，真的挺让人兴奋。

On incentives, if you have two different fundamental hedge funds, right?

说到激励，假设有两家不同的基本面对冲基金，对吧？

One Let's say they both trade consumer or health care.

比如说，它们都做消费或者医疗。

Let's say they're both in the same sector, and they're technically competing against one another.

假设它们在同一个行业里，严格说是在互相竞争。

Um

嗯

what position does that put you in?

那这把你们放在什么位置上？

Yeah, well, I mean yes, there's definitely and this comes up in every conversation we have with um prospective clients.

是的，确实有这个问题，而且我们跟每一个潜在客户聊的时候都会碰到。

Their investing is a pretty terrible market to operate in as an independent company because there are negative network effects.

对一家独立公司来说，投资是个相当糟糕的市场，因为这里存在负网络效应。

The more clients that we the company has, the less valuable that product is to each one of those clients.

公司的客户越多，产品对其中每一个客户的价值就越低。

Because it's all about differentiated insights into into this.

因为这件事的关键全在于差异化的洞察。

If everyone knows the same piece of information going back to our very first um point around outliers.

如果所有人都掌握同一条信息——这又回到我们最开始讲的异类那一点。

Like if everyone knows the same thing, then it's no longer interesting.

如果所有人都知道同一件事，它就不再有意思了。

You can't make money off of that.

你没法靠它赚钱。

And so, the And this is but there is also, ironically, how you make the most money as an independent company.

而讽刺的是，这恰恰又是一家独立公司最赚钱的方式。

That's how you scale is you build a product once, and then you sell it to 100, 1,000 different companies, and now you you made a lot of money.

规模化就是这么来的：产品做一次，然后卖给 100 家、1,000 家不同的公司，这下你就赚了一大笔。

We from the very beginning created our business model to be different.

我们从一开始就把商业模式设计成不一样的。

We are a mix between like a consulting company and a and a tech company.

我们是咨询公司和科技公司的混合体。

We, if you think about a a power law, like the the 80/20 rule, most like product companies that want to scale will focus on the 80% most common use cases.

你想想幂律，比如 80/20 法则：大多数想做规模化的产品公司，会聚焦那 80% 最常见的使用场景。

We chose to focus on the the 20% of the long tail of random one-offs.

我们选择聚焦的是那 20%——长尾里各种零散的一次性项目。

And we've built, including our technology layer, has been built around being able to do one-offs at scale.

我们搭的一切，包括技术层，都是围绕『能规模化地做一次性项目』建起来的。

And we love one-offs.

而且我们喜欢一次性项目。

Like one-offs for any product company are customer service nightmare.

一次性项目对任何产品公司来说都是客服的噩梦。

For us, we that's exactly what we're set up to do.

而对我们来说，这恰恰是我们生来就要干的事。

And that business model makes it easier to have those kinds of conversations with our clients, where we are not pushing out a report.

这种商业模式也让我们更容易跟客户谈这类事——我们不是在对外群发一份报告。

If we discover something cool on some company, we're not going to like blast it out all over the internet saying, "Hey, look at us. We we found this great insight."

如果我们在某家公司身上发现了很酷的东西，我们不会满互联网地嚷嚷：『嘿，看我们，我们找到了这个了不起的洞察。』

It stays with that with that with that client.

它只留在那个客户手里。

Having said that, we there are obviously benefits from us having worked on similar projects in the past.

话虽如此，我们过去做过类似的项目，这显然是有好处的。

As we work with with more clients, we've done more things, and every time we see a new problem, there are other similar problems that we've worked on in the past.

服务的客户越多，我们做过的事就越多，每碰到一个新问题，都能找到过去做过的类似问题。

And that is a big reason why clients like working with us cuz we have a ton of experience.

这也是客户喜欢跟我们合作的一大原因，我们的经验非常多。

And and lastly, to touch on a previous point, it still goes back to the creativity and the curiosity of the investors that we're working with.

最后，呼应前面提过的一点，这归根到底还是取决于我们服务的那些投资人的创造力和好奇心。

Like we have investors, like different funds that are invested in the same name.

比如我们的客户里有不同的基金，投的是同一只标的。

And oftentimes, by the way, they talk to each other.

而且顺便说一句，他们之间常常互相交流。

We and they'll they'll bring us in on the conversation, and and we can we so we've had situations where we'll like share all the the work between the funds.

他们会把我们拉进对话里，所以我们遇到过这种情况：把所有工作成果在这几家基金之间共享。

Like here's like everything that that you guys have been doing.

就是『这是你们各自一直在做的全部东西』。

At their request.

这是应他们的要求。

Like everyone was on the call, and we were we're talking about this.

所有人都在那通电话会上，我们就聊这个。

Even then, when you have this this like convergence, like both two clients are looking at the same thing, it inevitably ends up diverging again from there because each client is going to have a different follow-up question.

即便如此，当出现这种收敛——两个客户在看同一个东西——从那里开始也必然会重新发散，因为每个客户的追问都不一样。

You're going to have a different third parties.

你会碰到不同的第三方。

They might not care about that name as much.

他们可能没那么在意那只标的。

They might care about a different name.

他们在意的可能是另一只。

But the the rabbit hole that they go down afterwards is is very different.

但他们之后钻进去的兔子洞非常不一样。

And it's basically choose your own adventure.

这基本上就是『选择你自己的冒险』。

Depending on the questions that you ask, the follow-up questions to those questions, you end up in a very different rabbit hole than somebody else who asked a a different set of a follow-up questions.

取决于你问了什么、又追问了什么，你最终掉进的兔子洞，会跟另一个问了另一组追问的人完全不同。

And this ability of helping our clients like go down rabbit holes that ultimately get them to a deeper understanding about the company is a cool feature of the fundamental investment process, where you can have yeah, like two investors invest in the same company, but they're looking at it in in different ways.

而这种帮客户钻兔子洞、最终让他们对一家公司理解更深的能力，是基本面投资流程里很酷的一点——两个投资人可以投同一家公司，看它的方式却完全不同。

And that's ultimately where the alpha comes in.

而这最终就是 alpha 的来源。

And you see this in their in the results.

在业绩上你就能看到这一点。

Like people are invested in the same like same industries, similar types of funds, similar strategies, and up having very different returns because they follow different rabbit holes.

比如有些人投的是同样的行业、同类的基金、相似的策略，最后回报却差很多，因为他们钻的是不同的兔子洞。

They have a different set of curiosities that leads them in different places.

他们的好奇心指向不同，把他们带到了不同的地方。

So, and can yeah, it can be done, but it is it was one of the the hardest things that we we had to overcome in the beginning is getting our clients to trust us that we have their interests at heart, and that we are going to provide them with like great work that is differentiated and that is custom to

所以，是的，这件事能做到，但这是我们一开始必须克服的最难的事情之一：让客户信任我们把他们的利益放在心上，相信我们交付的东西够好、有差异化，而且是为他们量身定制、贴合他们

their interests.

的利益。

Would you ever productize certain aspects of what you know?

你会考虑把你们掌握的某些东西产品化吗？

So, we have that conversation often.

这个话题我们经常聊。

I mean, there are definitely a lot of reasons to productize.

产品化的理由确实有很多。

It would be great to have a lower price.

价格能更低当然好。

Our our price point is is pretty expensive.

我们的定价相当贵。

And I mean, like they were basically starting out at a million dollars a year in our in our contracts.

我们的合同基本上是从一年 $1 million 起步的。

And that is expensive even for very large hedge funds, which is is kind of counterintuitive.

这个价格即使对非常大的对冲基金来说也算贵，这有点反直觉。

They manage whatever, 40 billion dollars, they still think that that's that's a lot of money.

他们管着 $40 billion 之类的规模，还是觉得那是一大笔钱。

And we can, by the way, get into It was a really interesting um as as a quick aside there for anyone who's listening that is interested in selling it to hedge funds, and you hear all this and you're like, "Oh, it's great. They have all this money, and they're spending like a million dollars a year on on this on research."

顺便说一句，这里可以展开讲讲——插一句题外话，给听众里那些想把东西卖给对冲基金的人：你听到这些会想，『哦，太好了，他们钱那么多，一年花 $1 million 做研究。』

I think that yes, it does happen, but it is so freaking hard to get them to to to open up to do this.

我想说，是的，这种事确实会发生，但要让他们松口掏这笔钱，难得要命。

The more of my favorite stories was when I was working at this I was at a a for about about a decade and you know the fund managed like $20 billion or the founder was a billionaire and there was there's this one time that one of my friends was in the elevator with the founder.

我最喜欢的故事之一发生在我以前待的那家基金——我在那儿干了差不多十年，基金管着 $20 billion 左右，创始人是个亿万富翁——有一次，我一个朋友跟创始人一起搭电梯。

He asked the founder say, "Oh, what what are you up to?"

他问创始人：『你在忙什么呢？』

The founder says, "My my wife's birthday. I'm going to run out and buy her birthday card."

创始人说：『我太太生日。我要跑出去给她买张生日贺卡。』

My friend is like, "Oh, are you going to go to Papyrus?"

我朋友说：『哦，你去 Papyrus 吗？』

Which was like that's like this fancy birthday card store that was literally in our building on the first floor of our office building.

那是家很高档的生日贺卡店，就在我们楼里，办公楼一层。

And the the founder says, "Fuck no, that place is too expensive. I'm going to Duane Reade."

结果创始人说：『操，才不去，那地方太贵了。我去 Duane Reade。』

So, he was walking a few blocks away to go get a Duane Reade birthday card for his wife.

于是他要走好几个街区，去 Duane Reade 给太太买张生日贺卡。

Which to be fair Papyrus does have insane prices for birthday cards.

公平地说，Papyrus 的生日贺卡价格确实离谱。

But he's a billionaire.

但他可是个亿万富翁。

And and that I think summarizes the attitude of what makes a successful investor.

我觉得这件事概括了成功投资人的那种心态。

Like they're looking for value whether they're buying a birthday card or a company.

不管买的是一张生日贺卡还是一家公司，他们都在找价值。

And or a with service provider like System2.

或者买 System2 这样的服务商。

And so they they have this lens like even though they have a lot of money they probably have that money because of the way that they they look at everything and they are not very very easy to to open their their wallet.

所以他们看什么都带着这层视角：钱很多，但他们之所以有这些钱，多半正是因为他们这么看每一件事，钱包非常非常不容易掏开。

And so go back to productization question like we have yeah like it could be interesting to to to productize something and you'd have more customers.

所以回到产品化那个问题：是的，把某些东西产品化可能挺有意思，客户也会更多。

But it dilutes the alpha.

但那会稀释 alpha。

It dilutes this ability to for us to answer custom questions and to have our clients feel like we have their best interests in in in mind at all times.

它会稀释我们回答定制问题的能力，也会稀释客户那种『我们始终把他们的利益放在第一位』的感觉。

Um and and yet there are businesses like business models where this can be done.

不过，确实有些生意、有些商业模式可以这么做。

Like we're not fishing we're whale hunting.

比如我们不是在钓小鱼，我们是在猎鲸。

We're constantly like one of the metrics that I look for is not just the the growth in our top line revenue but also what is the size like the dollar amount of our largest clients.

我一直盯的指标之一，不只是营收的增长，还有最大客户的体量——具体是多少钱。

And I want that number to keep going up.

我希望这个数字一直往上走。

I want us to always have like be going after a larger and larger engagement.

我希望我们始终在争取越来越大的单子。

This is you know there's constant it's it's in in some cases like basically what that means is that I have some kind of like a large customer concentration risk where my top customer is I don't know like 30% of my my revenues.

这在某些情况下基本上意味着，我背着很大的客户集中度风险——最大的那个客户大概占了我营收的 30%。

and yet I think there's something awesome to be able to say like hey, do we have people that are paying us tons of money for what we're doing.

但我还是觉得，能说出『有人为我们做的事付给我们巨额的钱』，这件事很棒。

And they really really value us.

而且他们真的非常非常看重我们。

We're the best of the best when it comes to this line of work.

在这条业务线上，我们是最好中的最好。

And that that honestly has helped a lot in our go-to-market strategy where it's now much more of a funnel thing.

老实说，这一点对我们的 go-to-market 帮助很大，现在它更像一个漏斗。

Like we don't work with everybody.

我们并不跟所有人合作。

We work with the people that are very curious.

我们只跟好奇心非常强的人合作。

They have tons of questions.

他们有一大堆问题。

They can't get enough data and answers to those questions and they want a partner that can that can provide them with this very unique experience of what it's like to work with the best of the best in the in the data world.

他们对数据、对这些问题的答案永远不嫌多，而且想要一个伙伴，能给他们这种非常独特的体验——跟数据世界里最顶尖的一批人合作是什么感觉。

that's that's what we're going for.

这就是我们要做的事。

That's that's what's worked for us.

这也是对我们奏效的路子。

Maybe one day we'll we'll do something more scalable but for now we've been very happy with this approach.

也许有一天我们会做点更能规模化的东西，但到目前为止，我们对这个路子很满意。

It sounds like you think a lot about your clients about the potential business avenues.

听起来你对客户、对可能的业务方向想了很多。

It also sounds like you have a lot more options than you have resources to go after.

听起来你手上的选项，也远多于你能投进去的资源。

As in I remember speaking to you when I visited your office I think it was a couple of two weeks ago or a week and a half ago.

比如我记得去你办公室拜访时跟你聊过，大概是两周前，或者一周半以前。

Um and I was blown away by you telling me, "Hey, one client wants this.

你跟我说的话让我震住了：『嘿，有个客户想要这个。

We've already validated the market for that product but we just don't have the resources.

我们已经验证过那个产品的市场了，但就是没有资源。

We could theoretically build out, you know, and hire a couple guys, you know, have each guy own a particular project that we've already validated, roll that out and that in itself is an amazing business, right?"

理论上我们可以把它做出来，招几个人，一人负责一个已经验证过的项目，推出去，这本身就是一门了不起的生意，对吧？』

I guess my question is there's so much to think about there's so much that you're thinking about.

我的问题是，要想的事太多了，你脑子里装着的事也太多了。

What crosses your mind every day?

每天浮现在你脑子里的是什么？

Is there one thing that's repeated repeatedly in your thoughts?

有没有哪件事在你心里反复出现？

Yeah, most of the the coolest thing for me and about System2 in the position that we're in right now is that we have access to the minds of some of the world's like smartest billionaires.

对我来说，System2 和我们现在所处的位置，最酷的一点是我们能接触到全世界最聪明的一批亿万富翁的头脑。

It's crazy to think about that.

想想都觉得疯狂。

Like we are we're nobody in the grand scheme of companies.

在所有公司的大格局里，我们就是个无名之辈。

Like nobody knows heard of us and yet like we I personally know I don't know two dozen billionaires and there's a couple hundred in the in the US.

没人听说过我们，可我个人就认识大概二十几个亿万富翁，而全美国也就几百个。

And not only do we know them like I we're in these conversations where they're deciding what to spend their money on and we're playing our small role in helping them with that with that decision.

而且我们不只是认识他们——我们就在这些对话里，他们在决定把钱花到哪儿，而我们在这个决定里扮演着自己那一小份角色。

And the things they invest in are the things that end up getting built in the world.

而他们投什么，世界上最终就造出什么。

Right?

对吧？

Like we we see we read the headlines and we're like, "Oh, yeah, we worked with that client that led that round."

比如我们看到新闻标题，会说：『哦，对，领投这一轮的那家，是我们服务过的客户。』

It's insane to be to be part of that conversation to see how they're thinking about the world, what excites them, where they see their opportunities.

能置身这个对话，看到他们怎么思考这个世界、什么让他们兴奋、他们在哪儿看到机会，这太不可思议了。

And that is we're punching way above our weight class.

而我们这是在越级出拳，远超自己的量级。

We've been doing this from from the very beginning.

我们从一开始就在这么干。

It kind of goes back to that that business model like whale hunting and being

这某种程度上又回到那个商业模式——猎鲸，以及

you made the point about like I always think about what the client wants.

你刚才说到一点——我总是在想客户想要什么。

You we we we had these conversations like at what point like somebody asked me recently like at what point do you basically like say no to a client?

我们聊过这个话题，最近还有人问我：到什么时候你才会对客户说不？

And this is one of the big features of like if you try to build a product you have to say no to certain feature requests you have to figure out what what to prioritize.

这正是做产品的一大特点：你要做产品，就必须对某些功能需求说不，得想清楚先做什么。

For us that's like that's never been the case.

对我们来说，从来没有过这种情况。

Like I will do anything.

我什么都愿意做。

If the client wants us to spend all our time and attention on solving that one problem we will [ __ ] get to the bottom of that problem.

如果客户想让我们把全部时间和精力都砸在那一个问题上，我们就 [ __ ] 把那个问题刨到底。

We will figure it out.

我们会把它弄明白。

We will like scorch the earth until we find a solution that gets them an answer to to that question.

我们会掘地三尺，直到找出一个能回答那个问题的方案。

And that buys us like the goodwill of being brought into these circles where we're solving the toughest problems for some of the smartest investors out there.

而这为我们换来了信任——我们因此被带进这些圈子，替一些最聪明的投资者解决最棘手的问题。

And means like to your point every once in a while like we see opportunities we see things that we would have never seen before.

这也意味着，就像你说的，我们时不时会看到机会，看到一些本来永远看不到的东西。

We get ideas that are not even like our ideas for a new business.

我们会拿到一些做新生意的点子，那些点子甚至都不是我们自己想出来的。

It's something that we're seeing across our client base.

而是我们在客户群里看到的。

These are people are like much smarter and better connected and resourced than I am and they think this is a cool idea.

这些人比我聪明得多，人脉和资源也比我强得多，而他们觉得这个点子很酷。

And we're like, "Oh, we actually might be able to to do something in that in that space."

我们就想：噢，我们说不定真能在那个领域做点什么。

And that has led us to what you're you're alluding to is like we've we've always sort of dreamt of doing this and we're finally at a point where we have some resources to do this.

这就引出了你刚才暗示的那件事——我们一直梦想做这件事，现在终于到了手里有点资源、可以去做的时候。

It's there to create an incubator where we can take some of these things that start off as like one-off projects for let's say a portfolio company for one of our clients where we built out by the way that's another side of the the business where we are working we're getting brought in by the investors into the portfolio companies to help them.

就是做一个孵化器，把其中一些一开始只是一次性项目的东西接过来——比如给我们某个客户的一家被投公司做的项目，顺便说一句这是我们业务的另一面：投资人把我们引荐进被投公司去帮他们。

Like I mentioned that often times we know more about the company than they do themselves.

就像我提到的，很多时候我们比公司自己还了解这家公司。

And so the investors like, "Hey, you guys should talk to them and see if there are projects you can do directly for for the companies."

所以投资人会说：“你们该跟他们聊聊，看看有没有能直接给这些公司做的项目。”

Which is awesome by the way.

顺便说一句，这挺棒的。

It's it's it's a great that market has a lot higher TAM than the the hedge fund world.

而且那个市场很不错，TAM 比对冲基金那个圈子高得多。

And anyway so we're similar to Palantir in that regard.

总之，在这一点上我们跟 Palantir 有点像。

We're we're building solutions like tech solutions for these companies.

我们在给这些公司做解决方案，技术上的解决方案。

But we are coming at this from the investor mindset.

但我们是带着投资人的思维切进去的。

Like the investor brought us in like what we're thinking about we're not a bunch of like tech nerds that build software for the sake of building software.

是投资人把我们带进来的，所以我们琢磨的是——我们不是一帮为写软件而写软件的技术宅。

We're actually going and solving business problems which are often like how do you increase pricing?

我们真正在解决的是商业问题，往往是：你怎么把价格提上去？

How do you grow your EBITDA?

你怎么把 EBITDA 做上去？

How do you position for a sale to a strategic?

你怎么为卖给一个战略买家做定位？

How do you tell that story?

你怎么把这个故事讲出来？

And how do you get the the strategic to believe it and to see it in the data and the systems that you've built that there are synergies and there are growth opportunities and that's why they should pay a lot more money for this portfolio company.

以及你怎么让那个战略买家相信它，让他在你搭的数据和系统里看到：这里有协同效应，有增长机会，所以他们该为这家被投公司多付很多钱。

And so and these yeah and some of these systems that we've built like we can like use that system for another industry or other companies that have similar problems.

而我们搭的这些系统里，有些可以拿去用在另一个行业，或者其他有类似问题的公司上。

And the the challenge for us honestly has been finding like what I would love to to do more of is to essentially spin out some of these projects that we're like taking from zero to one and bring in a CEO to set up a JV and bring in a CEO and who can then take that project from like one to 100.

老实说，我们的难处一直是找人——我特别想多做的，其实是把这些我们从 0 到 1 做出来的项目分拆出去，成立一个 JV，引进一个 CEO，让他把项目从 1 做到 100。

And these are you know entrepreneurs that don't necessarily have an idea of what they want to work on but they know they want to be an entrepreneur they want something yeah they're excited about building something.

这些人是那种创业者：不一定想清楚了自己要做什么，但知道自己想创业，想干点什么，一想到能做出个东西就兴奋。

We've already done the work of taking it from zero to one and looking for people that yeah we want to take this thing from one to 100 off of an idea that again was we didn't come up with.

从 0 到 1 的活我们已经干完了，我们要找的是这样的人：愿意把它从 1 做到 100，而这个点子——再说一次——不是我们想出来的。

It's like came was come up with by some billionaire and they already paid to have like initial project built out for some portfolio company but we think this could be a much better software solution.

点子是某个亿万富翁想出来的，他们也已经付过钱，让人给某家被投公司做出了最初的项目，但我们觉得这能做成一个好得多的软件产品。

You're in probably one of the most unique vantage points of anyone in finance.

你所处的视角，可能是整个金融圈里最独特的之一。

I would agree with that.

这个我同意。

I have a question that's for you more personally, right?

我有个问题，更多是问你个人的。

System2's doing great.

System2 做得很好。

Um when you were telling me about the growth it's it's just growing growing growing.

你跟我讲增长的时候，那是一直在长、一直在长。

You're hiring more people and you're not raising money so it's fully you know, bootstrapped and um which I find nuts doing that in New York by the way.

你在招更多的人，又不融资，全靠自筹（bootstrapped）——顺便说一句，在纽约这么干，我觉得挺疯的。

Not only not only are we bootstrapped but we have enough money that we're now putting it back into our our clients through this fund of funds.

我们不光是自筹，钱还多到现在能通过这个 fund-of-funds（内部母基金）投回给我们的客户。

So it's yeah it's really cool cash position and then just like a financial position to be in.

所以这现金状况真挺酷的，能处在这样的财务位置上也挺酷。

Exactly and that's

没错，而且这……

nuts.

疯了。

Um would you say that System2 though it's doing well is more a stepping stone to greater things?

System2 虽然做得不错，但你会说它更像是通往更大的事情的一块跳板吗？

Do you think that that's what makes this kind of insane?

你觉得是这一点让它显得有点疯狂吗？

Yeah, look I mean I want to take over the world and this is I've had this conversation with with some of these yeah, of these clients and then people in my my network and the downside of hanging out with And and you see them and like they're phenomenal.

是啊，说白了我想征服世界。这个话题我跟一些客户、还有我圈子里的人聊过——跟这些人混在一起的坏处就是，你看着他们，他们真的太出色了。

They're very smart.

他们非常聪明。

They're hard working.

他们非常拼。

But then again, like so so are we.

但话说回来，我们也一样。

And the fact that you get to see them and work with them for many years and not just one by the way.

而你能见到他们、跟他们共事很多年，而且顺便说一句，还不止一个。

This is ⟨?Miracle of Life⟩.

这就是⟨?生命的奇迹⟩。

I mentioned this earlier.

我前面提过这个。

Like when you were at one fund, we have that one billionaire they reported into.

你在一家基金里的时候，只有那一个亿万富翁，所有人都向他汇报。

They're a god.

他就是神。

They're untouchable.

他高不可攀。

There is something extremely special.

那种感觉极其特别。

Like that guy right there was the first billionaire that I had ever met.

那个人就是我这辈子见到的第一个亿万富翁。

I grew up as like, you know, poor kid coming from Romania.

我从小是个穷孩子，从 Romania 出来的。

This was like I didn't even know this kind of money existed.

我当时根本不知道世上还有这种量级的钱。

And so yeah, like they're they're a god when you're there.

所以是的，你在那儿的时候，他们就是神。

But now I work with a bunch of them.

但现在我跟一堆这样的人共事。

And I say, okay, this isn't just like like a one-off thing.

我就想，好吧，这不是什么孤例。

Like there's many of them.

这样的人有很多。

And they come in all sorts of like sizes and flavors.

而且他们各种体量、各种风格都有。

And and and you see the the the awesome things that they're doing, but you also see differences in how they operate.

你会看到他们做的那些很厉害的事，也会看到他们在做法上的差异。

And it really is inspiring to be in the company of these types of people.

能跟这类人待在一起，真的很受鼓舞。

Like I mentioned earlier, I'm comparing myself to like Ken Griffin.

就像我前面说的，我拿自己跟 Ken Griffin 比。

He doesn't know who I am, but like I've I I've experienced a bunch of the the effects of the stuff that that he's built.

他不知道我是谁，但他建起来的那些东西带来的影响，我实实在在感受过不少。

And I know a bunch of people who know him personally, right?

而且我认识一堆私下认识他的人，对吧？

And so it is yeah, like I do want to do a lot more.

所以是的，我确实想做更多。

The there are um but yeah, like System2 is and will be a great company that connects these very curious people who want to be on the cutting edge of research and want to bring us along because they value the the type of work that that that we're doing.

但 System2 现在是、将来也会是一家很好的公司，它把这些极其好奇的人连起来——他们想站在研究的最前沿，也愿意带上我们，因为他们看重我们做的这类工作。

And so it's in some ways like I've aligned myself I've created a company that allows us to have mentors that would be insane to try to get otherwise.

所以某种程度上，我等于把自己摆到了这个位置：我做了一家公司，让我们有一批导师，而这些导师换别的路子几乎不可能请到。

And they're paying me us for their mentorship.

而且是他们付钱给我们，来给我们当导师。

Which if they're listening to this, I mean that's that's how they how you know how I really feel about this.

如果他们在听这期节目——我说的就是我心里真实的想法。

It's like I would have done this stuff for free.

这些事我白干也愿意。

And in the beginning like I actually didn't pay myself for the first number of years.

而且一开始，头几年我其实没给自己发过工资。

I really actually was doing it for free because I think there is something really cool about being in their company and learning from them.

我是真的在免费做，因为我觉得能待在他们身边、跟他们学东西，是一件特别酷的事。

And the fact that now we're able to yeah, like make a living off of the the mentorship that they are offering us is a very very unique experience to to to be in.

而现在我们居然能靠他们给我们的这份指导谋生，这种体验非常非常独特。

So I don't know if there's any piece of advice for people trying to to start companies or even go in this industry, I think having a clear sense of who you want to be like, the people that you want to associate yourself with.

所以，如果说对那些想创业、或者想进这个行业的人有什么建议，我觉得是：清楚地知道你想成为谁，知道你想跟哪些人在一起。

And aim high.

还有，把目标定高。

Like higher than you ever thought possible.

高到超出你以为可能的程度。

There's insane value in that.

这里面的价值大到离谱。

And if you find a way to get them to pay you to hang out with them, that's a goldmine.

而如果你能想出办法，让他们付钱来跟你待在一起，那就是一座金矿。

And then yeah, like there's always opportunities that can come out of of these relationships that can become like stand-alone businesses that can be more scalable.

然后，这些关系里总会长出机会，它们可以变成独立的生意，更能规模化。

It can be product businesses or whatever.

可以是产品生意，或者别的什么。

They can be a fund of funds.

也可以是一支 fund-of-funds。

I mean there's so many different things that that can that can come out of this, but at the end of it the core of it, it is about having these close relationships with a group of people that are very inspiring and that I'd want to hang out with even if I was paying them to do it.

从这里面能长出来的东西太多了，但归根到底，核心还是跟一群特别能鼓舞人的人保持紧密关系——哪怕要我付钱才能跟他们待在一起，我也愿意。

And in speaking to these two dozen billionaires that are paying you to provide you with mentorship, what's one thing that you've seen, one trait that you've seen um in each of them that makes them special?

跟这二十几位付你钱、又在给你当导师的亿万富翁打交道，你在他们每个人身上都看到的、让他们与众不同的一个特质是什么？

Yeah, I mean that's a great question.

这是个好问题。

So And they they really there are some commonalities, but there are a number of differences.

他们身上确实有一些共性，但差异也不少。

Like I'll um one conversation I had with uh with one of them was around this like how cheap they are.

我跟其中一个人聊过一次，聊的就是他们有多抠。

I asked the question of like "What should I do to diversify as my net worth has grown as these you know these things are achieving more things?

我问了个问题：“随着我的净资产变多、这些事情做出越来越多成绩，我该怎么做分散配置？

Like what should I do to diversify?"

我该怎么分散？”

And he looked at me like I was crazy.

他看着我，像在看一个疯子。

uh and then this guy he was uh he was an operator of um of a public company.

这个人是一家上市公司的经营者。

He was like the CEO chairman of a public company.

他是那家上市公司的 CEO 兼董事长。

And he's like, "You told me you one day want to be like me."

他说：“你跟我说过，你有一天想成为像我这样的人。”

And I was like, "Yeah."

我说：“是啊。”

He's like, "Don't diversify.

他说：“别分散。

You should bet it all on the thing that you're trying to do."

你该把全部都押在你想做的那件事上。”

And I said, "Yeah."

我说：“嗯。”

I guess this I guess that makes sense.

我想这确实说得通。

Like stop wasting your time trying to like be whatever talking different whatever like ETFs like mutual funds whatever your private bankers trying to pitch you on.

别再浪费时间去搞什么 ETF、共同基金、你的私人银行家想推销给你的那些东西。

He's like, "Bet it all and try to like focus on on the thing that you're working on to to get there."

他说：“全押进去，专注在你手上做的那件事上，去做到那一步。”

Um but then yeah, the other ones I think one of the biggest commonalities is how curious they all are.

至于其他人，我觉得最大的共性之一，是他们都极其好奇。

To the point where it's like there's this this one um one one client who is um he's really old.

夸张到什么程度呢——我们有一个客户，他年纪很大了。

He's probably in his like like 50s, maybe even older.

他大概五十多岁，可能还更老。

So he's been he's been around the block.

他什么场面都见过了。

He has achieved a lot.

他已经取得了很多成就。

And I found out from one of one of his other guys that he's been taking coding classes at like, you know, the the local university, which the local university for him is a very well-known university.

我从他手下另一个人那儿知道，他一直在本地大学上编程课，而对他来说，“本地大学”是一所非常有名的大学。

Um and and he's like, "Why are you taking Python classes when you're like in your late 50s?

然后就有人问他：“你都快六十了，为什么还去上 Python 课？

You've achieved so much.

你已经取得这么多成就了。

Like what are what are you trying to do?"

你到底想干什么？”

And his answer is that he goes, "I just I want to learn.

他的回答是：“我就是想学。

Like this is where the world is going.

世界就是往这个方向走的。

I want to understand how to do this."

我想搞懂这东西该怎么做。”

And and yeah, like one of his one of his like this is right-hand man described him as the king of side quests.

而且他的一个左右手把他形容成“支线任务之王”。

Loves like going down rabbit holes and like learning different things.

特别喜欢一头扎进各种兔子洞，学各种不同的东西。

And and that is if you boil it down, like a core skill set that is needed when you are a fundamental investor.

而这个东西归结起来，正是做基本面投资必备的核心能力。

It's like are you just curious about the world?

就是：你对这个世界到底好不好奇？

And and to them like kind of going back to like what would you do if they don't they didn't pay me to to to learn.

对他们来说，这又回到那个问题：如果没人付钱让我去学，我会做什么。

Like their job is basically but they're getting paid to go out there and like ask all sorts of questions about these companies, learn about the direction of the world is going in.

他们的工作基本上就是——有人付钱让他们出去，围绕这些公司问各种各样的问题，去了解世界正在往哪个方向走。

And it is um yeah, and I it's something I can definitely relate to.

这一点我完全能感同身受。

Like I love a lot of random things.

我喜欢一大堆乱七八糟的东西。

I love learning about many different things, which is why like we set up System2 the way it is.

我喜欢学各种不同的东西，这也是我们把 System2 做成现在这个样子的原因。

Like we're not working on one product.

我们不是在做某一个产品。

Like every single day we're answering a different question about a company in an industry I didn't even know existed.

而是每一天都在回答一个不同的问题，关于某个我压根不知道存在的行业里的某家公司。

I know it's something I told you last time.

我知道这个我上次跟你讲过。

Like we have one of our largest um corporate clients is a sand mining company.

我们最大的企业客户之一是一家采砂公司。

We get what what in on that that phone call the first five calls.

我们被拉进那通电话，最开始那五通电话——

I didn't know sand was mined.

我以前不知道砂是采出来的。

I just thought you go to the beach if you need some.

我以为需要砂子的话，去趟海滩就行了。

You know, this is like massive industry.

结果这是个巨大的行业。

Apparently it's like the second most mined natural resource on the planet after water.

据说砂是地球上开采量第二大的自然资源，仅次于水。

And it's it's because everything construction like cement, glass.

因为建筑相关的一切——水泥、玻璃。

Like there's so many use cases of of sand.

砂的用途实在太多了。

Um but yeah, you go into like something like that.

总之，你就这么一头扎进这样一个东西里。

You didn't know this thing existed as an industry.

你之前都不知道有这么个行业存在。

And then within like days, hours, like maybe on the same phone call, I have to tell these people something about the industry they've worked in their whole lives that they don't know.

然后在几天、几小时之内，甚至可能就在同一通电话里，我得告诉这些干了一辈子这行的人一件他们不知道的事。

And I didn't even know this thing existed 5 minutes ago.

而 5 分钟前我连这东西存在都不知道。

So how do you solve that kind of problem?

那你怎么解这类问题？

I love that, right?

我太喜欢这个了。

Being able to like go with your I'm being thrown into the deep end of the thing I didn't even know existed.

就这么被扔进深水区，扔进一个我压根不知道存在的领域。

And on that freaking phone call, I have to tell them something that they're like, "Oh, I haven't thought about that before."

而且就在那通该死的电话里，我得说出一件让他们觉得“噢，这我以前没想过”的事。

So that they can hire us then do an engagement to prove out in this case yeah, we're we've helped them with their their pricing.

这样他们才会雇我们，然后做一个正式项目去验证——在这个案子里，我们帮的是他们的定价。

Like helped them increase pricing by leveraging a bunch of a bunch of data.

靠一堆数据帮他们把价格提上去。

And yeah, like that's the that's the beauty of of this job.

这就是这份工作的美妙之处。

And it's very similar to how these investors think.

这跟这些投资者的思考方式非常像。

Like they just love learning about different things, going down rabbit holes, asking tons of questions, connecting the dots between different things you've learned in the past, and then hopefully getting paid and getting paid well to to do the thing that you love.

他们就是喜欢学各种不同的东西，一头扎进兔子洞，问一大堆问题，把过去学到的不同东西连起来，然后最好还能因为做自己喜欢的事拿到钱，而且拿得不少。

I should have framed the question more like this because I think this one this

我应该把问题这样问才对，因为我觉得这个……

way is a lot more interesting.

……的方式要有意思得多。

And you've touched on it there.

你刚才也提到了这一点。

What are some of the differences between these super successful billionaires that surprised you greatly?

这些超级成功的亿万富翁之间，有哪些差异是让你格外意外的？

The differences um You know, one of the big differences that we noticed and this is where what surprised me the most about these about these these folks.

差异……嗯，我们注意到一个很大的差异，也正是这些人身上最让我意外的地方。

And it's then one of them was like, you know, basically how cheap they are when buying birthday cards for their wife.

其中一条是，他们给太太买生日贺卡的时候有多抠。

Another one is for how much time they spend thinking about businesses versus how little time they spend working on their own business.

另一条是，他们花在琢磨各种生意上的时间有那么多，花在打理自己这门生意上的时间又有那么少。

And what I mean by that is that they are in many ways like doctors.

我的意思是，他们在很多方面就像医生。

Like my my brother my dad is a doctor.

比如我哥……我爸就是医生。

They love being doctors.

他们热爱当医生。

They don't love running a doctor's office.

他们不爱经营一家诊所。

They do it reluctantly.

这些事都是硬着头皮做的。

They call that you know, if you're an administrator, it's like they look down on the business aspect of running that thing.

他们会说，你要是个搞行政的——反正他们打心底看不上经营这摊事的那一面。

They just love being doctors.

他们就是纯粹热爱当医生。

And that's why they've been rolled up by private equity firms who love the business side of it.

所以他们才会被私募股权公司一家家整合（roll-up）掉——人家爱的恰恰是生意的那一面。

Well, it's very similar in the the hedge fund world, especially in this in the the the fundamental single manager world where the main the founders love being investors.

对冲基金圈里也非常像，尤其是基本面的单一管理人那一块，创始人热爱的是做投资人。

They don't actually really like running a business.

他们其实并不怎么喜欢经营一门生意。

And and you can see that when like they are still in the day-to-day the the weeds of of the the work.

你能看出来，他们到现在还扎在日常工作的细枝末节里。

Like they will think about companies.

比如他们会去琢磨一家家公司。

They'll think about every single investment.

每一笔投资他们都会琢磨。

Like every one of their investment team like analyst is funneling information to them about that they're ultimately decision-maker.

他们投研团队里的每一个分析师，都在把信息汇总给他们，因为他们才是最终的决策者。

They they think about this.

他们会琢磨这些。

They ask questions.

他们会提问。

They poke holes in in the thesis.

他们会在投资逻辑上挑刺。

If you think about like I don't know like the CEO of Disney like is he working on the next like Pixar film?

你想想，比如说 Disney 的 CEO，他会亲自去做下一部 Pixar 电影吗？

Is he working on the next like amusement ride?

他会亲自去做下一个游乐项目吗？

Like you're not actually you're working on and as an entrepreneur I can totally relate to this like I used to do a lot of engagements myself in the early days cuz I had no choice.

其实你做的并不是这些……作为创业者我特别能体会——早期很多项目都是我自己上，因为别无选择。

But now I spend the vast majority of my time working on the business.

但现在，我绝大部分时间都花在经营这门生意上。

Like how do we grow?

比如我们怎么增长？

What are the strategic directions you want to be in?

该往哪些战略方向走？

How do you manage a team?

怎么带一个团队？

And um and so yeah, that was a very surprising thing when it comes down to these just like management of of a team.

所以是的，落到带团队这件事上，这一点非常出乎我的意料。

How much time they spend thinking about their own teams, their incentive structures.

他们到底花多少时间去想自己的团队、想激励结构。

Like and it is um yeah, that was it was it was very surprising cuz they would just rather be left alone.

这真的非常让我意外，因为他们宁愿谁都别来烦他们。

Like they'll bring in a CEO or somebody to try like work on the business side of it.

他们会找一个 CEO 或者别的什么人来管生意那一面。

But I know that you yourself are interested in like the operational side of of hedge funds.

但我知道，你自己挺关注对冲基金运营那一面的。

In some ways like the CEOs of the CEOs of hedge funds are like CEOs of any other business because they are the ones that are focusing on actually running the thing as a business.

在某种意义上，对冲基金的 CEO 跟任何一门生意的 CEO 没两样——因为真正把这摊事当生意来经营的，是他们。

And this has consequences when you're looking to potentially like exit.

而等你想退出的时候，这件事是有后果的。

Most of these funds can't exit cuz there's no value.

大多数这类基金退不出去，因为里面没有价值。

There's no like IP in the thing beyond the founder.

除了创始人本人，这里面没有什么 IP。

And yeah, they they try to like and they bought it all often don't think about this until it's very late until the the very end of it when they're starting to think about retirement.

而且他们往往要到很晚、到最后开始考虑退休的时候，才会去想这件事。

And they're like oh wait, I should like think about who else I'm promoting or talk to my investors about saying hey, it's okay, you don't have key man risk here because if I leave this entire organization and they will play our part.

然后他们才想到：哦等等，我是不是该考虑再提拔谁，或者去跟投资人说，没事的，你在这儿没有关键人风险，因为就算我走了，整个组织也会各就各位、顶上我这一份。

Whereas in these like larger the the pod shop model you do they do spend a lot more time thinking about the the actual business of it.

而在那种更大的 pod shop 模式里，他们确实会花多得多的时间去想这门生意本身。

And the the the head folks are not actually like managing every every investment.

而那些掌门人其实并不去管每一笔投资。

They've built all these processes to surface interesting ideas and it's it's much more and much more automated.

他们搭起了这一整套流程，让有意思的想法自己冒出来，自动化程度高得多。

And so yeah, that was one of the most surprising things is how um how little time they spend on their own business or even if they spend the time on it like how reluctant they are in doing it versus what they would rather be doing which is to be left alone and to answer answer questions.

所以是的，最让我意外的事情之一，就是他们花在自己这门生意上的时间有多么少；就算花了时间，做起来又有多不情愿——他们更想要的，是没人打扰，只管回答问题。

And to take that to an extreme like we have we have we have a client by the way who has who hires the analysts on like a two-year program.

把这个推到极致——顺便说，我们有一个客户，他招分析师是按一个两年期的项目来招的。

So after two years they basically don't have a job.

所以两年之后，他们基本上就没这份工作了。

They they know from the very beginning they're only going to be there for two years and then they rotate off.

他们从一开始就知道自己只会待两年，然后就轮换出去。

And the idea is like you just want people to like do what you tell them to.

背后的想法是，你只想要那种你让他做什么、他就做什么的人。

You don't want these like analysts to then like start having their own like super strong opinions and feel like they know better than you.

你不希望这些分析师慢慢有了自己特别强的主见，觉得自己比你更懂。

Like they and then so AI that's why AI is so fascinating to or interesting to a lot of them is because you can essentially have a bunch of analysts doing exactly what you tell them to do in your way, your process and not muddying it up with whatever ideas that that they have that you ultimately don't really care about.

所以 AI 才让他们中很多人这么着迷、这么感兴趣——因为你等于有了一群分析师，完全按你说的做、用你的方式、走你的流程，不会拿他们自己那些你根本不在乎的想法把事情搅浑。

Cuz you think about it like what is the the job of a PM?

因为你想想，PM 的工作到底是什么？

It's like weed out tons of things that they're being pitched by their internal teams all the time and only focus on the stuff that they really like.

无非是从内部团队天天推给他们的一大堆东西里筛掉绝大部分，只盯住自己真正喜欢的那几个。

And that that fits their business like their model, their view of the world, their framework for for investing.

而且要契合他们那一套——他们的模型、他们看世界的方式、他们投资的框架。

But if you can program that framework into a bunch of agents that will now everything they do filters that like everything they present to you will already have that filtered.

但如果你能把这套框架写进一群 agent 里，那它们做的每一件事都带着这层筛选，呈到你面前的东西都已经筛过了。

You can end up with like an army of helpers that you don't even yeah like you don't need to hire for anymore cuz you can just have the the AI do it.

你最后就能有一支帮手大军，你甚至都不用再去招人了，让 AI 做就行。

So yeah, that was surprising to me.

所以是的，这一点让我很意外。

Not surprising to me as well that these guys though they are billionaires and though they are so interested in businesses they neglect their own in a sense.

同样不算意外的是，这些人虽然是亿万富翁、虽然对生意这么感兴趣，某种意义上却疏于打理自己的生意。

But yeah, they they love what they're doing.

但没错，他们热爱自己在做的事。

Kind of going back to like I I think the doctor example is so perfect.

回到刚才——我觉得医生这个例子实在太贴切了。

Like they just love being a doctor.

他们就是热爱当医生。

Like there's nothing else they like they like interacting with patients.

别的都无所谓，他们喜欢跟病人打交道。

They like the work.

他们喜欢这份工作。

They like reading up on all the things.

他们喜欢去把各种东西读个明白。

They don't like being like oh let's schedule the next whatever like nurse's shift or dealing with the front desk person who called in sick or quit or like figure out how to work with like the billing company to get paid more for these things.

他们不喜欢的是：哦，来排一下护士下个班次吧；或者处理前台的人请病假、辞职；或者去琢磨怎么跟账单公司周旋，让这些项目多结点钱回来。

It's like that's not what drew them to that industry.

当初把他们吸引进这行的，不是这些。

Like what drew them to the industry is like this curiosity and passion for the the role itself.

吸引他们进这行的，是对这个角色本身的那份好奇和热情。

And yeah, like in some ways these doctors or these fundamental investors are like the old school artisans.

而且在某种意义上，这些医生、这些基本面投资人，就像老派的匠人。

Like they're just doing this for the sake of the craft.

他们做这件事，纯粹是为了手艺本身。

And that's all they love.

他们爱的就只有这个。

Like they love the craft of that job and everything else is just a way to to that to that end.

他们热爱的是这份工作的手艺，其他一切都只是达成它的手段。

Um that's what makes them good, right?

他们厉害就厉害在这儿，对吧？

You want them to actually like you want the head person still involved in thinking about these things.

你希望他们真的……你希望那个一把手仍然亲自去琢磨这些事。

And versus be like oh like let's move on to to the next thing.

而不是「行了，下一件事」那种。

Um so yeah, I know it's it's I would say it's it's actually it's it's a feature not a bug, but the result of it is that yeah like it leaves opportunities for the management of the business.

所以是的，我会说这其实是特性、不是 bug，但结果就是，它给经营这门生意这一块留出了机会。

There are other people that are very talented like COOs and other folks that are thinking about this or like people are managing the research teams and there are tons of um whatever like CTOs and other folks that are dealing more with the operations of the of the business.

还有另外一些很有才干的人在想这些事，比如 COO 之类，或者带研究团队的人，还有一大堆 CTO 什么的，他们更多是在处理这门生意的运营。

Um but yeah, like their their heart is in the art of the job.

但说到底，他们的心在这份工作的艺术上。

I love that and I think that's great advice advice.

我太喜欢这句话了，我觉得这是非常好的建议。

Fall in love with the craft.

爱上这门手艺。

Fall in love with the actual work and you'll be successful.

爱上真正的活儿本身，你就会成功。

Thank you so much for coming on Odds on Open ⟨?with say⟩ this was wonderful.

非常感谢你来 Odds on Open ⟨?with say⟩，这期太精彩了。

Thank you so much Ethan for having me.

非常感谢你的邀请，Ethan。

It was a it was a great conversation.

这是一场很棒的对话。