How To Generate Alpha in Venture Capital | Albert Azout, Level Ventures

22:07Turner Novak:So actually, one thing you mentioned earlier was that luck plays a pretty good role or pretty big role. I feel like a lot of investors would not like to admit that. Well, maybe people can't admit that in certain senses, but what role does luck play? Yeah, we don't know by definition what it plays because there's no counterfactual. There's no way to re-stimulate what occurred. Can you measure it somehow? Or how do you underwrite it or incorporate it into your process? Yeah, it's more like we think of it as randomness. And so that's why portfolio construction really matters. I think because, you know, you, you, there's always an inherent uncertainty on things.

22:54And so, and luck is really when like you're sort of, you know, there's an opportunity and you're prepared and like you're able, you know, so even luck itself is like, there's a context to it. You just sit, you know, they sit at home, they don't do anything. It's hard to have luck. So there's some, there's some action that needs to happen.

23:09Turner Novak:That could be a measure of like a pretty defensible, like persistent return network though, You sit at home, you do nothing, and you still are getting the grand slams. Great for things to come your way. Yeah. It's very hard. I think you just got to just be comfortable with the randomness and just the randomness of things. We cannot predict the future. There's no way to predict the future. It's inherently unpredictable. What you can do is try to position capital that will benefit from future states of the world, I think, is the way we think about it. And some of those could be just luck, you know, or yeah, just being at the right place at the right time.

23:46So, you know, like just those things you just, you just never know.

23:50Turner Novak:And this is a question from a little bit different topic for my friend, Sasha, Creator Ventures. Is there anything you're seeing on data versus spin out versus new, newly created or kind of like outsider type funds? Have you guys done any research on that? We haven't done specifically research on that topic. I think there's a set of personas that we like. We gravitate to. We do like the outsider, either operator, turned investor. We like that a lot. We think it's a really good persona. I think it attracts founders. I think there's just a different way of looking at the world often. We also like investors that have worked their way up.

24:39Not necessarily like partner GP level, but hungry early, have really deep networks, but they want to go on their own. We think that's a really good persona. And we think that that strategy has done well. Versus like a large GP leaving and trying to build a big firm. And it's just not the persona we tend to like as much.

24:57Turner Novak:How do you kind of suss that up? Because I know a lot of people will say, Like there's one argument to say, you know, they're they're unshackled from the partnership and the investment committee and they can maybe put some of their own deals through. There's another argument that says, oh, they don't have the brand of the mega platform. And, you know, the way they get their deal flow is probably going to change when they're not attached to that. How do you guys kind of underwrite or or get comfortable with how that'll change? Yeah, I think we it's a lot of chemistry also with the GP and just getting a sense for them.

25:26There's a lot you can do. We do look at a lot of data, depending on where they've come from, like if there's attribution or not attribution, or if they have, usually we're looking at a portfolio that's already sort of extant, like they were already developing a portfolio and we can look at it that way. So even with a few data points, we can start to get an assessment of it. And then we do a lot of backchannel on that particular thing with founders and other GPs and other people on market to get a sense for whether the GP is a signal for them in the market. and that's kind of how we think about it.

25:58We sort of ask the inverse question.

26:00Turner Novak:You ask the inverse of what? We ask the market, if this GP were to show you a deal, would it be a high signal for you? Just like, yeah, is that the key thing? Because we want to see this, that we want to see that the GP actually is a signal for the market. And you can't fake that. you really can't fake that don't people kind of do like hey man I'm gonna you know I'm raising my phone when people reach out tell them this like here's your talking points people kind of do that don't they? we don't ask them for references we go out and we talk to people that would give us an unbiased opinion oh yeah fair yeah and there's people you have to really everybody communicates differently you have to kind of read between the lines of a lot of what people say and just try to understand the picture and try to figure out what you think you need to de-risk about this particular GP.

26:55But yeah, we try to just get as many counter data points as possible and try to form an opinion.

27:00Turner Novak:And then a lot of it is also just in life, I think you get the investors you deserve. And a lot of it is just the chemistry and the interactions over a period of time that will make us comfortable just generally. At the end, we're investing in people. Actually, Amanda Robson at MTF mentioned that you had the most intense reference process or most references out of anyone else. I don't know if that's a good or bad thing, but it seems like you're doing that at least. Yeah, I think we're trying to be very objective. If you look at our data approach and understanding the embeddedness of a GP in the network instead of talent flow, and then being able to back that up with in-market checks, I think is important.

27:51That just creates a more objective approach versus just how else would you do it? You have to look at their presentations and their advisors and all these things that are maybe more intangible. So, you know, we try to get to the bottom line. Yeah.

28:08Turner Novak:When you're, when you're like meeting a new manager or, or doing diligence on someone, like what is the general process look like? Like, is this a one zoom call one day type of thing, or does this take you five years? I'm assuming it's somewhere in the middle of that, but. You know, it really depends on how fast things are moving, to be honest. Yeah. We try to like pace ourselves with how fast an opportunity is evolving. And if we have more time, then we'll, we'll take more time. Right. I don't think most LPs will tell you that, but it's true. If we're not in a rush, we'll try to take our time.

28:41Turner Novak:Yeah. I feel like a lot of people mess that up on the fundraising side is you just say you don't give it a time in a box. For sure. But also, maybe the other way around is they're also somewhat short-termists. And they try to squeeze people to an answer, but they're not yet ready and it takes time. So you have to figure out each LP is different and each GP is very different. But yeah, we try to, you know, obviously just kind of synchronize with the timeline that's set forth that we think is fair. Because, you know, we see a lot of opportunities, so we have to kind of manage that. And then manage back with a team, the only two of us on the investing side right now.

29:21Turner Novak:So I really want to do a demo of the software. Maybe we'll do that in, I don't know how long this will take us, 10, 15 minutes later. But just thinking about, you mentioned something earlier that I know you've written a lot about, criticality investing. Did I say that right? Yeah, criticality. Yeah, so what is that for people who don't know? Yeah, so we've been, you know, I've been, like the team and us, we've been thinking for a long time, like, how do markets work? How does the world work? That's kind of the question we always ask ourselves. And there's certainly, you know, sort of higher level phenomenon about the way the world works that you can ascertain and sort of use as a guiding principle or a set of, like, a frame of reference when you're investing.

30:02a mental model right it's not to say that it's oh it fits every situation but generally speaking i think there are things that are generally true as like true phenomenon about the world

30:13Turner Novak:so what would be an example about something like this yeah i guess with criticality i'll give you an example like you know we we know that for for example in the world um you know there's things organized into networks generally speaking which is like the relationships between things. This is not even investing related. This is even like... Just generally, yeah. Like if you think about like, yeah, for example, social networks and I don't know, protein networks, you know, sort of chemical signaling networks, the neurons in the brain, you know, like even human-made networks like the grid, et cetera, et cetera, you know, airports and sort of like travel paths, et cetera.

30:53All those are essentially networks. You know, there's entities in the relationships between them. and networks that form in real life, they have a set of properties about them that make them really interesting and difficult. You know, for example, you know, like the degree distribution in networks, meaning how many friends or how many, you know, sort of outlets you might have for a specific edge is parallel distributed, right? So there's going to be ones that have, you know, hundreds and hundreds of relationships and there's going to be many that just still have very few.

31:26Turner Novak:You say people, people that have many relationships. It could be people, but it could be anything. It's not just people. It could be even in protein signaling and all kinds of networks. That's how they merge in that form, generally speaking. That's just like one sort of property of them. And there's many other kinds of properties about them. But these networks are very complex, and they evolve in complex ways, and they're unpredictable. Because if you think about, for example, like a tweet on Twitter, we don't know exactly which tweet will cascade. and suddenly all of Twitter knows about it and which ones won't.

31:58And a lot of it depends on the structure of the network where it started, who tweeted it, when, and the context, et cetera. So a lot of the behavior of networks tend to be very unpredictable. But one of the things networks do is they end up organizing networks, and I was talking about markets, the same thing. They end up organizing themselves in a way that they're both in a state of order but also can change very rapidly. And we have a whole article about how that works. But whenever you have a situation where there's a rapid change and the network's reorganized, for example, you think about the AI wave.

32:36Before this, and we'll talk about that in the context of criticality, but before this LLM moment, the network was organized as such. And it was maybe focused on vertical software and fintech and all these different things. The network of investors was organized in a specific way. then you have this moment right and now the whole network needs to reorganize and change because now you have different talent clusters you have different companies that evolve you have talent that goes in different ways you have new investors that have come in because like they have relations with open ai etc so the whole network sort of evolved so you could say the networks really was really you know stable before and then suddenly there was like a new event and then now it all shifted and the only way for that to happen is that the network itself had to have in it the ability to change with rep because all these behaviors are decentralized it's not like people are coordinating at large scale they're you know just making local decisions and then suddenly there's like a sort of a structure that forms so that's the that's the context so the way we think about criticality is that a couple things one is that we the way we we believe that technology evolves it's similar to the llm moment is that there's very very slow or what is apparently kind of slow, under-the-radar progress.

33:50And then there's a moment where there's a punctuation event, and then there's a big shift. And the reason why that happens is because a lot of these changes were under the radar, per se, or under the attention of most. And then there's something that happens that changes the way people perceive things.

34:06Turner Novak:And then there's a really big shift. And so the way we think things evolve and why we call it criticality is that networks are in a state of criticality. There's a market that maybe has a lot of incumbents in it and hasn't changed, hasn't iterated. but something that sort of necessarily happens, and then the whole market sort of reforms and rechanges. And it's due to the technology innovation, but it's also due to other factors. So that's what we call criticality investing. So what we're basically doing at any point in time is we're looking for where in the world are there latent stresses that maybe the market hasn't perceived and trying to figure out if those may change at some point in the future.

34:43And that's how we think about things.

34:45Turner Novak:So it's like your favorite networks to invest in. You think about yourself, you're investing in networks. You're trying to find networks that are near some sort of a massive changing moment of some kind. Networks and markets that are under a lot of stress and duress that's sort of unobserved. It could be like there's a big disequilibrium in the market. There's a lot of friction that maybe the market hasn't yet perceived. Internal people perceive it, but the market doesn't perceive it. Or when I say marketing, I mean capital hasn't perceived it yet. And then something we think will change, whether it could be even a company that comes online, like, for example, like Palantir, you know, or even Anduril and Defense Tech, you know, before Anduril, there really wasn't even a focus.

35:28You know, there wasn't a lot of capital going into Defense Tech at all. In fact, VC's never invested in Defense Tech, you know, as a rule.

35:34Turner Novak:Yeah. And I remember even like, I remember Google was going to do a big deal with the Pentagon in like 2017, 18. This was like a round of Anduril. Yeah. And they like, people protested, like, you can't work with the government. It was like, or work with the Department of Defense. I guess. It was like anti-signal. You could say the military-industrial complex was under a lot of disequilibrium and, let's say, stress to some degree, because they had this cost-plus kind of approach, and they were sort of behind on innovation. There was a lot of dependencies and frictions in manufacturing and scalability and legacy systems.

36:12You can say all these different things and that existed. Right. And the reason why it stayed that way is because there's just a lot, there's a lot of, besides technology, there's a lot of social technology that's in place that creates friction. And all these things were already there and accumulating, accumulating, accumulating. And it wasn't maybe until like there was geopolitical crisis, bundling with China. And then maybe then, you know, all of that together and suddenly everybody's rotating capital over. So if you had positioned capital in the right companies before, or if you had your kind of ducks in a row right before that change, you know, that's probably where the most asymmetry is, right?

36:45Because you're, you know, that's kind of, if you think about it, that's where the most money would be. So it's almost like in a way, this is going to sound really bad, like, like front running capital

36:55Turner Novak:flows almost like, that's really what it comes down to is like, it's your point, it's like, allocate a bunch of money to defense, a bunch of capital flows into defense, and you benefit from being in front of all the other capital. I mean, maybe that's not the only thing that you do, Well, it's the truth. So if you think about the markets, generally speaking, it's very rarely that markets are in a state of efficiency. Usually, they're in a state of – they're usually in a dislocation of value and price, almost always. They're always in this equilibrium of price and value. So they're either undervalued or overvalued.

37:31Yeah, exactly. Fear or greed. There's either fear or greed or whatever. and they're almost never in this state of equilibrium. And so if you think about the returns, even over time in the markets, in the S &P, et cetera, most of that has come from multiple expansion. 90%, I think, has come from multiple expansion versus actually earnings growth, right? So a lot of money is made by capital flows. And in fact, if you knew where capital flows were going, that's even more important than investing. If you were a value investor in the last 15 years, you've lost money, right? So anyway, so if you think about it that way, so capital flows is pretty much everything.

38:12And I think even more so these days because people's attention, the complexity of information processing and attention, et cetera, it's just much more simple to invest in an ETF, you know, and sort of passive investing in factors as opposed to thinking about individual companies or like these sort of big scale changes. And so capital flows really is what drives price and price being sort of completely dislocated from value. And it can be dislocated for very, very, very long periods of time. And so I think capital flows really drives almost everything.

38:47Turner Novak:And capital flows almost a downstream of attention or interest or... Yeah, exposure. I mean, eventually, asymptotically, of course, you need to invest in a company that has higher earnings potential, high pricing power, can have economies of scale, et cetera. And eventually, they merge at some point. There's a come to Jesus moment. It's like, do you make money or not? Exactly. And then for sure, it will happen. But I think it's more important to predict what the next, not so much if something's overvalued, but where are the valuations going. Right? Right. And that's, that's, which is hard to do.

39:26I'm just saying that that's kind of like what we've noticed. And so criticality that, that amplifies, that's why criticality works is it ampli it's amplified by floats and narratives and attention. Right. Cause the markets are not rational. There's, there's no action. Yeah. We know that.

39:43Turner Novak:Yeah. Well, I mean, I feel like then there's a big kind of storytelling component of it. I've always thought Elon did a good job with this. Like I know people say like, Oh, you could criticize a lot of things he's done. how by the books he did things, etc. But like, he did a good job of getting attention and specifically in the capital markets, like reducing the cost of capital, the cost of equity, whatever form you choose to do that. I feel like he, I feel like it's kind of like more of like played out and well-known now. And like a lot of other people are kind of using this strategy, especially in 2025, you know, second half of 2025.

40:19Turner Novak:But I feel like he did a really good job of that probably over the past 15 years. using that to his advantage when maybe he didn't have enough there yet, right? Like you think of like Tesla being overvalued for like such a long period of time, but now kind of have higher operating margins like every other car company. Like it's, you could argue it's a better business, even if it's always been overvalued. Yeah, hopefully, yeah, you can use narrative as a way to sort of bootstrap a business, right? But yeah, but a lot of things move with narratives. It's very common. I think most narratives are overly coherent.

40:56They sound good and they make sense, but if you think about in practice what that means and what does it say about the process it's talking about, for example, maybe humanoids, just not to bring into it. If you think about that, okay, what does that actually imply? Yeah, we're going to a place where there's labor shortage and there's going to be more automation and these robotic foundation models. But what is the underlying thing that needs to happen for like there has to be a lot of regulation? So the timescales might be very different than the markets are responding to narratives. But that does give a lot of these humanoid companies the ability to get more capital and to sort of build a future that they want.

41:41So it works both ways. Yeah.

41:44Turner Novak:I mean, when you think of the narrative of that, it's like labor. Like how much your GDP is just people doing work? And we just played out like digital labor is open AI and anthropic. Now the humanoid is physical labor. Like the entire economy, basically, AI is going to transform. I actually saw it was probably the worst argued point I've ever seen. But someone said the AI market was like a$370 trillion opportunity. And it's kind of one of those things where you're like, this has to be the top. Yeah, for sure. I mean, there's always truth, you know, in a lot. Like there's always, yeah, for sure, it's completely disruptive.

42:29And it's going to be large scale changes in society for sure. And at the same time, the time scales and what it means is going to be different than what people expect. Right, I think.

42:42Turner Novak:Is that something you think people get wrong a lot with this is just like the timing? like it's that whole adage of less happens in one year than you think but more happens in 10 years yeah i think criticality has a lot to do with that you know and these cycles of narratives and sort of like the overly coherent narratives and then sort of the breaking down of the narratives to like realism and then back around and back so that that whole process has a lot to do with time scales right because the reason why there's convexity is because you know some of these things is because time scales are off right yeah that's the whole point you know they're they're Essentially, people, when they're investing in something that's overvalued, they're saying that they're bringing the returns from the future deeply into today with heavy, heavy, I guess what they call hyperbolic discounting.

43:26Turner Novak:I've never heard it described as hyperbolic discounting, but okay. So yeah, that's a behavioral mechanism. It's a behavioral finance term, but it's like you're discounting deeply the future. And that is just a function of distorting time. You can think of it that way, as opposed to thinking about time being sort of over the years. as you can think, but it's like a distortion of time. And so, yeah, that's something that investors do, right? It's something I do. And sometimes it works. So. There's a really interesting, probably related to some of this in a way, from Martin Casado at A16D. I'll read the tweet.

43:59Turner Novak:We'll throw it up on the screen for people too. It says, the idea that non-consensus investing is where the alpha is, is actually quite dangerous in the early stage. Follow-on capital tends to be more and more consensus aligned. I agree. I think that's a very, very solid statement. But wouldn't you say, looking for alpha, you want to be non-consensus? No, what I'm saying, that's why I never used the word non-consensus. I never used it in that definition because at the end of the day, you need to go where capital is going. We just spoke a little while about capital flows. You can't ignore capital flows.

44:32You want to position yourself to where capital flows are going. There's many ways to do that with alpha, or there's several ways to do that with alpha. And you want to do it in a way also that you get paid for the risk. Right. Because, you know, because if you, especially with fund sizes, like you were paying$50 million caps on your seeds, like you're, it's going to be very hard to return upon. Right. So there's something about price.

44:51Turner Novak:You almost have to be like early consensus. Like you, you want to be like the first of the consensus, and then you benefit from the follow-on flows. And that's actually where the, the, the highest form of. Yeah. So that's what we call criticality. That's what we call it in our business. Like what we call like criticality investing, which is the way to do that. You can't, the world, You can't ignore the fact that technology needs to solve problems, right? It needs to solve problems for sure, right? So there are areas where there are pronounced problems that are not yet in the capital rotation, but they're going to be.

45:24And so that's sort of one way to think about it. Another way to think about it is like there's, you know, the first order stuff, like we know that, you know, sort of like the AI platforms. And then there's a second order, right, which is you have to sort of predict like because of this, what do I need? Well, we need to think about MCP servers. And then there's a third order, right? And you want to figure out where the consensus will be in second and third orders before others do, right? And so that's the other way to think about it. But yeah, but you can't say I'm not like, we're going to be non-consensus, but that doesn't make much sense, I don't think.

45:59Turner Novak:Yeah, no, I mean, you have to be consensus. Capital is more and more concentrating. It's like saying, oh, we want like NVIDIA, everybody's after NVIDIA. Why do we invest in NVIDIA? Because it's everybody's doing NVIDIA. That makes no sense because capital is concentrating more and more and more on the winners. And most, yeah, so that wouldn't make sense. But what matters is like positioning your capital to where capital will be rotating to in a correct way. Is there a time to think about getting out of a trade? Like, for example, the NVIDIA one, right? It's like NVIDIA has been the biggest winner in AI.

46:34Turner Novak:It probably has more than 100 % of all free cash flow generated in AI, right? They've obviously been the big beneficiary. But is there a time to think, how do you figure out capital flows changing? Is there a way? I think it's impossible. I really, truly think it's impossible to predict the markets. Anybody who predicts it, you should just stop and fuck you off. It's impossible. An economic forecast we know over time, it's better just to pick things randomly. No one knows the future. What I think you can say about Nvidia and what you can say about the market structure today is that it's mostly driven by passive capital flows.

47:16And passive capital flows are inelastic to price. So as things, for example, go up, they continue to go up because these ETFs need to rebalance. and you have a situation where there's elasticity. And if there's any negative news in that sort of environment, there's going to be a big sell-off.

47:35Turner Novak:A big shock because the people who are active immediately get out. Yeah, everybody will get out. It will just roll and it will just roll. It just continues to roll. I wouldn't be surprised if you have a 30 % or 40 % or 50 % drawdown. I don't know, in the market. So we just don't know when it happens or if it happens, etc. But that's the structure of the market today. And so, you know, if you have enough gains, like I guess you just want to manage risk. You sort of want to manage risk. I actually saw a chart. There are now more ETFs than there are individual publicly traded companies. I don't know if you saw this too.

48:08I didn't see that, but we track, a lot of the data that we track, we also track a lot of the public markets. And we look at a lot of the ETF flow, not the flows, but looking at the ETF holders in different companies and trying to understand, you know, how that kind of works. It's interesting to us to kind of understand that, especially because we're looking at the intersection of the private markets. But yeah, there's an overexpression of both active and passive indexers and mutual funds. And it's just, it's a very noisy, very noisy market.

48:41Turner Novak:So talking about noisy and we talked about randomness a little bit, we were talking before about kind of persistence of returns in venture. Do you think are returns persistent? I think you mentioned you don't think they are. There are for very few, but they're mostly not. I guess, you know, our data shows it's really not persistent. Top quartile persistence, you know, from year one, I think is very, very low. I forget the exact numbers. It's like, it's less than random. Like, it's like random. And also, by year three, it starts to improve a little bit. Like, so if you're like top decile in year three, I think you're like 40 % likely to be in the top quartile.

49:23and then it gets better and better over time. But there's very little persistence. In fact, when the GP is raising their fund, their new fund, they're usually banking on a time when persistence is the lowest in returns, persistence of returns. And we look at it by cohorts. We look at it like yearly cohorts as well. So we don't just look at the funds. We look at how they persisted over time based on cohorts of investments. Yeah, we should publish more about this. But yeah, persistence is very low. And the question is why? You were going to ask probably why. We don't know. We can probably see a couple of things.

49:58One is strategy drift, which I think is like, as they get larger, it's harder to maintain quality, adverse selection, things like that. I think that's sort of one major thing. I think the other thing is that markets change. The nature of complexity is what we spoke about is like the actual structure of the market. And then opportunities change. Networks get stale. Opportunities go elsewhere. and capital rotates elsewhere, et cetera, et cetera. So it just means that your next set of investments may not be as sort of important as the first set, for example. So that could be it. And yeah, and there's other kind of dynamics that happen, I think, in the markets that just change how different courts and investments perform.

50:45Turner Novak:And you mentioned there talking about networks and the importance. So what is more persistent, returns or networks? Because it sounds like they can both change. We always say networks are more persistent than performance, right? So that's your answer, I guess. Yeah, networks are persistent. The quality and the importance and economic viability of networks might change, but the networks are persistent. But yeah, so we actually believe networks are more persistent. So you may have like an incredible network of early WeWork executives and employees. And like that was maybe a good network in 2019 or something.

51:27Turner Novak:And maybe today it's less of an exciting network. Yeah, exactly. And our models, which predict like we try to predict top-desk performance based on these network features, we know that our models are much more persistent, which we can publish as well. So our models are much more persistent. So even in the year, you know, like, for example, I think in year three, if you're top decile with our models, you know, I think they're like 75 % likely to stay, you know, sort of the top quartile. So there's definitely something to be said about networks as a feature for understanding persistence and performance.

52:02Turner Novak:What are some of the most interesting networks right now that you're seeing? When you talk about top decile networks, are there like things that stand out when you look at it today just based on your data? Like the actual qualities of those networks? Yeah. What are you specifically looking at and what shows up in the data as being a high-quality network? Yeah. The problem with our models is since they're deep learning, it's hard to understand, explain them completely. I think there's motifs of things that work. I think there's some networks that are really tight and they do well. Some networks are just people that are just investing in many things.

52:33There are different kinds of networks that we've seen. You know, from a thematic perspective, you know, in our first fund, we thought a lot about, you know, like physical world, for example. We thought a lot about, you know, where would the next kind of wave of innovation be? And we felt that, you know, the application of, you know, AI as well as like hardware and hardware software enabled services to, you know, sort of the value chain in the physical world and defense would be important. We did a lot there. which I think was prudent. I guess it did play out that way. So a lot of capital did rotate there.

53:13We also felt, for example, now we're looking at networks within computational life sciences which has been sort of depressed a bit in the public markets.

53:25Turner Novak:Yeah, some of them trade for less than cash, right? Yeah. Which is interesting. It's like there's so much innovation taking place and hopefully it's going to be a better regulatory environment. and so much more data upstream, especially around discovery. And there's all kinds of new kind of modalities, therapeutic modalities. And there's also gene editing and sort of gene synthesis and all these areas. Like the expectation is that, you know, there's going to be some unique things that happen in that space. And so, you know, it's kind of like non-contrarian, I guess, to use that word. But so we're looking at those networks and then everything around like data infrastructure, we always like, like whether it's like the way you develop and deploy software, you know, especially with AI and sort of security sort of as a kind of a primitive is always evolving.

54:21And I think that's like an evergreen kind of area for us. And then we love like these days we're looking at some consumer networks as well. uh we you know like we're getting to a point where we can maybe have like 100x better experiences and new modalities in consumer prosumer as well um and like application level which we didn't do like in the last one but we don't know no one knows anything yeah yeah it's kind of interesting

54:46Turner Novak:on the consumer side i go back and forth on it having done having done some consumer like i try to actually avoid investing in consumer because it's just a tough category like i feel like with like a like like like a b2b company you can just kind of like make a spreadsheet and you can kind of be like this will probably work or whatever with consumer it's just like binary like it can seem like it's working and just people change their behavior yeah you have to have a good investors like you know you mentioned sasha i think sasha is really just made very good instincts really good and sasha and casper like really good instincts on consumers and what they will like and what's durable and they have a good sense for like, yeah.

55:27So you have to have someone who really good instincts.

55:29Turner Novak:You taste as the kids call it nowadays. Yeah. A hundred percent taste. Yeah. It's hard to find really good. Yeah. There's a few, I think there's a few funds that are, that are good in that space, but not that many. What would you say is maybe it looks like a good network on paper, but probably turns out it isn't based on the data or structure of a network or something. You know, maybe some of the ones around YC could be very false positive. You know, like, there's a lot of, you know, like demo day kind of networks that, you know, or like you have all these investments that you made, kind of random kind of sparse networks, things like that, I think.

56:07You know, unpacking YC is very hard. It's kind of a hard problem.

56:12Turner Novak:You'd probably rather just own the YC, like you'd rather just invest in the YC fund probably versus like a... Yeah, I mean, there are funds that are kind of indexing against YC. Some have grown. you know there's there's like soma and there's others and but like there's and there's others that are around it you know that have maybe made a couple bets demo day and been bright and whatnot but those are not sustainable you know i think networks and they're usually paying like high caps and things like that so i i think you know they in a way that's a hard network to kind of underwrite generally do you have a bar when somebody says like high or low cap?

56:47Turner Novak:Is there like a point in your mind where you're like, that's low, that's high? And if I'm telling you, oh, I do pre-seed or seed and I say this number, you're like, eh, that's probably not pre-seed or seed. We like to see high look-through ownership relative to the fund size. We think that drives the most volatility. Other strategies can work. People can argue that the outcomes are getting larger. So what was once a billion our outcome is now 10, 20, 30 billion. But those are still like 0.01 % of the outcomes. And so I think the expectations, I forget who did this thing, it was Josh Kopelman, but there's the venture arrogance score or whatever.

57:30Turner Novak:Yeah. It's like the exits you need is the percentage of total exits in there. Of the market, right? So which I think is like, yeah, we have seen that. Like people with very high venture arrogance scores, you know it's a rare event it's a rare event we prefer there's something about pricing power that's unique as long as it's not adverse of course it can't be adverse selection but there's something unique about being able to get in early and have a good price and I think it says a lot about the sources, it says a lot about the channels of alpha you know one of the ways in which you can really sort of manifest alpha because i could in theory invest in all of the hottest seed rounds but under the hood it's like an uncapped safe after the round or it's like 2x over price of the last round's price and like that's not really price and power necessarily because no it's not it's also not going to return your it's not going to like you could have nice logos but it won't do much for your fund unless it's like a tiny fund like a$1 million fund or$2 million funds.

58:38So it just won't do much. It won't do much at all for you.

58:41Turner Novak:So on your guys' software side, I know we were talking about doing a demo. Yeah, it'd be fun to just kind of walk through it. I don't know. I don't know. We can kind of think about it as like you're demoing it for a smart person who maybe has never seen it before. Maybe that can be the assumed audience for this. Yeah, sounds good. I can give you, maybe I can show you a few things. And just to maybe highlight a couple of things, like we have a set of a few dashboards that we use. We also do a lot of tracking of like our own companies, our portfolios and whatnot. And just maybe to level set, you know, like our data, where it comes from, we have a lot of data that comes from a lot of different sources.

59:18And it includes not only like private market data, which is like transactional data, but also, you know, we track a few hundred million people profiles.

59:27Turner Novak:A hundred million people? 400 million. It's around 20, 25 million that kind of interact with venture in some way, shape or form. Oh, wow. That's a lot of people. It's a lot of people. I can tell you more about that. I can tell you where we have the sources from. So is this like an employee at Apple is interacting in some way? Yeah. For example, looking at LinkedIn profiles and things like that, like understanding work experiences, flows of people, where they're coming, where they're going from. We do a lot of that. We also track other what we call communities. We look at, for example, all of GitHub and PyPy and also scientific journals, a lot of crypto market data, public market data, and things like that.

1:00:14We've also been doing more like Discord. We're trying to do some Slack now, trying to understand the community. So where people hang out and what they're saying, I think, is important to us. but but generally speaking you know this is kind of our one of our investor dashboards and what we use to kind of diligence managers there's a lot here maybe i'll just show you a couple things

1:00:35Turner Novak:yeah show me the most interesting stuff like you used to make a decision i mean some of the most interesting stuff is like you know we we for every single transaction we have a lot of different data points that we've curated over time so everything from like how early the you know we have an earlyness factor, how early they were in the company, and we can estimate returns, the quality of the actual syndicate, whether the company's maybe, is it likely to graduate or not graduate based on various population statistics. We do a lot on just understanding different updates. This is all private data for us, but different updates that we're getting from the market on the company is to have multiple data points.

1:01:14And then we have a lot on talent. So this is the basis of understanding you know, the SOI of the manager. We also do, we classify every single company into different categorizations, like high level ones, like what we call verticals, but also like very, very low level, you know, categorizations across the market to see what's like changing and evolving to get a sense for managers exposed and where they have like created value. So we do that sort of stuff. So we do a lot within like trying to understand topics, topic areas and evolving topics.

1:01:46Turner Novak:So it looks like that one in particular So there's a lot of data center related and like machine learning. Yeah. It's one of our funds. It's a data infrastructure fund. So yeah, it's a lot. You're going to see a lot of that for sure. And then network is really more around, you know, focused on the actual co-investor network of the manager and how that's evolved over time and what it looks like. And is this like some off the shelf software you can buy on like the, on like a Google marketplace or something or like AWS marketplace? or is this like built it all in-house, custom data? Yeah, everything's in-house.

1:02:21Yeah, we built everything from scratch.

1:02:24Turner Novak:And it's on unstructured data too, I think. So you literally take LP updates and just forward it in. Well, some of it's structured. We obviously have licenses with a lot of data providers. So some of it's structured and some of it's unstructured. But this is just showing you what it looks like underneath the surface. Okay, so what am I looking at here? So you're looking at a relationship between one of our funds and Dresen. and the flow of information between them. And in this particular view, the arrows represent the strength of the relationship and the flow of it as well, and the frequency recency.

1:02:57There's a lot of different features that go into this, but this is just giving you a sense for the scale of how we do network reconstruction and how big these networks are. So if you were to open up Sequoia, you would see how big the network could be of the interrelationships between different entities. And this just forms the basis of some of our algorithmic work that we do on trying to understand the position of a manager in the network.

1:03:18Turner Novak:So there's a lot of arrows pointing into Sequoia. What would that tell me about Sequoia specifically? Yeah, so it could be arrows both ways, which means there's a lot of co-investing taking place between the entities or there could be some after investing and pre-investing and different ways in which the investing takes place. But they're obviously very central. And do you want to see that? You want to see certain configurations that are persistent is, I guess, the way I would answer that. And then when you're looking at adding a new manager, are you thinking about exposure to a network that doesn't exist specifically or a network that you don't have in the portfolio yet?

1:03:58No, we don't do that. We actually don't do that.

1:04:01Turner Novak:Oh, you don't? No, we've thought a lot about it. I mean, we like to spread out for sure, but it's unlikely. I mean, there's going to be overlap in the networks. I don't know, there's a few thousand deals done a year, seed deals or whatever, and there's going to be overlap in the networks. And we're okay with overlap. And in some cases, we have kind of a distributed network. But it's unlikely there's going to be a lot of overlap between networks. So you mostly just like good, high-quality network? Yeah. Yeah, the models are really focused on good, high-quality networks, close to the core, good single for the market, and good patterns of behavior.

1:04:38so you know good persistence and recency and things like that and that's how we kind of get a signal and then from there we do our work yeah makes sense and then i mean i think this this maybe

1:04:48Turner Novak:begs the question like why why launch a fund to fund around the software like shouldn't you just you have this data like you see who's the best investor just like follow on to them versus like have a fund fund like what was the general thing maybe this gets back into like the history of stuff a little bit but yeah yeah certainly so i mean we sort of felt insistent in some ways but But in reality, we felt that obviously building a VC is hard. And for us to try to compete in series A, B deals with no brand when we first started would be kind of impossible and it would be adverse selection. And we felt that the highest convexity part of the market was the emerging VC funds.

1:05:29And there are other fund-to-funds that are good. We felt like there was a way to do it very, very systematically. and where you can add value to the GPs in unique ways, especially through technology, be very tech-enabled and sort of tech, like have really interesting tech primitives. But yeah, so the idea was, so if you select really good managers and those managers select really good companies, then coming out of your portfolio should be really good companies.

1:05:55Turner Novak:Yeah, in theory, yeah. If that's the case and the managers want to continue to follow on on their winners and we're a capital source for them, then we should have access to a very asymmetric high alpha portfolio. And that was the idea. And in order to do that well, you have to have a good data lens on being able to track your look-through portfolio, track the market generally. And that was sort of the original idea. And being a fund of funds puts you in an interesting structural position because you see so much. and yeah and then there was like i thought there was a paucity of like good fund of funds or good lps smart lps in that market you know i just felt like there was very few and there's a lot of lps but there's just very few that are like really you know kind of intently focused on on sort of being tech extremely tech enabled so like the return profile of something like level do you think of it as like you're getting like kind of like diversified seed exposure but then there's also because you do some co-investing follow-on secondary type stuff later is it like series b exposure blended with like some diversified seed or pre-seed exposure like is that generally how you might pitch it to an lp generally or yeah we've evolved but i think yeah we think of it as pitching it as holistic so it's like a holistic uh it's a holistic way to get exposure to this part of the market like the frontier of technology innovation and the the buckets are essentially primary LP commitments, opportunistic LP secondary commitments, and that's in a similar vehicle, and then doubling down into co-investments.

1:07:38With co-investments, our approach is we try to empower the GP to put more money into their best companies and that we can be a partner for them there. and the way we like to do it is if we can sort of get capital in before a round happens, you know, it's kind of like the best opportunity for us. But in that particular bucket, we want to be, you know, we're trying to manage risk as well, so we don't want to be too early. So usually the entry point is as early as A, but typically B, we just did a C. So, you know, we like to kind of manage risk there. and just the combined platform should deliver the results.

1:08:22Turner Novak:When you talk about getting in before a round, so is this, how do you generally do that? Is it like some kind of like uncapped or like discounted safe or do you like buying secondary, primary from employees? I don't know, I'm just curious. I mean, this is going on the internet. Millions of people hear this, so I guess you're... I don't know. It's not that hard. You know, it depends. I think if we know that the round's coming together, and we like the company, we can definitely do it on cap safe. We're happy to do that. We've done that before. If there's a period of time between... Because we also don't want to anchor price.

1:09:02Just because we're not a lead investor, and we also want to be careful about the relationship and about their own financing process. So that's what we'll do.

1:09:11Turner Novak:So it's kind of a way to just signal, we'll participate in the round. And we're in, And we don't care who leads it. Like, consider this as part of the round when it comes together. But like, we're securing our ticket. And of course, like our data comes from our GPs. And we have a lot of asymmetric access and information about the companies themselves. And, you know, and we've developed, you know, mindset. Like, we've developed information and a history with the company ourselves internally. So that's how we think. And you probably kind of knew, like, for a year or two or three or four, you're like, oh, it's pretty interesting.

1:09:41Turner Novak:Like, we kind of want to follow this. We like it more and more over time. And so it's not like a shotgun, like someone sends you an email and like, you know, you're making the decision. No, no, no, no. Yeah. No, no, we had no, that's, yeah, it's much more. But that's, that's what we try to do. You know, it's kind of like, and it's not, it's never perfect. Like, you know, everything's like a moving target in this business, but that's what we try to do. Yeah. What do you think an ideal kind of LPGP relationship looks like then? In this case specifically, but then just more broadly, you kind of mentioned like not enough people are taking a data-driven approach.

1:10:12Turner Novak:What does kind of the ideal look like? what do you want from an LP, I guess, is where you'd want to start. I think you'd want to have strategic guidance with an understanding of the market and how it's evolving, right? Because I think the benefit of having an LP that is investing in the market is that they understand the market and they maybe have perspectives that you don't have as sort of a single instance in a big market, right? So I think that's one thing and that can be informed by lots of data. So that's one thing. The second thing is essentially helping with capital formation. If you have conviction on a GP, helping them with being a signal for the market so that if there's a fundraising, it takes place pretty quickly.

1:10:56The third is to the extent that you can enable them to have more information than they would have had on their own. So whether it's like having some of our tools that we use for like network search, you know, for maybe for sourcing or for sourcing support or for portfolio support, hiring, things like that. Like if we can enable them with our data so that they can do things that they couldn't do before. That I think is a really amazing relationship to have with a GP over a long period of time. So those are some of the ways we think about it.

1:11:32Turner Novak:Do you think there's things that maybe other fund funds maybe get wrong with our approach? I think you do have to look at, I do think a data approach, however you want to do it, I think it's really important to have a data centric view on things so that you're not making decisions based on presentations and maybe some ad hoc calls to your network or whatever. So I do think you need to have a data centric view. And you just think people just don't do that enough on the LP side? Yeah, a lot of people don't do that. They do a lot of work on the actual GPs, but they don't look at the market. They don't look at the holistically, the market that the GP is operating in and how that's changing and evolving and what those networks look like, and et cetera, et cetera.

1:12:25So I think there's a blind spot, I think, for many there. And it doesn't mean that you don't make good decisions. It just means that you're missing information that may be important or relevant. So I think that's sort of one thing.

1:12:39Turner Novak:Yeah. Speaking of using data, what do you think about benchmarks? I see a lot of people... I mean, it's a big thing. People say, oh, we're a top decile or quartile in whatever certain benchmark or category. Do you think are they important? Are there things to be aware of or poke around with benchmarks? benchmarks? Yeah, benchmarks, it's all, I mean, by definition, it's relative. And there's a lot of ways to game it. And, you know, I'm not sure how correlated DVPI is with DPI. And, you know, there's, there's, yeah, yeah, we don't know. They're somewhat correlated, I imagine. And so, yeah, there's just, it's, it's just, it's, it's, you know, it's a rough measure.

1:13:20It's a coarse measure for something that you're trying to understand better. And it is what it is. At the end of the day, what matters, of course, is like cash on cash returns. over a period of time that makes sense. So from a cash flow perspective, which you can try to ascertain as a proxy by benchmarks, but it's just one of several potential course measures.

1:13:42Turner Novak:Yeah, I think it's kind of interesting. You might outperform a private market benchmark or whatever, but then pretty much every venture manager probably underperformed NVIDIA, like the biggest, most liquid asset in the world too. So it's like, you know, I don't know. It's all interesting how much you will obsess over this stuff. Yeah, yeah. Although, yeah, you have to like, if you're going to do a public markets equivalent, you have to definitely look at the cash flows. You can't just say the cash on cash 10 years. It doesn't work that way. You have to look at the cash flows and see like, what does it look like over time?

1:14:17And yeah, I'm sure you're going to find that most funds are going to be worse than NVIDIA for sure. Like almost all of them. But you're going to have funds that could look like 20%, 25%, 30 % IRR, etc.

1:14:32Turner Novak:But would you have predicted that it was NVIDIA that was going to be the one that did this? I think at the time, we were coming off the crypto Web3 bubble where they were just being used to mine Bitcoin and there's no other use case. And LLMs basically weren't invented yet, for lack of a better way of describing it. The information was there. I mean, I didn't act on this, of course, but I remember meeting people really early on when I had my last startup and we were doing machine learning and meeting some researchers that were starting to deploy models on GPUs. And we're like, what's a GPU? So I remember the information was there.

1:15:11It was starting to gather. but the market could never have foreseen like you know the impact of of that and that invaded would win that and etc etc so yeah or the similarity between like matrix computation in video games and sort of large-scale matrix computations in in sort of deep learning models etc like i don't know that's how you but that's what investing's about i guess yeah well so

1:15:39Turner Novak:So speaking about AI, have you seen anyone interesting, maybe outside a level, anyone using AI and data internally in an interesting way? Any funds? This may be like a selfish thing. So I'm trying to, besides just like chat GPT to summarize things and, you know, make some things faster. Yeah. I mean, I've seen some of the, there's a Quantum Light guys that the founder of, oh yeah, founder of Revolut. He has a fund in Europe that they're building. Yeah. Yeah. I think it's called Quantum Light. Yeah. We saw some of their work there. It's good stuff. They basically track the market, trying to figure out where to invest from a series B perspective.

1:16:13Similar to our infrastructure. I thought that was very good. Very good work. Yeah, to be honest with you, we haven't seen so much out there. On the LP side, almost nothing. There's almost nothing. People have reporting interfaces where they report on their portfolio, but there's very little general market work. Yeah, it's surprising that for such a tech-centric ecosystem that there's very little tech out there. But it's like a lot of them just don't need it. Some of them have automation tools and things like that, but they're just using their networks and sourcing. It's not something that's front and center.

1:16:53Maybe more and more with LLMs and the ability to do more analytical work and research work, I'm sure a lot of people are using those tools. we use it ourselves we have a lot of tools that we that we use and we integrate with our data sets for getting up to speed quickly on a market those kinds of things i'm sure there's a lot of that

1:17:09Turner Novak:happening you know i don't know if you know samir kaji at allocate they kind of do some so i think i described it as like it's like software for private market investors family offices giving them access to certain stuff and tools for their portfolio i know they made kind of AI investment memo underwriting tool. Like I put in my DAC and it spit out a bunch of information and like gave me a grading and a score. It was kind of interesting. But again, it's just kind of like the, you know, investment memo summarizing use case. So what have you seen just generally over the next decade? How do you think venture is going to change as an asset class?

1:17:46Turner Novak:More data? Maybe that's a big one, hopefully. And I think there'll be more data out there for sure. Maybe people using data? Hopefully. Yeah, definitely more exhaustive data, more and more pockets of data that you can leverage if you're able to combine it together to get a better view of things. Yeah, the later stage market, I think, is becoming much more institutionalized for sure. I think that'll continue. Maybe we'll see one of these firms go public. It become sort of established as a public asset class. You're saying like an investment firm? Yeah, like an Andreessen or something like that.

1:18:27Yeah, I can see that happening. Like a Blackstone kind of. Yeah, we can see that. The secondary markets will evolve. I think, I feel like that's like, there's not really a good market making approach there. Like, you know, matching buyers and sellers was on the LP side as well as like direct side, et cetera. I feel like, and people are going to want to be able to be flexible on when they exit oppositions and those kinds of things. So I think that market will evolve, especially as venture just becomes like, you know, the timelines are so much longer and companies will stay more private longer because there's a lot of benefits to staying private, especially if you can get liquidity, you don't even need to go public, right?

1:19:08Turner Novak:Yeah. Yeah, Databricks just did their Series J or K. Yeah, Series K. You're going to see Series Z one day. Yeah. Yeah, exactly. So if you can sell, there's a way to exit positions and sell and, you know, investors can sell. There's really no reason to. and so I think that will happen. Well, plus two, if you're an investor and you have a public market fund, I mean, you probably charge 1 % management fees roughly and like 10 to 15 % carry versus private markets, you can charge 2 in 20 and sometimes more than that, honestly, especially if you're an open AI. If you look at like the look through fully blended costs of some of those like triple layered SBDs, it's like, I don't even know the management fees they're charging and the carry they're charging.

1:19:50Turner Novak:Yeah, it's crazy. It's just so much more profitable as an asset management firm to be doing private markets. So the incentives are there to just continue doing it. Not even respect about the company side, but on the investor side, the capital is there. For sure. So those are some of the things that we think will happen. Cool. Well, this is a lot of fun. Thanks for coming on the show. Yeah, it was a lot of fun. And I hope that you had fun. And a very fun thank you to Ramp for supporting this episode. Head to ramp.com slash ThePeel for$250 on your first set of cards. If you missed it, make sure to check out last week's episode with Will O 'Brien at Ulysses where they're building low-cost autonomous robots for the$2.5 trillion of economic activity in the world's oceans.

1:20:31Turner Novak:If you like this conversation, please like, comment, subscribe, and name your next form of alpha after me. If you don't want to miss a future episode, subscribe to my newsletter, The Split, linked in the description to get each episode plus a transcript emailed directly to your inbox every week. Thanks, Dan, for listening. See you next time. you