The Exchanges

Q What did the summary of that analysis look like? Like, what are the three or four key metrics that you were calculating?

A I'd say that the analysis was probably split into three separate sections. The first and most important was really what we called the individual manager analysis, so it was a five pager on every single manager that looked at Independent calculation of the manager's performance using public regulatory files. A lot of assumptions in place, but that was very valuable for a lot of our clients, just as an independent check on what the manager was reporting. We looked at the batting average, the win-loss ratio, the alpha generation of a manager. We looked at key statistics like liquidity, their market cap focus, so things like median market cap, number of positions, concentration, you know, top five, top 10, et cetera. And then we do a deep Dive and break it down by sector, by market cap, by geography, and so forth. So that was the first main block. The second block was an overlap analysis. So we would take all the manager positions, and we would overlay them on top of each other's, and we would calculate the percentage of each manager's portfolio that was identical to other managers. And the results were incredible, because sometimes you'd find managers where 50, 60% of the portfolio was the same as an allocator I think one of the key things that every allocator tries to do is find uncorrelated streams of alpha. And if your managers were overlapping with each other, well, they weren…

AI assessment note: “We looked at the batting average, the win-loss ratio, the alpha generation”

Q Does that slippage come from what I might expect performance chasing?

A Exactly. See, I think that it typically comes from, I think, a conflict between wanting to maximize your return, and so having as much exposure as possible at any given time, and protecting the firm. Effectively, the sensitivity is not just a function of your fund's performance, a function of your firm's performance to short-term shocks. And so when you go back and you look at periods like August, September, 2011, March, April, 2013, or even 2014, you look at October, 2015 through February, 2016, what you find is, is managers go into these periods of time sort of at full capacity on exposures. And so really the only thing they can do is reduce it, when they really should be increasing it. And so it tends to be, as you stated, a reactive behavior and chasing, and as a, as a function of that, it tends to be detractive.

AI assessment note: “Exactly. See, I think that it typically comes from, I think, a conflict”

Q And how, so how would that play out?

A I think the challenge always is when a large portion of your thesis as an allocator, and I'm not saying this is the case for every allocator, but it is for some, when a large portion of your thesis is a function of what the manager's historical returns are. Well, by definition, if the returns turn bad, Then your thesis turns bad. And so what ends up happening is you end up redeeming often at the worst possible time, or you end up contributing at the worst possible time. To give you a statistic, rolling three-year performance periods for hedge funds, correlation between different rolling three-year periods is about -.15. In other words, there's actually a little bit of mean reversion in there. By being able to introduce a new framework that helps allocators understand the underlying skill sets of managers, I think first and foremost, they can be more long-term oriented, And understanding those are long-term skill sets, and being able to attribute performance to different things. So for example, if I'm an allocator, and now I understand that actually security selection, position sizing, and trading is still strong, but the drawdown is largely a function of a bad market timing decision, or a bad Allocation decision. I might double down, but also what we didn't really talk about is the fundamentals that support those skill sets. For example, let's just say you and I ran analysis on…

AI assessment note: “By being able to introduce a new framework that helps allocators understand the underlying skill sets”

Q So, and turn a little bit to conviction. The way you described assessing skill as requiring a large number of data points, or big N, Seems like it'd be the opposite in a high conviction portfolio. So you think about highly concentrated managers or activists where there are big position sizes, but probably not that many of them over time. How do you use the conviction metric?

A It's actually a great question. What I will say is that the number of observations is really dependent on your number of positions at any given time, and it's also dependent on your turnover. So high conviction doesn't necessarily mean low turnover. You could have a manager 20, 30 names at any given time, but turning those over a couple times a year. You could also have a situation where a manager has 20 positions for 10 years. And so what I would say is it's much, much harder to assess skill on a lower number of observations. By definition, if my holding period was 10 years, and I had 20 names, it's gonna be really tough to analyze skill. You're just not going to get the data. I would also say that that's rare in our industry because of the, probably because of the institutional imperative that has evolved over the last 1520 years, and that there's very few managers that can withstand running that structure. And so, more often than not, when you're assessing skill on the long side and then separately on the short side, you know, I'd say if you have five years worth of data on a manager, you've got enough observations.

AI assessment note: “number of observations is really dependent on your number of positions... and... your turnover.”

Q Before we turn to the closing questions, I just want to ask, what's the most creative application that you've seen from one of your clients from using the data?

A I don't want to get too proprietary. I think a lot of our clients are very sensitive about that sort of thing. I'll give an example that I think is relatively benign, but still very powerful in the sense that one of the things that I've seen more and more managers do is have a thoughtful position sizing framework that is sort of a multi-factor position sizing framework. For example, factoring in crowdedness as an actual score in how you size a position. And so one client that has a, maybe a five factor model, crowdedness is one of them. And they found that incredibly successful. So they still end up owning crowded stocks, but man, it's the bar high to get to a seven or 10% position on a crowded name. And I think that's had a tremendous impact since they implemented that on basically minimizing their max drawdown. So it's been, it's been really very special.

AI assessment note: “factoring in crowdedness as an actual score in how you size a position”

Q What did you see in that initial data research analysis?

A What I saw is that I had an edge in being able to objectively measure manager skill and fundamentals for the purpose of doing what ultimately every single investor does, which is forecasting the future. Ultimately, if you make an investment in a manager, you're forecasting how they will perform in the future. You're not investing in their historical Track record. Investing in their future track record. And so I felt like by having access to all this data and understanding the skill sets of managers, so for example, wasn't quite as sophisticated as this, but we started to look at things like batting averages, not win-loss ratios, not yet, that came later. We looked at alpha generation by sector, by market cap, things like that, and we looked at fundamentals, and I felt like the fundamentals were really important. It gave me the confidence to be able to sit down in a meeting with folks that were much more experienced than I was. Having them effectively give a hypothesis or an opinion based on their most recent meeting with the manager and saying XYZ manager is feeling really good right now. I could tell that they've got fire in their eyes. I could tell that they said they're seeing more value than ever. We asked them about their liquidity and the manager said liquidity is incredible. We asked the manager about the opportunity set In large caps, even though they've invested in sma…

AI assessment note: “What I saw is that I had an edge in being able to objectively measure”

Q I'm right. I'm on the edge of my seat. What are those three? And then walk through each one.

A Well, the first one is sort of the bread and butter. There's idea selection, stock picking, security selection, whatever you want to call it. And it depends how you measure it. But I think If you're being objective about it, a manager picks a security. It's like picking a marble from a marble jar. Now you can define the marble jar any way you want, but that's a universe of similar marbles. I think a good approach is to, you know, if you picked a large cap tech stock, well, you picked it from a marble jar with large cap tech stocks. You can apply a geographic frame to it. You can apply a market capitalization frame to it, but irrespective of what frame you choose, at the end of the day, you are selecting something From some sort of universe. And so we can measure that. And we found an incredible amount of persistence on idea selection. The fourth is position sizing. So it's, once you select that marble, how big do you make it within your portfolio? The perfect example I always like to give is, is we could have a group of 20 people, uh, start on January first, and we could say, here's a million bucks to each. You're forced to own the Dow Jones Industrial Average, all 30 stocks. But you have to own all 30 at every single point in time. But you can size them any way you want. Well, you could have a huge dispersion in performance. And sizing is typically fully in your control. The o…

AI assessment note: “Well, the first one is sort of the bread and butter. There's idea selection”

Q because like performance, any of these metrics, stock selection, skill, or capital allocation, it's all going to come from looking at past data. And I think what you're suggesting is if you can map How people generated the returns in the past with their environment. So those fundamentals, what was the liquidity profile that allowed them to elicit that return? That you start to draw some conclusions about the future.

A Persistence important concept. I have a general philosophy, which is you sort of have to sometimes measure people based on the game that they're playing. So intentions matter. And so for example, if a manager's intention is to pick stocks and make most of their money picking stocks, well, they should be measured on stock picking skill, not on market timing skill. But in a similar vein, we often measure managers based on time. And so for example, monthly returns is the really the key metric in our industry. Take January's return as an example for a manager. What was that a function of? Was that a function of any conscious intention for the manager to generate a return in January? Not really. It was a function of ideas they came up with two years ago, ideas they came up six months ago, and ideas they came up with a month ago. It was a function of what the market did. It was a function of what individual sectors did. And so when I think about persistence, I think about it in terms of N, not in terms of T. So I think about number of observations as being a more important predictor of what will happen in the future than whether or not there's consistency in T. And I think that's a really important foundational frame. And so, for example, when I look at idea selection, I won't name any manager names, but I'll sort of give high-level examples that are descriptive. There is a manager t…

AI assessment note: “I think about number of observations as being a more important predictor”

Q And what other markets do you see the Novus analytics applicable to?

A Yeah, the two big initiatives right now are really multi asset class and long only. The reason why it's What's easier to do for equity managers is because pricing data is much more readily available. So if, for example, if you, to calculate trading, or even selection, you need really good pricing data. Now what we've been doing our last Three to five years is really expanding our multi-asset class coverage. So most recently, I'd say the biggest investment that we're making at Novus right now is in fixed income data, which is extremely expensive, extremely complex, but will allow us to do this effectively the same exact framework on non-equity managers. And then Long Only is a big industry. I think they're struggling as well from an active management perspective, and so there's a big need within Long Only to look within, and definitely something that we've started ramping up The client base on. So last year, say we got a few clients. This year, we've gotten a few clients, and we anticipate that will be a big percentage of revenues in the coming years.

AI assessment note: “the two big initiatives right now are really multi asset class and long only.”

Q What life lesson have you learned that you wish you knew a lot earlier in your life?

A One of the sort of recurring themes in my life is failure and rejection. I think the first time probably was that experienced boarding school, getting rejected by all those boarding schools when some of my friends had been accepted, when my parents expected that I would be accepted. You know, sitting down in my dad's office as a fourteen-year-old kid with 10 offer letters, sort of, that just came in from the mail, all of them thin, obviously, and opening those up in my dad's office, that was such an important moment In, in my development, and then I think also similarly at Novus, I think when you look over a 10 year period, it's been a great success, but there have been periods where I, I felt like I was really failing. There have been periods where it's been very difficult. I think I've made some really big mistakes, and one of those mistakes was sort of the balance between talent recruitment, scaling that up, and talent management, and making sure that you had the talent management systems in place, In order to be able to develop these people and integrate them, onboard them in the right way. I'm glad, genuinely, that we made these mistakes in the sense that that's how you learn. Do I wish I'd known them earlier? Sure.

AI assessment note: “I'm glad, genuinely, that we made these mistakes in the sense that that's how you learn.”

Q What information do you read that you get a lot out of that you think others may not know about?

A Well, I definitely read a lot about psychology, and I think most people in our business at this point have read a lot about psychology, but I think that it's such an important area. And what's interesting is more recently I've really turned to physiology. And I've spent a lot of time in physiology. I've been on a two, three year journey just to understand my own physiology, and to understand That ultimately, your physiology is what leads to feelings. Feelings lead to emotions. Emotions lead to thoughts. Thoughts lead to behaviors, and behaviors are what lead to results. And so when you think about looking at some of the great athletes that are out there, you see that they're really able to be cool, calm, and confident under pressure, and that has a tremendous amount of impact. And running a company, that becomes really important, because sort of your first day on the job, it's like you print business cards, and you're like, Like, oh my goodness, like huge, everybody, high fives all around. And so your first days are really filled with a lot of good news, because anything is good news. And you, when you become a bigger organization, now a hundred people, as a CEO, every day you get good news and bad news. And being able not to get overly excited or overly depressed about it, I think is important.

AI assessment note: “more recently I've really turned to physiology. And I've spent a lot of time in physiology.”

Q And what are the core components of measuring the skill of a manager using this data?

A Well, this ended up developing over a much longer period of time. We've developed a framework called Novus Framework, and it's a particularly simple framework, which basically says that managers have five degrees of freedom to generate alpha. The first one we call exposure management. So it's moving around gross and net. Very simple. Now, what's interesting is we found over time by now having access to private data, and we have about 1500 hedge fund managers that now give us their position Level transparency directly from administrators, and so we feel like that's a really valuable data set, and what I'm about to say is based on that data set. Exposure management, on average, that tracks approximately 200 basis points a year from a manager's performance.

AI assessment note: “We've developed a framework called Novus Framework... managers have five degrees of freedom”