Misha Malyshev - The Quest for Alpha

00:00:08 [Misha Malyshev]
One can say, well, if market is getting more efficient, is the pool of alpha getting smaller and smaller every year? Well, and the answer is the pool of known alpha is getting smaller and smaller every year, except those pools kind of multiply. As the world grows, as beta grows, the alpha that follows that beta also grows. So number of inefficiencies or places for inefficiency increase.
00:00:39 [Vincent Weber]
Hello everyone, I'm Vincent Weber and this is Resonanz Spotlight, the podcast where we explore investment strategies, learn the stories behind them and meet the experts who create them. In today's episode, we explore the concept of alpha, a key driver behind daily investment decisions involving trillions of dollars. Alpha's importance in the investment world cannot be understated, yet it presents unique challenges. We'll discuss its practical definition, where to find it, and why its scarcity creates difficulties for both small and very large investors. Small investors may struggle to access alpha, whereas for very large investors, it might not significantly impact their portfolios. Join us as we dive into these topics with Dr. Mikhail Malyshev, widely known as Misha.
Misha is the CEO and CIO of Teza, a niche multi-strategy platform known for its high-frequency trading infrastructure and its team of over 80 PhD-level professionals across five offices. Before founding TESA in 2009, Misha served as a global head of high-frequency trading at Citadel, a position he took up in 2004 after roles as a management consultant at McKinsey and as a researcher at Bell Labs. Misha earned his PhD in astrophysics from Princeton University and holds degrees in theoretical physics and mathematics from the Moscow Institute of Physics and Technology. Misha, welcome to the show and good to have you here.
00:02:13 [Misha Malyshev]
Thank you, Vincent. Glad to be here.
00:02:15 [Vincent Weber]
Before we deep dive into our main topic, would you mind giving us a brief introduction of TESA and your current role there?
00:02:22 [Misha Malyshev]
Sure. I think you've done it perfectly. I'm CEO and CIO of TESA. which focuses on very unique and niche strategies with the value proposition to investors as a very orthogonal set of alphas to add to their portfolio.
00:02:45 [Vincent Weber]
Now I find you have a very fascinating background. I mean, it's not unusual to find Science P&G doing investment, but you managed to go from a research lab to management consulting, high-frequency trading, and investment strategies. So how could this all happen?
00:03:01 [Misha Malyshev]
yes that's true and i like to say that half of my adult life i've been a scientist in moscow fistic and then princeton then bell labs and and then i turn quote unquote turn to the dark side and went to mckinsey It was an interesting experience because in science, you deal with logic and facts and strict definitions. And in management consulting, you deal with emotions and humans. And that was a very unique experience kind of after doing physics for so many years, diving into the human psychology.
00:03:51 [Vincent Weber]
So what were the biggest change or technology shift that you witnessed in markets, if you have to compare your early days to today?
00:03:59 [Misha Malyshev]
Interestingly, we are participants in the markets from a very unique perspective. We are scientists and technologists who are playing investment game or participating in investment activities. We'll look at the markets through the prism of science and technology. So when you ask what are the biggest changes that happened to the market over our lifetime, over the lifetime of our company, I would say those are definitely technological changes. When we look at the markets, and it's actually tied to your initial question about alpha and beta, the market efficiency that is sort of relentless evolution, right? Like market is getting more and more efficient. And one way market getting more and more efficient through actually development of technology. So technology delivers. higher information dissemination, higher information gathering, higher information processing, higher information accumulation.
So when we talk about higher information dissemination, we're talking about like ultra high frequency world. And since we've started TESA, the frontier of high frequency changed from being a microsecond time reaction. Microsecond is obviously one millionth of a second to a hundred of nanoseconds to currently 10 nanoseconds. And nanosecond is one trillionth of a second. So we're talking about times where computers, your normal computers are not fast enough. So we're talking about special applications called FPGAs. Those are like tiny silicon chips that allow you to react to something happening on the market, not in a blink of a second, but in a blink, not in a blink of a millisecond, a microsecond. It's a few nanoseconds. So that definitely happened, right? Like the speed of.
of reaction time to what's happening to the market by variety of different participants increased the speed of data delivery has increased we're talking about since we started the company in 2009 the speed of data delivery between two major trading point between chicago and new york between aurora illinois and exchanges in new jersey increased from several, from give or take 15 to 20 milliseconds to initially to just under 5 milliseconds and then later on even further.
00:07:10 [Vincent Weber]
Thanks, Misha. So, my knowledge of high-frequency trading or market making is really limited to the movie Trading Place. that 18th movie with Eddie Murphy. So, when you talk about speed getting faster, Is it just a replacement of before the ability to shout louder or physically push your peers just to get in front of the order flow?
00:07:35 [Misha Malyshev]
That's a perfect analogy. I know old bond traders used to be all six feet tall and huge and former football players for the reason of being able to push themselves closer to the pit. yes currently currently working on nanoseconds is is an analogy of of shouting louder or or or making making this gestures with your hand by by cell much faster than than the person nearby would do so it's definitely it's definitely a technological way of making market more efficient So rather than humans doing it, right now it's done through ways of technology.
00:08:22 [Vincent Weber]
Okay, right. But now if you're not a market maker, you're an asset manager, how does this affect you? Is it something you can benefit from?
00:08:31 [Misha Malyshev]
For sure. If you think about various horizons of investment activities, like starting with the very, very large horizon, which is buying gold. So buy and hold, that's where beta leaves, right? Like all the way to high frequency where the smallest of the alpha sleeve. So each sleeve of that time horizon provides opportunity for another sleeve to be in more efficient market. So competition and high frequency allow fund managers to trade with better liquidity or at more precise prices. Without market making or high frequency activity, people who trade on the daily horizon may have not have gotten the right trade or the right price, right?
So the fact that market makers and high frequency traders remove inefficiency in price discovery by making sure at any point in time price is as close to fair as possible, actually providing the service. to an intraday traders. And intraday traders who take positions maybe like overnight, they provide the service to longer horizon traders. And that's how the whole chain works. Of course, you have to pay for that. And on the other side, when you provide the service, you demand payment. Asset managers do pay to high frequency traders and market makers for liquidity that they seek.
00:10:14 [Vincent Weber]
So you mentioned initially the more intense competition or efficiency at the very fast level. But what I've also noticed is that at the medium term level, with the proliferation of tools like Python Pandas and some open source statistical analysis library, the barrier to entry for doing things like classical medium term trend following has been dramatically lowered over the last decade. So where do you see your competitive advantage in this spectrum ranging from the very super fast training strategies to the medium term 200 day type of moving average?
00:10:52 [Misha Malyshev]
Yeah, actually, it's a good question. And the way we do it and the way we always have done it, our competitive advantage always come from the intersection of different areas, from intersection of different horizons or intersection of different activities. So we are not trend follower. right we are not doing simple cta like approach and and and we believe right like at this point it's a commodity right like you said anybody anybody with the python notebook any intense grader with the python notebook can create a cta like strategy so so the question is in in and and therefore those strategies are not not high quality strategies they don't carry high sharp ratio behind them.
So what we do, we try to get units of information from one area to transfer this information to another area by focusing on microstructure and by focusing on other people's market impact. when people trade they create market impact it's super important concept in the market we like to say that when when people say there are only two things death and taxes we like to say well there are only three things death taxes and market impact and market impact is the one that you cannot cheat so if you have to buy something you have to pay up So when you decide to buy extra amount of gold or oil or particular stock, you are going to push the market higher. And when you decide to sell it, you are going to push the market lower.
Understanding that market impact and its dynamics on the very, very granular level allows us to take advantage of inefficiencies which people create by creating that market impact.
00:12:53 [Vincent Weber]
So coming back to our main topic. we wanted to discuss about alpha. And the first thing about alpha is, everybody has its own practical definition. So as a starting point, I'd be curious to hear about your take on this.
00:13:11 [Misha Malyshev]
Yeah, I think when we talk about alpha, probably it's worth first defining beta because that requires probably less controversy, right? So by beta, we essentially define strategy that is buy and hold. So, for example, there is beta of U.S. market. You can buy S&P 500 in one form or another and hold it for a long period. Obviously, beta is the biggest investment game in town because that's the largest opportunity in terms of capacity to make money. And alpha is. trading around inefficiencies.
So rather than buying every single stock in S&P 500, if one can figure out that one stock is going to perform better than the other stock, then the typical alpha game would be to buy a better stock and to sell a stock that is much worse, in which case it would be market neutral, so no trace of beta in this strategy. And it would be defined as alpha. The necessity of alpha typically to investors comes for two reasons. A, it can be completely orthogonal to beta, right? Or at least I would define alpha completely orthogonal to beta. As some funds would mix beta and alpha together, but most of the sophisticated investors would not go for it.
so when you're looking for alpha what market are you typically looking at as i said we are we are starting with microstructure we studying market impact of of of other people and knowledge of microstructure the data set that i described the uniqueness of the data set the the our ability to see through kind of our ability to have a particular microscope to look at this data allows us to identify patterns on any market that is governed by economic laws and there is only one economic law in the market is supply and demand it's that balance between supply and demand pressure that determines the change in price there is no there is no other you can say sentiment or fundamentals or or any kind of quantitative pattern but in the end of the game In the
end of the day, it's the balance between supply and demand that pushes price up or down. So if you can focus on this most fundamental concept of balance of supply and demand, and if you can say that the supply will exceed demand or demand will exceed supply, if you can focus on that, you can determine where the price is going. So therefore, your question is, which markets do we trade in? Anything. Then any market that allows you provision, allows you collection or identification information that you can get about the activity on that market. Any market is good. It can be U.S. equity, Chinese equity, Indian equity, which is traded through futures mostly, global commodities, gold, FX, anything. So as long as you have electronic.
00:16:44 [Vincent Weber]
limit order book to which we can connect and listen to we can get the information from this market and we can trade on that okay great to hear given your extensive experience in quant finance across various cycles could you share your insight on the major trends currently shaping the market and the competitive landscape the interesting interesting thing in in this whole game the high level high level view on this
00:17:13 [Misha Malyshev]
that there is not enough alpha in the world. And not enough alpha in the world leads us to the question of capacity. If you talk to any manager who has a particularly niche strategy or high Sharpe ratio strategy, you will probably very quickly discover that capacity of that strategy is limited. That strategy is what is called capacity constraint. You look right now at the most successful multi-strategy funds, most of them are closed, most successful are closed because they're operating at their capacity. So the main trade or the main trend in that space is that lack of capacity with increased demand for those strategies.
this increased demand for alpha and everybody everybody everybody looking for another multi-strat another successful multi-strat the reason being is when when you put when you put several different orthogonal strategies together right like you can significantly improve the quality of your return the sharp ratio of your strategy so even though you are neutral And that's what happened in August of 2007, right? Like that was kind of the quant earthquake when most of the market neutral funds who thought they are completely risk-free before that, like experienced Armageddon type of event. Now, why does it happen? Well, it happened exactly because that space is already capacity constrained. Most of the people, one way or another, pushing in the same direction, right? Like taking position in the same direction.
So when the last person who cannot compete begins to liquidate, it's actually in that process, it creates an earthquake for everybody else.
00:19:16 [Vincent Weber]
I think you wrote an interesting paper on that topic. Where do you see the sweet spot?
00:19:22 [Misha Malyshev]
Yeah, I think, probably the the the sweet spot in terms of the size of investor the sweet spot is probably from a billion dollar to 10 billion dollars right like that that kind of a horizon that that's where you can afford to be very very sophisticated in your due diligence at the same time you're not too big that you don't know where to just to put your capital, right? Obviously, your typical sovereign funds with the trillion-dollar-plus portfolios, right? They are destined to beta investment, right? Because there is just not enough alpha.
00:20:04 [Vincent Weber]
So now I have a somewhat nerdy question for you. As a scientist, there's likely an investment channel that you hold dear, one that you find yourself returning to repeatedly, almost like a pet problem. in the best way possible do you have such a problem a pet problem
00:20:26 [Misha Malyshev]
Yeah, we are trying to build a multistrat. We are trying to do it in a very different way from other people. We believe that just recycling of managers, managers coming from one place to another, then to another, to another, is not an interesting way to go for us. We are trying to figure out what What can we invent which would be different from other people? And it's a challenging problem. The reason it's a challenging problem is that at the same time, you're trying to focus on what you do well and diversify. So diversification basically means do as many things as possible. Focusing on what you do well means do as little things as possible.
So that balancing act of how you focus on only where you can really make the difference and at the same time make it as diverse as possible is an interesting problem. That's actually another interesting segue because one can say, well, if market is getting more efficient, is the pool of alpha getting smaller and smaller every year? And the answer is the pool of known alpha is getting smaller and smaller every year, except those pools kind of multiply. As the world grows, as beta grows, the alpha that follows that beta also grows. So number of inefficiencies or places for inefficiency increase, right?
00:22:14 [Vincent Weber]
Yeah. I have to ask you about your take on AI. That's the hottest topic of the day. so yes there is the burst but beyond that do you see any anything significant any significant change for your domain are there particular opportunities or are we just talking about increments it's it's a it's a fantastic question and we've been into call it ai or what we call neural nets or deep learning kind of more specifically
00:22:48 [Misha Malyshev]
for a while right now and if you if you think about neural nets there are kind of there are three stages of neural net attack on wall street the first one happened in in in 80s where neural net first appeared on wall street and everybody thought oh my god like what a great concept if it does something good you put a little bit more sugar if it does something bad you put a little bit more asset and you keep on training it until it until it supplement a human trader so that failed then in in 90s there was a second wave of neural nets like the let's call it support vector machines where people say okay like like that whole thing that we tried before didn't work because because we were we were sort of
like it's hard it's really really hard to train black box so let's kind of like a stick the electrodes into the black box and control this training process and people tried it again and that failed And then there was a period where people kind of forgot about neural nets. And then, of course, in 2012, Hinton and the crew revived it. And in 2017, of course, there was a first paper on transformer. Attention is all you need. And that paper basically was the beginning of what right now called AI chat GPT and all this stuff, transformer-based technology. We've been into that game for a while, except cautiously. Cautiously in what sense? It is very, very easy for machine to deal with complexity when there is a high certainty of the outcome.
For example, chess or even Go. The outcome is very, very certain. Is it a draw? black wins or white wins right like there are three states there is nothing in between it's not like oh oil can be a little bit lower or maybe a little bit higher or the price of gold is is that way so it's very very certain state while complexity of getting there is very high so humans don't deal with complexity but humans deal better with with sort of like less certainty so there's joke is that that human traders see a trend where there is not even a trend. And a machine, of course, cannot see that trend even when there is a trend because of high noise.
00:25:29 [Vincent Weber]
Misha, thank you for being with us today. I really enjoyed our discussion. And to our listeners, thank you for tuning in to Resonanz Spotlight and for making it this far through these episodes.
00:25:41 [Misha Malyshev]
Thank you, Vincent.