In this episode of The Acquirers Podcast Tobias chats with Mikhail Samonov. He’s the Founder and CEO of Two Centuries Investments, and has extended the value backtest all the way back to 1825. During the interview Mikhail provided some great insights into:
- Value Investors Should Double Down
- Studying Historical Markets Makes You A More Resilient Investor
- Value Crashes: Deep History
- Value Has A Long History Of Never-Ending Drawdowns And Tiny Intervals Of Outperformance
- Value Backtest To 1825
- Benjamin Graham’s Backtest
- OSAM: Value Is Dead, Long Live Value
- Value Investing: Even Deeper History
- The Difference Between Volatility And Risk
References In This Episode:
You can find out more about Tobias’ podcast here – The Acquirers Podcast. You can also listen to the podcast on your favorite podcast platforms here:
Tobias: Hi, I’m Tobias Carlisle. This is The Acquirers Podcast. My special guest today is Mikhail Samonov of Two Centuries Investments. Just like the name of the firm, he’s extended the value backtest all the way back to 1825. We’re going to talk when Vanderbilt got his steamships, when Charles Babbage invented the difference machine, when value outperforms, how it draws down, right after this.
Value Backtest To 1825
Tobias: Yeah, so maybe you could tell me a little bit about why you did this research because as we were discussing earlier, I saw your Momentum research. Looking back over two centuries of momentum, thought it was absolutely fascinating. I was really hoping somebody would do it for value, so I was really excited when you did. I think the first time I saw it was just after you published in May. So, you’ve taken several different data sets, and you’ve knitted them together and now you’ve got data for the value factor going back to 1825. What inspired the research project and where did you track down the data?
Mikhail: Yeah, the inspiration actually goes all the way back with the long-run investing, actually just long run in general. I was an all the way undergrad and my macro professor who is the founder of this development theory at Brown University, he posed a question to our class, “Why is there no single economic model that explains 20 centuries of growth?” And the first 17 centuries is basically zero growth. One model explains it, [unintelligible [00:01:47] and then we have the Solow growth model for the next 300 years, but there is no complete model that explains both. I was fascinated that you can just ask questions like this. And it became my thesis and we invented this whole method about evolution and how it referenced different duration of life before Industrial Revolution, after, based on knowledge transfer and physical work, etc. He continued publishing books on this topic and I walked away just feeling that science is not just about cramming equations into your head, but you can actually ask really cool questions and then answer them. And then, also this idea of long term really fascinated me.
Then fast forward to 2007, I’m at Wharton Business School. I’m also quant in Wallstreet, doing factor stuff which we would call now factor investing. I looked at Jeremy Siegel’s book and the breathtaking graph of stocks and inflation and bonds over 200 years. For me especially, it’s like very aesthetically pleasing to see human history and finance all in one picture. Also, growing up in Soviet Union, where you have every decade, complete wipeout and currency stocks and bonds and everything, seeing this magic compounding, you can just sit back and passively compound your wealth, too good to be true, just breathtaking.
As soon as I saw it, this question popped into my head, if the indices go back to 1800, why does the longest quant go back 1925 with CRSP? By that time, it was lucky in hindsight is that, I built my first backtest on 20 years of data. That was part of my AIG experience and then right away we funded the strategy, so it was complete and it started to grow. And then we bought data back to– the was first data set was 1980 to 2004. Then, we bought 30 more years of data, got to 1950. And for me, those 30 additional years was the closest I could get to out of sample experience. Watching some factors crash and burn, others go flat, some continue working, I realized how much chaos there is in this whole factor modeling thing in real time. That couple decades of data is just really not enough to know anything about the stuff we’re doing. And I got obsessed with this extension idea, and then the CRSP data back to 1925.
On the background, also Momentum, the investment philosophy that was within, I was the first quant inside a fundamental shop. They were very comfortable with value and quality, these fundamentally sounding ideas. We innovated like crazy around it, but at least at essence, they sounded good to the fundamental PMs. Momentum was always like this thing where, it’s too technical, just too technical. And then Eugene Fama, of course, in Chicago wouldn’t approve it either, the academic quant. So, I had this dose of doubt, even though the results were really benefiting from adding Momentum, both correlation-wise and return. And so, I was just always thinking about Momentum.
When we got back through this data extension to 1925, all of a sudden, I see that massive crash of momentum right after Great Depression in 1933, massive drawdown. And that was like a whoops moment. Because I’m living that history as if it’s coming to me, as if it’s future, not like some distant past I can ignore, it left a big impression to discover that. And risk management, expectations, how much weight to put in Momentum, I put a tiny, tiny weight. All my factors are very small weights.
And so, when I saw Jeremy Siegel’s graph, it’s right away momentum, momentum. A, two questions, is it real? Is it going to exist on this untouched data? Then you can say it’s data mined. And B, how often does it crash? What does the true distribution look like? By December 2008, like I mentioned, that’s the first little backtest I did. Yale International Finance Center there, Professor Goetzmann, they’d very nicely just put on their website, the stock level data. It’s been sitting there since at least 2008, which is when I discovered it, I think it’s been there longer. And it was enough to run something really quick and dirty. But it was there, the spread was statistically positive, lower than recent history, but the crash has just jumped off the page.
Once I got that little result and Chris Geczy at Wharton saw that, as my independent study advisor there. And right away, he’s like, “We can publish this.” And then, we went to GFD, Global Financial Data, and Richard Sylla’s website, who is another historian also published free stock data, blended it all together, and we got Momentum. And then it took nine years to make it sound very academic and get all the feedback, all the other robustness tests. But crashes are there, and then that momentum crashes, of course, became a popular topic.
But the thing with long history is that a lot of people either hate it or love it, either they say, “Why do you go so far back? It doesn’t matter,” or, they just obsess over it. I’m sort of like fascinated by it. But I take a moderate approach. I say, long history helps us protect us from short history. Short history can be really misleading, especially about risk and things like crashes. We definitely set very high expectations for factors, I believe, both momentum and value, and many others when we just studied 1980 or 1950, to 2000, where a lot of us grew up as quants. And then 20 years later, we’re all like now struggling figuring it out. But if you go back 200 years, 1950 to 2000 becomes this unique period, and longer history has a more subtle way of seeing how good these things really are.
Value Crashes: Deep History
Tobias: Momentum is perhaps a little bit easier to calculate because you’re only looking at price when you’re doing that calculation. Value becomes more difficult because there’s a fundamental component and you have to find that fundamental data and marry it to the price data. And so, the challenge, I guess, is you have book value data going back to 1920 in the Fama-French– I think it’s roughly 9 to 20 in the French data. There’s two posts of yours that I just wanted to talk about you. The first one, you go back to 1871, and that’s using the data from– I’ve just forgotten. The Cowles data. Cowles who famously produced the cards that– he looked at all of these whether anybody had any stock-picking skill at all. I’ve written about Cowles before. So, how did you marry the Cowles data with the French data?
Mikhail: Yeah, the fascinating thing about Cowles data is that they lost all the underlying stocks because as soon as I had this idea at Wharton, the first place I went to was Yale. And I emailed them, and the Professor Goetzmann said, unfortunately, the industry data is still there, thank God, but all that hard work of punching those cards in, like some– they blame some administrative assistant, I don’t know if that’s true or not, but there was no computer and so it just disappeared.
There’s really a couple ways to go about it. If some academics want to pick it up, and maybe somebody is already doing it right now, there is a way to go more fundamentally in terms of data collection and look at company annual reports and just collect the data. That’s really painful for many reasons. Then, there’s a shorter quicker way with Moody Manuals, but they only go back beginning of 20th century. That’s what Fama and French used to get back to 1925. By the way, I asked them why they started in 1925 and they said, “Well, it was just a couple of years before the Depression and we couldn’t go too far back.” And that’s random basically.
Tobias: But at least it captured that. At least in the price to book data, you captured the Great Depression, which I always looked at like it was this complete anomaly. The book value breaks down and then book value doesn’t perform at all for this sort of 16-year period. And I just look at it as one of those things potentially you can just dismiss that as something that might never happen again. But the interesting thing is when you include that longer data, it really becomes clear how– they’re very rare the events, but they are still pretty regular.
Mikhail: That’s absolutely right. Exactly. In fact, if you think about the whole stock market and things like a 60/40 portfolio, these insights really carry over. Before 2008, nobody looked at Great Depression drawdowns or 60/40, which are like 65% for 60/40.Aand then, the 30 something percent drawdown for 60/40 in 2008, pretty much cleaned up a lot of people and they’ve had to rebalance it. So, yeah, long history is fascinating if you start taking these things– is repeatable events and hard to predict one.
So, in terms of your question, values. I’m sitting there in the spring of this year, value continues to get hammered. I’m feeling bad for all my friends who are value investors and I don’t do direct value like this investing right now. For the past 10 years, I don’t do traditional factors at all in equities, but I’m still a big fan of that research and keeping track of them. Value was on my radar. And this industry data from Cowles is sitting there filled with the P/E ratios, all this P/E ratio data there, for all the industries on their website, and they have total returns. And they have a lot of– I believe, 68 Industries. So, it’s quite a wide number of industries that they provide. It almost becomes very related to stock specific at some point when you go really granular in industry groups. It’s almost really pure value in the sense that the risk maybe factor more than alpha.
So, yeah, it’s using these P/E ratios and the total returns for industries and we rebalance once a month, we rank on the P/E, I do the lagging so that the earnings are lagged, to make sure that information was there hypothetically in this long deep history. And so that data goes back from 1871 to 1925 when Fama-French price to book data starts. So, that was the first part that I did.
Tobias: And you construct– so that the test is, it’s long short, long, the cheap third, short, the expensive third, rebalanced monthly, and you run it from 1871 to the present. That’s interesting thing. I looked at what happened in 1871. It’s six years after the Civil War. The Gilded Age starts in 1874. The telephone gets invented in 1876. The light bulb is invented in 1879. And value investing works, despite all this technology.
Mikhail: I know. I love the way you put it. When I give talks on the Momentum paper, which started in 1800. It was fascinating to say– the stock market got started a few years after the Revolution and the Independence and then started under a tree on Wall Street.
Tobias: The Buttonwood tree, Buttonwood Agreement.
Mikhail: Buttonwood tree, two dudes traded some stuff, and then it moved to this Tontine House, I think it’s called which was a gambling–
Tobias: The Tontine Coffee House.
Mikhail: I love that whole gambling and investing. Like during the day, you’re a commodity trader, at night you’re just– [crosstalk]
Benjamin Graham’s Backtest
Tobias: Yeah. I love it too, 100% agree. The really interesting thing– I saw this in your blog post, you talked about Benjamin Graham referred to a backtest. You think that he was talking about the Cowles data, when he was talking about his backtest recently. He looked back about 50 years and he thought that value had produced about a 15% return. Presumably, that’s the long side that he’s talking about there.
Mikhail: Right. That’s when it appears to be some of that early backtesting is taking place. I’d be shocked if they didn’t know each other. You can ask Jamie, the guy who does all actual historical– look up the evidence in the newspaper. Catherwood, I think, was his last name. I’m sure they were familiar with each other’s work. I don’t know if he used exactly his data or not, but it appears to be at least in spirit, that’s a period that was being looked at. It, in many ways, echoes, we as quants around 2000 were backtesting, a really golden age of value. And if you happen to have very few drawdowns and then you got started, you’re going to have a lot of conviction in value, and then whatever happens next will either help a lot or not.
Graham had to live through some drawdowns really early on in his investing career. But then, they have this really nice long run, and then he was able to liquidate on the top. And then, Buffett took over and also had this really nice three, four decades. On the upside, we can argue there’s a lot of tweaks and things they do, obviously an expertise in depth and leverage and quality, but on the downside side, when you just have like value as a risk factor more, less of an off idea, more just risk correlation, and that really helps when you have that crash for– 30%, 40%, 50% crash for several decades, and now it’s coming back.
Tobias: It’s interesting to contrast their careers because Graham started– Graham had been going for a little while, but the Great Depression and the crash were closer to the beginning of his career. And then he had, what you describe as that golden age of– Golden age probably for both factors, for value and momentum, but value obviously relevant to Graham. And then, Buffett begins in the 50s, I think, he similarly gets that golden age for most of his career, but now he’s encountered what is at least as bad as– so the worst one that you had in that data was 1904. I think it’s the long short factor draws down 59%. I don’t know where your data ends, but I think it’s May or April or something like that. That was about a 59% drawdown, so are we through the 1904 trough yet?
Mikhail: I think we’re really, really close. There’s two posts. The first post just uses industry data, even Fama-French use industry data, which are price to books and I connected to Cowles price to earnings, but it’s all thirds. The max throwdown was 1904, 59%, long, short industries, which is crazy if you think about this industry spread. The 2000s were negative 42% spread, that makes sense, the dotcom, right before the bubble burst. That one recovered really quickly and that’s what we’ll remember. That’s when the value came back really strong and you felt really justified and validated after three years of tech pain, to get that back. But now it’s has been really going on since 2006 plus or minus. It’s been in these waves, but recently it’s really nose-diving. I haven’t updated it through July or August, I mean June. But I believe we’re very close to that 59% max industry level crash by itself. It’s crazy to be living through it.
OSAM: Value Is Dead, Long Live Value
Tobias: You had some really interesting findings. The one thing that you find, the full data set values positive generates about a 7%– and this is long-short, market-neutral, weighted equally to both sides of the trade. The worst drawdown is the 59% in 1904, which we’re sort of confronting now. It took 14 years for that to resolve itself. We’re 14 years or maybe a little bit longer into this one, does that mean we’re getting close to the end of it?
Mikhail: Well, so it took 14 years– Yes, the maximum drawdown in 1904 took 14 years to build up, and then another 9 years to recover. So, that’s the length, it’s long. And, yeah, this time around, we’re also 14 years and really close to this 59%. Looking at this data, and my blogs say this, it’s not a very sophisticated statement just to say, “Look, it happened in the past, [unintelligible [00:19:27],” but it’s just so tempting to say, “Hey, it’s going to be reversed.”
If I were to Bayesian adjust the strength of my belief, this is the only input I have, except the other one being that if things get cheap enough, eventually capital just has to flow there because of expected returns get high and we just can’t have this mispricing, if it does get to that [unintelligible [00:19:47] last forever. But, yeah, it looks like we’re there. But people debate this all the time. Why is this happening? I have some thoughts on that and I’ve been actually designing strategies to prevent that for the last 10 years, prevent this fundamental value that’s already published, and everybody knows about it
Tobias: What causes it and how do you avoid it?
Mikhail: Right. Well, it’s hard to tie it to long history because I have no idea– actually, so the OSAM guys, I’ve read their piece, that was really interesting. Industrial Revolution is concept, where you have a massive shift. They count, I think, five of those revolutions. Those periods line up pretty interestingly with even the extended value tests I have.
The way I interpret it is just the value of intangible capital becomes really much higher, intellectual capital, IP, brands, innovation, customer satisfaction, all these things. They eventually feed to the fundamentals. But now when they get priced a lot sooner and earlier, it’s much less about the tangible capital like book values, but also even the earnings You can see a lot of negative earnings are still getting beat up. Of course, it’s always two sides. At some point, it does become overvalued in a bubble, and maybe we’re already there. But during the last– at least since this 2006 period, I think that’s one factor is that things that are part of company’s value have become very different than more tangible value.
Tobias: One of the questions that I had looking at– just referring to the Jamie Catherwood, OSAM piece where they looked at that– they said those 16 years, I think that they said it was the sort of rollout popularization of the automobile, which meant that they needed to build the infrastructure for cars They needed roads, they needed fueling stations, and so on. And they said that’s comparable to the current situation, except you substitute the internet for the car. You build out the infrastructure and then you have these layers built on top. So, does it start in 2006 or does it start in 1999 and we had this little catch up midway through, but now we may be much longer into this?
Mikhail: It’s a good question. You could might as well– because if you go that far, if you basically ignore the 2001, 2002, and 2003 where value really caught up and recovered this drawdown, at least if you look at industries, which is maybe slightly different from– probably going to be a little different from stock level. But if we just keep talking about the very granular number of industries and the backtest, that goes all the way back to mid-80s, the drawdown. Maybe you can even say that– Yeah, the entire 90s and we’re still in that 30-year window of transitioning to the digital world. Of course, now it’s accelerating.
Value Investors Should Double Down
Tobias: So, the last thing that you discuss in that paper, you say that there are a number of reasons why people think that value works. I think that you said, “I don’t have any view on whether it’s a risk argument or whether it’s a behavioral argument,” but do you have any thoughts, why does value work?
Mikhail: The risk story is very good in many ways academically, conceptually, is about all factors, value and momentum, anything else. Even the stock market. The premia story, compensating for risk, we all speak that language. It’s very good for risk management. It’s very good for diversification. It’s very good for attribution if we can decompose Buffett’s returns and having these core attribution models. And we’ve gotten really good at it, it’s also very engineering friendly, and it just fits a lot of math and models.
But I think there’s a big risk of saying that, if something is risky in terms of covariance, correlation, we’re going to tie it to some theory that there’s a compensation necessary for that risk and correlation. And then we’re going to say, we’re going to put it in a portfolio in a massive way, because we expect, just like the stock market, to deliver an equity premium, and then have these other factors delivering premium because, look, they crash and they correlate stocks that have those exposures correlate with each other. So, they influence the risk matrix for sure. But is it really like safe to really count those premia always being there?
I always go back to being the quant. I started in 2004, building the first quant models. We launched in 2005. By that time, 75% of the market was in the hands of top three players and the whole quant space was less than 10% of the whole industry. 90% was fundamental. Within that 10%, 75% was the top three quantum firms. We were doing a lot of stuff.
We weren’t doing one factor with a simple formula, give it to everybody. And it was brutal to compete. Then, the top three blew up in ’07, and then again in ’09. And then, the whole quant industry went dark for a while until this revival through smart beta enhanced factor investing and ETF, transparency, liquidity, all that story came about, which in one way, we’re all happy, is this old quant saying, “Hey, quant is popular.” On the other hand, I was just going, “Oh my God, how is this gonna play out? Because all of the sophistication we used to have didn’t beat index.”
Our strategy had 125 factors back then, and it continued to outperform for 15 years. It did positive 2007, positive 2009, [unintelligible [00:25:40]. So, I navigated this turmoil of both my firm falling apart, market falling apart, quants falling apart, and just focused on innovation, innovation, innovation. Yeah, that’s why a lot of my blogs come out, evolve and stuff. But why value works? Why anything else works? And sharing with you how I was discovering the early history as if it was live and out of sample, you become really humble and you start to figure out how many ideas do I need to have? How accurate are they going to be? What is my real goal here? Is it to beat my competitors or is it just to at least have one basis point above the benchmark? That was my goal back then, now it’s different. But back then, I was a young quant. There was nobody above me. I was on the hook. So, I just said, “I don’t care about how much I beat this benchmark. I just need at least one basis point a year to say I’m positive.” And it ended up being hundred basis points.
I didn’t care if my individual idea was right or wrong, or Fama and French would laugh at something I did there. I didn’t have to write a dissertation over it or do a mathematical proof. It was intuitive, gut — If I think of it and I test it ad it’s positive, that’s good enough, take it in. Worst case, it is going to dilute the positive ideas with random noise. Absolutely worst case, it’s going to flip and become from a positive t-stat to a negative t-stat, which is really, really unlikely, most likely by data mining too much, it’s going to become random. But because I’m not just pressing buttons, I’m constantly thinking, reading, and innovating, there’s going to be enough of these gems, I just don’t know– if you take a bunch of sand, which one is a diamond and which one is a sand? Only time will tell, as I discovered with these extended decades backwards. So, I had to have enough ideas in there to continue staying positive.
And with value, we’ve seen this with quality. Around 2000, there was a lot of literature, Sloan and those guys in quality factors. It kind of went flat for a while. I remember looking at that, and it was really like an inverted hockey stick. If you did a long-term backtest, it’s the scariest type of outcome. It’s one thing where it just keeps zigging and zagging and then you add other stuff to diversify, that another one is pretty straight and then it just goes horizontal.
Tobias: And why is that? It’s just been flooded with capital?
Mikhail: That was my thinking, yes. A lot of quants were doing it. A part of it is also being a little contrarian– that’s even unpopular. Contrarian, I think, gets misused a lot. It’s like you just contrarian to the market prices. But even contrarian can either be popular or unpopular view in given times. Being unpopular is really, really hard. That’s when you say something and people look at you like you’re crazy. Why are you doing this? It’s much easier– It’s very popular in the spring of this year to be saying, it’s all going to hell, and I have that instinct, I have that bias. It’s a contrarian view, when the market keeps going up.
So, with quality, I looked at it. It wasn’t hurting our models. It wasn’t adding to our models for that early period. I decided to double down research and we went really granular into the balance sheets and cash flows and modeled very subtle things, put it all back together, quality 2.0, backtest got better. And then, I just said if people are going to hate it and see this hockey stick and pull money out, we want to be there, pluses defensive. And then around, ’08, ’09, and actually for many years later, that part of the model delivered a ton of value. So, right now, with value, if you’re a value investor, I just say double down– do your research double down, you’re forward-looking. Definitely don’t pivot out to growth or anything like that at this point.
My thinking led me to intangible capital in terms of value for quite some time, and I think that’s where I’m very comfortable, but let’s say if there was a more fundamental element there like direct, then it’s just about at this point, I think, A, waiting for mean reversion, just natural cycles. B, waiting for enough people to abandon it, which I think we started to see that–
Mikhail: That was fun. Yes, capitulated. [chuckles] And then, there’s the question about just long-run mean to the whole thing. That’s the third equation. But first is writing it, tweaking it, getting upside when it happens. At this point, I would say the odds for the upside for traditional systematic value is much higher than the downside. Of course, very famous people recently made the same statement and go their– [crosstalk] [laughter]
Value Investing: Even Deeper History
Tobias: You’re referring to Cliff with his Valuesberg Address and so on. The second post that you put up which I think is fascinating, this is the one where you take Goetzmann’s data and you go back to 1825. And just so folks know what was happening then, Charles Babbage invented his difference machine. John Quincy Adams is the President of the United States of America. Vanderbilt has the new industry of the day, which is steamships. He’s operating a fleet of steamships, and he goes on to become one of the wealthiest men in the States. It’s regarded as the first Industrial Revolution. It’s one of those early periods of technological advancement that potentially is very bad for value. It seems like that’s the thing that hurts value the most.
And so, the way that you’ve constructed the data through that earlier period, and you say that’s a very rough approximation of the value factor, you’re forced to use the dividend yield. So, how do you adjust when you’re using the dividend yield and what’s wrong with using the dividend yield?
Mikhail: Right. Both of those data sets are Goetzmann, the price data and the dividend data. Dividend data is annual. The number of companies that have dividend data is much, much smaller than the monthly price data. You can’t really tell whether the companies that don’t have a dividend don’t have it because they didn’t pay it or because they just didn’t find it. So, the safest thing to do is limit the entire universe just to dividend-paying stocks that Goetzmann identified, and that becomes our universe. It’s much smaller. It’s 256 companies that are in that. 1825 is when they start, 1871 is when they end.
Tobias: Do you have an idea– the other companies that weren’t paying dividends, were they listed companies? Do you have any idea how many of those there were?
Mikhail: In their data set, there is, I think, a total of– I might be wrong, 671 companies with price data. But then, of course, global financial data has a lot more companies in there, but that data set, you have to purchase and I’ve worked with a data set when I was at Wharton. So, there’s actually a lot more companies. Just as a side note, what’s interesting is when the CRSP data starts, the number of companies in that first month, the second month and first few years is so much smaller than when global financial data ends in that same period. You have multiples of number of securities– [crosstalk]
Tobias: So, they just weren’t tracking them. They just missed them when they started tracking.
Mikhail: Yeah, survivorship bias issues and all that stuff. So, all of this is highly, highly approximate. What I get out of it is general patterns of drawdowns, correlations. But interestingly, the average return, the spread here is still very similar to the other periods. It came out 3.7% per year for this ’25 to 1871. And then comparing to the 1872 to 1926, it’s also 3.9% a year, that’s industry spread. And then Fama-French is 2.7. Actually, it’s lower a little bit. Depends how you look at it. I try and make the most consistent definitions for each period.
Studying Historical Markets Makes You A More Resilient Investor
Tobias: There’s a few philosophical questions. Why use a very long time period to– What does that give you?
Mikhail: First thing, it gives me a better sense of the truer distribution of whatever we’re looking at. We all know things are not normal in finance, even the basic S&P 500 returns are not perfectly normal. If you look at left tail, which is what we’re worried about, the crashes, we know they’re not normal. So, if we just look at short history, we’ll totally miss them, most likely. Especially if we accept the strategy and start investing using a strategy, it’s very likely that a left tail was not in the recent past, unless you’re very unpopular and you’re just going to go at it.
Starting with Momentum, when I discovered the first left tail in 1933 and Fama-French data, I just got obsessed. What other left tails can we find? And [unintelligible [00:35:25] 200 years of history or 125 years of history, we’ll give that a shot. The second is this thing that keeps– When I was at college, I was a more theoretical econ guy, definitely didn’t believe it was possible to beat the market.
That was actually– I was very influenced by markets are efficient and forget about it. And then, when I got that first job as a quant, where I had to come up with a model to beat the market, I was terrified. Almost out of fear there, I’ve just kept like innovating. So, this idea of data mining where these factors are data mined and they are real, when you add untouched– it’s not pristine in terms of quality but it’s pristine in terms of nobody messed around with those 130 years of data. And you test something and it comes back with a positive statistical t-stat, it’s like, “This thing is real. I can almost touch it.” At least it might be dead now but at least it existed and I’m not just miraged by somebody fitting in history. And so those are the two major, major reasons.
If I continue to studying when something happens like a crash and trying to explain how there are factors in the macroeconomy or innovation or Industrial Revolutions, or whatever it is, that help understand the spreads. But I think it’s very hard to time them just in production point of view, even to implement these trades on timing factors. It’s very hard to give up a lot of correlation benefits because they don’t correlate.
So, if you start overweighting one, you’re just sacrificing a lot of diversification and that’s not even to mention the fact that predicting when one does better than the other. It’s really, really hard. So, I like history from– it gives me confidence about the left tail. And the thing with left tail– my whole philosophy just for a second about what is risk? What is volatility? It comes in here and that’s what drives Two Centuries Investments, the firm I started.
I think about it as an asset owner, less as an asset manager, making products, raising assets, that’s all great. As an asset owner, my own capital is long strategies, what is risk for me? Volatility is different. Volatility is something is moving up and down. If I understand it, if it’s expected, if it’s normal, I don’t sweat things. Risk is when I give up on something. Risk is when I have to change a strategy, I’m going to lock in permanently some losses. I’m going to give up the unrealized gains and cancel out long-run compounding, start over. Risk is when you’re lost, you don’t know what to do, you give up.
It’s kind of behavioral for me, but looking at left tails, those crashes for value momentum or in my case, asset allocation like 60/40 or other stuff, I want to really put in that worst-case as if it is really a worst-case. Like 60/40 portfolio, worst case has been almost 70%. Doesn’t mean the future can’t be worse. Hopefully is better, but can be even worse than 70%.
But if I can’t even stomach 70% loss, and statically keep rebalancing– when that’s happening, that means you have to be selling bonds and buying stocks more and more during that depression. If I can’t honestly stomach it, and I have a long-term investment horizon, and I want my capital to move through decades or centuries for my kids to keep doing this, I’ve got to put in these long-term expectations that history gives me has actually that it could happen tomorrow, or it could happen 200 years from now, I don’t know. Building around that makes you more resilient and there’s less risk in the system, that risk in terms of giving up.
And then, you set up these expectations based on those bad case scenarios. If real life, you’re still within those expectations and you can sleep at night, it’s just volatility, that’s the kind of experience I want to have versus setting up very great high expectations, low risk, low volatility, low drawdowns, and then being constantly surprised. This stopped working, that stop working, what did I do? It turns on this like survival brain, which is you’re either regretting stuff or is worried. Fear, agreed. The words I like is anxiety and regret. It’s all this primal brain. When it turns on, it just doesn’t let you go. You’re going to be feeling it. You have to do something. We see people doing that all the time across levels of experience, committees, all the way to retail.
This year really amplifies both the value and asset allocation are static, this problem. And so, the more I can be in my frontal lobe, the executive brain, and all this stuff helps me stay there. Being rules based, systematic, study long-term history, be comfortable being unpopular.
Not being too afraid of having original ideas, because at the worst case, they’re going to be random noise. Versus popular ideas, at the worst case, get crowded and then you get this big capital movement. So, there’s that correlation with other managers. So, you could actually swing from a positive t-stat to a negative t-stat. It’s a long way of answering, why do I study long-term history. I don’t obsess about it too much. It’s just one of the tools I have, but it fits nicely within the risk management approach.
Tobias: Yeah, I love that. I couldn’t agree more. Just looking back over the 200-year history that you’ve created for value, what are the takeaways for you? What surprised you, what can we learn from that 200-year period? For example, one I would have thought was that at 59% drawdown in a long-short was entirely possible. And here we are, we’re about to confront it again. So, it’s not a theoretical possibility. It’s very real and we’re living through it right now.
Mikhail: Exactly. It’s very similar to Momentum. So, what’s shocking is how similar it is to my Momentum experience. When I looked at those results in ’08 and then ’09, and boom, there’s this massive crash that you never thought you’d see but you were afraid you will. Same things happening with value right now. You take all that long-run history and you think that’s really relevant. We’re living through something very similar. It was so tempting to say it’s totally irrelevant.
The second thing that was surprising is how safe this second half of 20th century for value was, and you have these great value investors that are like literally riding a wave. Of course, they outperform traditional value and add a ton of value on top of that, and there’s many other reasons they succeeded. One of the important ones being the psychological where I believe when you really go deep into something, you’re going to hold on much more in terms of risk of giving up. I didn’t give up in the 90s, kept going, got paid.
But zooming out, there was definitely a lot of positive wind pushing them in this value. And now, you see the cracks. Not even cracks, but the pain, the real deal. When you have a 60% drawdown in value, you start thinking about permanent loss of capital and because things are getting priced into that are looking really, really scary. So, I think that’s a real test for a lot of the value folks out there and I sympathize with that.
Tobias: Yeah, I love the way you frame it up. You said 1940 to 2006 was exceptionally safe, and I love that. You mean that in its literal sense. And then, the takeaway probably should be that value investing is not safe, that you can expect these gigantic drawdowns every now and again.
Mikhail: If you’re thinking about volatility, absolutely not safe. That’s where the risk premia argument often I think is just wrong. When people say, “Well, it’s volatile, it crashes, hence it must be risk premium.” That’s a very slippery slope theoretically, academically. But on the other hand, if your process is value investing, you just feel great owning undervalued companies, you have a process to do it, and then you look at 200 years and say, “Yep, one decade out of a century, it’s a 59% crash. I’m okay with it, we’ll just get through it.”
As an asset owner, you can do it. If you really believe it and you and you feel great, who cares? It’s still total return. I mean if you doing long-short, it’s really painful, but you can still hold on to that crash. If you’re doing long-long, you still have this just general market data pulling you up, and you’re lagging S&P for a while. Who cares? As an asset manager, it becomes a different story. It’s career risk, it’s clients get impatient, and all that.
One big thought, I’m working, recently expressing to people finally just to share is just like people either love or hate long-term history, people tend to split love or hate, active versus passive. They just fall into these two camps. They love alpha or they hate alpha. Or they love beta or they hate beta. As an asset owner, that’s an asset management headache. As an asset owner, yes, those are useful risk management concepts. But really, I care about total return with survivable cross risk. Survival meaning no margin calls, and I don’t give up on the strategy, that’s the main– [crosstalk]
And the total return can come– When it’s coming, both from alpha and beta, that’s even better. There’s two sources. If for a decade, it’s all beta and for another decade, it’s all alpha, it looks like a horrible passive management product. But as a total return concept, it’s great. Why not? Because you’re sticking with something, versus just having a passive index and then you still give up. Somebody comes in– like with Jeremy Siegel, he was in the stocks for a long run. His book, I loved it. And then he moves everything to dividend investing. And then to 60/40 and now 75/25, and there’s very strong intellectual reasons. But for me, that’s risk of investing. You start on path A and then you’re on path B and then you’re on path C.
So, to avoid all this, my takeaway from all of this is, yes, studying history, building your personal resilience, and just keep doing what you’re doing. Evolution and innovation is also part of my DNA. But if you blend things well, you continue to innovate, without having to switch what you were doing. There’s another misconception, I think people just obsess with investment process, and it’s just been these safe guardrails for people, which were good. I guess investment management back in the day, you could just switch around and go crazy.
So, all the asset owners put on this tracking and benchmark and don’t switch your style, stick with the process. That’s all great. But that’s all risk management. But where’s innovation in all of this? Where’s your edge? How do you keep moving forward? So, that’s balancing all that. And it’s never black and white one way or another. If you overdo one of those things, you’re going to get hurt.
But as an optimizer, you’re pulling different forces together. A good balance, [unintelligible [00:47:22] when I say they’re balanced, there’s really this battle going on inside of it, like sell, sell, sell, buy, buy, buy, and optimizer just finds that point, utility function, even if there’s some noise there. I like it when there’s tension. So, innovation versus investment process is one of those things that are pulling and pushing. But I think that’s where some of the success can be found.
Value Has A Long History Of Never-Ending Drawdowns And Tiny Intervals Of Outperformance
Tobias: Yeah, it’s absolutely fascinating. I just wanted to get your comment on that. Finally, you have this great line about value. You said value has a long history of never-ending drawdowns and tiny intervals of outperformance. Is that value?
Mikhail: That’s value. It’s like a melancholy normal. You go to the ocean and start fishing for this big whale. When is my glory day? The drawdown is over and it feels so good. I think the idea is to disconnect this emotional brain that feels so good when things are working and just to set things sailing, and there will be some really, really awesome periods for value ahead of us that will I’m sure get us out of drawdown and generate positive return. I’m really sure about that. The question is when and how much patience you need until then.
And yes, value especially, but everything like stock market, I obsess with drawdowns because, again, as an asset owner, that’s what you feel. You see your highest point recently and you’re like, “Ooh, I have this.” You don’t look at calendar, I mean you do, but secondary calendar. There’s a lot of ways to show performance that feels appealing to a client if you’re an asset manager, [unintelligible [00:49:06], if you’re an asset owner and you log in and you see the highest value that you lock it in, that’s drawdown. And pretty much will spend the vast majority, a lot of time in drawdown mode.
That’s the time where we stick with the process and continue to innovate, and then when the winds align and we reach the new heights. But surprisingly, that’s when you’re living month by month, year by year, that’s how it feels. But as soon as you zoom out, all that disappears and you just start to see– if you pick the right general horses and you’re not giving up, the risk is not ruining your switching, then the compounding kicks in.
The clients who have been with us for a longer time now, things scare them a lot less. A, they know our strategies are working as expected. But, B, because their starting point now has compounded into something material, the drawdowns even of the same magnitude, now feel less scary on the cumulative line.
So, a lot of it I think is about your brain and how you set up the scorecards. We have this awesome way to customize scorecards that have multiple ways for you to win. You can tell us, “No, this doesn’t apply.” But it’s about holistically measuring success, not in any sort of narrow way. But it’s getting from A to point B, which is decades from now with all this uncertainty. There’s going to be a lot of drawdowns and they check a rainy day, I’m going to get too depressed and give up on where you’re going just because it’s raining. Sometimes, you might change your schedule, but if there’s a big storm, like this March, it was definitely a lot of self-awareness time. Even everything’s working, but you still feel the primal brain, [crosstalk] brain kick in. You call your friends, you call the pros, you call the experts, you shake things out, and then you keep going, even when everything’s risk-managed and we’ve seen this before, etc. But yeah, drawdowns are the pain of our existence and we have to deal with that pain in any way we can to keep going.
Tobias: Yeah, I think that seeing them happen historically is certainly a way to prepare yourself mentally, at least for the ones that you’re currently enduring or ones that will come into the future. It’s absolutely fascinating. Mikhail, if folks want to get in contact with you or follow along with what you’re doing, how do they go about doing that?
Mikhail: Thanks. Yeah, I have a blog and website, twocenturies.com and it’s T-W-O centuries dotcom. My email is Mikhail@twocenturies.com, M-I-K-H-A-I-L. Twitter: @msamonov. We have a lot of conversations with our readers. There’s all sorts of exciting research projects, consulting things that are coming alive. Definitely, part of my passion is to help the industry get better. I’ve been in finance for a long time. From day one, I was going to conferences and seeing things.
I just wasn’t inspired by traditional quantum finance or finance in general. I would maybe look at other industries and feel all like, “Oh, this cool stuff happening there. There’s a movement, there’s purpose.” And that’s definitely drives what I do and that’s why I try to share as much as I can and collaborate with other buyside even firms that would do more embargoed, but it’s still IP sharing, just so we all can get better. There’s a lot of win-win situations from collaboration more in this industry, more creativity, more innovation, and learning from each other. And I really liked your work which you do, which is really fascinating.
Tobias: Thank you.
Mikhail: I’ve been listening to a lot of them and seeing you on Twitter. That’s the kind of thing that was missing for a long time and now, it’s getting more democratized, I’d say. But there’s this great benefits of this community and you’re kind of learning from each other, and there’s more purpose behind all that I think, and I really enjoy that.
Tobias: That’s very kind, Mikhail Samonov, Two Centuries Investments Thank you very much.
Mikhail: Thanks, Toby. Have a good day.[outro]
For all the latest news and podcasts, join our free newsletter here.
Don’t forget to check out our FREE Large Cap 1000 – Stock Screener, here at The Acquirer’s Multiple: