When does a rules-process become risky Quant?

Jason Zweig, one of my favorite financial journalists, has a fascinating article today about the dangers of quantitative investing - https://blogs.wsj.com/moneybeat/2017/05/05/quantitative-investing-a-crisis-waiting-to-happen/. It’s worth a read.

I’m calling attention to this in the wake of a meeting I had earlier this week with other investors who simply won’t touch anything that isn’t completely reliant on human judgment. Once you’re in the “rules-based” camp, they say, well that makes you a quant, and, summarizing their viewpoint, unless you’re the “mostest” with the “fastest” - i.e., unless you possess the most data, most brainpower, and are processing it all with the aid of the fastest computers and high-speed trading - well you’re destined to lose. Lose to the investor at the top of the quant food-chain, and presumably, to “judgment-oriented” investors like them.

Well not so fast. Personally I think all of these criticisms are intellectual straw-men. Why does one have to be the “mostest” with the “fastest”? For that matter where is it written that complexity is the only path toward investing success?

Furthermore, I think their criticisms miss the point. Given the grim performance of active money managers - the vast majority of whom rely on their investment judgment - it’s worthwhile to remember that we are all trying to beat a simple index, i.e., a market-cap index. That’s the bogey. Note I said “simple index”. But I didn’t say it’s necessarily easy. It’s difficult to beat that index. But is complexity and speed the only path? I think not.

I think it helps to get back to basics. For example: what are the flaws with a market-cap index? How can those flaws be bested? Once better performance is discovered, how does one accomplish superior performance with less risk/volatility? Etc., etc. Now, I think it’s true, as suggested in Jason Zweig’s article, that many quant investors go astray with their complex models, and that they don’t realize how they “become” the very market they are trying to beat. But my point remains - a few systematic, uncorrelated investment strategies, each anchored by nothing more than just a few time-tested metrics (say, 2-3?), just might do the trick. Now, is that quantitative investing? If so, how is the S&P 500 not a means of quantitative investing? How many metrics does it take before a simple rules-based approach becomes dangerously quantitative? And who decides the answer to such questions?

Before writing this I went back and re-read Warren Buffett’s speech at Columbia University, the “Super-Investors of Graham and Doddsville”. A classic. Granted, his focus is solely on value-investors and the non-random success of disciples of Ben Graham. But, in the context of this thread, I have to say that some of the successful investors he profiles in that speech, like Walter Schloss, would probably be branded as “quant” investors by critics today. If so, consider me guilty of the same “quant” label. I’d be happy if, at the end of my investing career, my methods and performance are as “flawed” as the least successful investor among the group highlighted by Buffett in that speech. But until the “judgment” crowd proves the superiority of their approach, I’m not going to lose much sleep over their critiques.

Alright, time to get off my soapbox. But I welcome input from others here.

Ed

Nice post.

I am simply reminded that playing the market is a zero-sum game which has a risk-neutral expectation equal to the market’s rate of return less costs. Costs (direct, indirect , and
otherwise), and under-diversification (due to cross-sectional negative skew) explain why most active managers under-perform.

Given a zero-sum game, one must be cognizant that the market is by definition capital weighted; participants with the most capital tend to be the best dressed, most well-connected, the smartest, and to have gone to the best schools. So, the likelihood is that on the opposite end of every transaction there is a buyer/seller who has a justification which is just as rational as my own.

In my mind, this is no different than poker. In order to be a poker player, you need to master the rules. But that will not confer any edge since all other players (at the grown-ups table, at least) also have mastered the rules and the betting strategies. In order to be a good player, you also need to master the psychology of your opponents. Whereas in poker you usually sit across from your opponents, present day markets are almost totally disambiguated. The only forms of evidence as to who might be taking the sucker’s bet are circumstantial: SEC filings (forms 4, 13F/D/G) and disambiguous market footprints (trend, volume, short interest, etc). The only alternative is to have an arbitrage (e.g., insider information, better strategy, better data, superior subject-matter expertise, etc…). Do simple quant systems offer any hope of providing arbitrage?

It’s like the Buffett proverb: “If after 10 minutes at the poker table you don’t know who the patsy is, you’re the patsy.”

Ed, my subscription to the WSJ has lapsed (I don’t like their cookie tracing policy).
Can you copy into this thread the essence of what you think Jason is saying from his article? Not all, just the essence.

I am reading Hagstrom’s book right now on Buffett. From what I can tell so far, Buffett is actually pretty rules based and process oriented in deciding what he invests in and he has a lot of discipline and little emotion (other than maybe for the newspaper guys like Graham Holdings) in his decisions. He likes good ROE, expanding book value, low capex required, little debt, good margins and a ‘fair’ valuation.
So was Graham with net current asset value.
Both sound like they could have used a ranking system; they just did it in their head.

I think the advantage we have with P123 is the breadth of stocks we can invest in (since we are mostly pretty small investors where we won’t move the market) and a mechanical approach takes the emotion out of our decisions.
We don’t get hung up on defending our decision to invest in Valeant or Gold or Copper. Active guys have to deal with heuristic justification. We don’t.

As to Zweig’s article, he interviews Richard Bookstaber, former hedge fund manager, author, and who currently oversees the University of California investment portfolio. Brilliant guy. One quote stuck with me, I don’t have it quite right, but it’s something along the lines of “once you model an anomaly, you lose.” I’m not sure I totally agree, but I intend to read the book.

Primus, good point, I should have acknowledged the zero-sum game.

I should add something that keeps nagging at me about that fact (zero-sum game), which may hold an opportunity for those of us a bit less well-dressed. We are told that the outcome of active investing is analogous to a series of coin-tosses, with a distribution of outcomes approaching 50/50 (though for investing, when factoring in fees, the probability of a superior outcome is, at best, say 45%). Here’s my question - why is the performance of active management so far from that figure? Depending on the study, at best only 15-20% of investors outperform, and if the recent Standard & Poors study is to be believed, that figure is closer to 5%. Is that proof, as the academics would assert, that the EMH is working as advertised? I think not. I think it exposes a flaw in their reasoning. Yes, it is unlikely that 45% of investors will outperform, as all sorts of factors - emotions, changing strategies at exactly the wrong time, etc., etc., will mean fewer than 45% will outperform over the long run. But only 5-20% outperform?? I think the very poor performance of active managers exposes a flaw in academic thinking. It tells me that behavioral biases are very much at work. In other words, investors are not the rational agents that academics claim. And yes, the same goes for the well-heeled, well-dressed, well-schooled, well-capitalized folks. They make the very same mistakes.

I think therein lies the opportunity.

Looking back to my high school poker days I can say there were some games where I could not tell—at the time—who the patsy was. Older now and with Buffett’s wisdom I have finally figured out who the sucker was (me).

Each poker night might have been zero sum: we buy our chips, play till I lose them all and I go home. The number of chips stays the same, during the game, and we call it zero sum.

But what if I did not realize I was not as good as the regulars in that game (or even that the deck was stacked) and I came back a few weeks in a row. I would bring more money. Is it zero sum over time?

I keep reading books that suggest value investing works because the retail investors get shaken out at the bottom leaving some money on the table as they leave. And I keep wondering how that explains things in a zero sum game. This is equivalent to one poker night. The amount of money is limited here.

But what if those same investors (that got shaken out) get excited when the market rebounds, maybe having remorse at not being in the market and buy-back-in with new money that they earned with hard work at their jobs. This is like multiple poker nights and there can be a continuous flow of money into the market from some and out of the market to others.

We focus on the money on the poker table or the total market cap of the stock market and call it zero sum. But unlike Buffett, some people pay some capital gains once in a while and buy a home or a college education.

Even in zero sum game—like poker—there can be a continuous flow of money to the smarter investors (or the ones who stack the deck while you are getting a beer) over time.

I am a big fan of Game Theory and I am not trying to come up with some stupid hypothetical. I think it is a zero sum game over short time-frames as David suggests.

But there are multiple plays of the zero sum game: multiple poker nights with new patsies and old patsies with new money each night.

-Jim

Those guys that rely completely on human judgement? Thank goodness for them. We need someone on the other side of our trades.

If you ever read about Ed Thorp, for example here, you might be forever convinced to leave your human judgement out of it.

All:

Buffet makes the point that over the long haul the US Economy (US GDP) has and will continue to grow. The stock market is a way to participate in that growth. So, if you believe in what Buffet’s saying the market is not a zero-sum game.

Bill

Here is an example of why it is not zero sum (ie, all participants either are better off or no worse off):

http://www.pragcap.com/is-the-stock-market-a-zero-sum-game/

In reading this again, the author of this example says it is not zero sum because one of the participants, who has a smaller net worth, thinks he is wealthier than that, ‘at least on paper’. So maybe this is not a valid example…

I think that David is letting us know that “zero sum” is a useful simplification and that the sum is actually expanding at the “market’s rate of return.”

Clearly “risk-neutral expectation” is a mathematical formulation slightly more complex that “0.”

In any case, I did not mean to oversimplify David’s ideas in my post and I am inclined to think he is on to something. I think that some of the complexities and subtleties that David seems to understand better than I do—like money flows, expansion of the money supply etc–might be important in some markets too.

-Jim

Another link to the article. At least, it has the same headline.
http://jasonzweig.com/quantitative-investing-a-crisis-waiting-to-happen/

Thanks. I can read that one.

The first thing after I read this though, was everyone crowding into SP500 index funds, not quant funds. Read the article with that in mind as well.

Take a coin that’s weighted 45% heads, 55% tails. Equate “heads” with the probability that a given fund manager out-performs the market in a single year. Tails equates to under-performing. In any given year, the probability of beating one’s benchmark is 45%. But when you repeat those odds over and over again, the cumulative odds of beating the market (i.e., having a win rate >= 50%) is much lower than 45%. This is the essence of binomial (i.e., Bernoulli) probability .

Take the often cited case of 20-year track record. Given these odds, the probability of beating the benchmark (i.e., winning > 50% of tosses) over 20 years is just about 25%.

So even in a perfectly behaved “normal” world, we can begin to see why fund managers’ under-performance grows over longer time frames. If you add in non-normality, those results become even more striking.

Non-normality exacerbates the under-performance because the cross-section of equity returns exhibit negative skew (i.e., a few winners are responsible for much of the mean returns) and excess kurtosis (extreme events are far more common than what is implied under a normal distribution which can be completely described by mean and variance). By definition, active managers hold more concentrated holdings than “the market” and are therefore likely to miss out on the few securities which provide most of the returns.

I do not think that any of our explanations necessarily exposes a flaw in the idea that “markets are hard to beat”. But I think it does, however, suggest one pragmatic principle: in order to be in the top 90% over the long-run, one simply needs to be in top 60% consistently. I.e., if you flip a coin which is 10% in your favor 30 times, there is about a 90% probability that you’ll win at least half of the time.

The idea of risk-neutral pricing is that there is at least one probability distribution which does not result in arbitrage. That distribution is the risk-neutral expectation. I simply meant this as a way of demonstrating how active money managers are able to earn some return even if they lack skill.

You are giving me way too much credit. For the record, I do not understand macroeconomics. I do think, though, that the central banks have played in role in supporting high asset prices and that easy money has been detrimental to stock & bond pickers’ abilities to sort out the good and the bad. Just look at hedge fund returns in the last decade since 2008. There is littler incentive now to find mis-priced risk than probably ever before.

I also hold ETFs responsible since they cause everything to move together. When someone, like me, buys an ETF, he/she creates an unnatural bid for everything in the ETF. On the way up, passive investors are pleased. But on the way down… if we ever get there… their pain could be an opportunity for us value-minded guys (and gals).

From what I’ve read, it seems that one of the biggest reasons that active managers find it hard to beat a benchmark is the inflow and outflow of clients’ funds. When the market is doing well, fund managers have an excess of clients and are forced to invest near the top of the market, when they can’t find many bargains. Then, when the market is doing badly and there are bargains galore, managers have far fewer clients and can’t buy the securities they think will beat the market. The benchmark, by contrast, consists of an imaginary fund whose “clients” never withdraw or deposit money. That gives the benchmark a huge leg up.

An alternative way to measure the performance of active managers would be to look at the net asset values per share of individual closed-end funds. My gut tells me that their performance will be better than the performance of mutual funds. Who knows–perhaps lots of broad equity-based closed-end funds actually beat the market when you look at their NAV. Unfortunately, I cannot extract that information from P123. I can get it on a graph by using P123’s discount-to-NAV data, but not on a spreadsheet. It would be nice if closed-end funds were available for custom series or if the data used to generate graphs on the fundamental charts were downloadable.

Interesting read. I’ve been a “quant” for 25 years. In the early days, I was considered a crackpot. By the late 90s, the quants had done so well that all the firms were hiring geeks. In 2008, (and Aug 2007) there was a massive quant meltdown. Value stocks had done so well 2000-2008 that the trade was way over-crowded with quants chasing the obvious anomalies. Deleveraging in '07 and '08 led to a slaughter of long/short quant funds. That was a shock since they were market neutral so should have cruised through the bear market.

As for the quote “once you model an anomaly, you lose.” …

In the 90s, almost all the anomalies I modeled were winners. Now, you need to be careful because there are so many quants with so much money chasing the same thing. But, that creates opportunity for the clever and the small.

As for “beating the benchmark”… Warren Buffett doesn’t seem to understand that Hedge Funds aren’t trying to beat the S&P. They are generally maximizing Sharpe Ratio. As a quant, my Sharpe Ratio has beaten the S&P in every 3 year overlapping period except for the 2010-2012 period. I thought for a while that maybe I couldn’t do well anymore. It is certainly much harder than the 1990s when I just picked up dollar bills in empty parking lots all day. Now I’m picking up nickles in a busy parking lot… but that’s a lot better than throwing darts or picking up pennies in front of trains.

The game always changes but in some ways, it is always the same. 90% of quants will probably fail just like 90% using humans judgement fail.

This is a link to a quant hedge fund that has “beaten the market” over 15 years but has also had double the Sharpe Ratio. They made 11% in 2008 and 12% in 2009 which was a much better ride than Buffet who lost 34% in '08 and made 28% in '09… which BTW, left him still down 16%

https://ctaperformance.com/qim

Ed,

Your sentiments are on point. This is an important and frustrating issue as what we do is very badly mis-understood and under-understood. Actually, I should walk back my use of the word frustrating. It frustrates the hell out of me in dealings I have with other professionals. But in terms of opportunity for us, the more “they” fail to understand what we’re about, the more rom we have to profitably do what we do.

Anyway, back to the main point.

I believe quants in general have made an embarrassing mess of what they do and deserve scorn. They don;t understand the markets .They don;t understand the data. Their study designs range from lousy to downright stupid (I just read an scholarly article on equity income ETFs in which the professors never thought of looking at summary prospectuses and didn’t realize that about a third of the ETFs they studied were not really equity income). Data min ing and over-optimization is not just something we discuss here; it;s become so widespread and well known as a problem, I often, when writ5ing on Forbes or Seeking Alpha, expand my discussion of testing to show why what I do is not data mining. And there’s the issue you raised; the so-called quants have allowed the world at large to assume that the most speed wins, which is not at all true; the main proponents of more-speed-is-better are the tech writers and companies that sell the technology (I hear remarkably little from people who actually invest or trade).

What we do and what they do are worlds apart; we’re caught in the middle. Active human-judgment investors think we’re quants. Quants think we’re active investors. The sort of rules-bassed modeling we do is something most don’t understand. In the outside world, I’ve taken to calling it moneyball investing hoping to latch on to the Billy Beane launched popularity of sports analytics. That really is what we do. We start with a very human decision making process and use objective analytics to improve it. One way to look at it is to say we do exactly what human-judgment active investors would be doing if they could somehow or other remove emotion and bias from their decision processes and limit themselves to objective facts.

Again, though, my reaction to all this is mixed. Professionally, I’m often frustrated. But as investors, we’re in good places. Those of you who have been succeeding with p123 have been helped by the fact that you’ve been seeing ideas, and jumping on them, that are being missed by those who don’t understand us. Since, at the end of the day, money talks, then let’s be happy as quants and actives continue to beat the daylights out of one another and continue to fail to understand out place in the world. Or to put it another way and paraphrase the old kiddie rhyme: Sticks and stones will break their bones and the money they drop will fall to us. :slight_smile:

OK. Time to get ready to watch the Bachelorette.

Whoa! What a cool analogy. Maybe I’m like Willie Mays. No too far. But I really do like that!!!
-Jim

Jim - Like Willie Mays you are certainly “out standing” in the field :slight_smile: