On 1/19/19, Bloomberg published an article by Satyajit Das likening quantitative asset managers to “alchemists seeking to transform base metals into gold” based on “pattern recognition” (correlating “past periods of superior return with specific factors”) but being foiled by “hindsight bias” and “practical matters, such as what data is or isn’t available.” Yes, there are quants who work that way. But by defining quant investing based on their activities is akin to defining medicine in terms of those who thought application of leeches to bleed patients was the pinnacle of medical practice. (By the way, when searching for stock photos, I actually found a bunch that suggest there are folks — healers — today who still do this. Eeew!)

Can Stock Photo, Ollyy

Who Is A Quantitative Investor?

Sadly, at present, a quantitative investor is anyone who claims he or she is one, especially if one can lay claim to any one of a number of certifications now available in today’s credential-obsessed world.

But self-pronouncements, job descriptions, degrees, certificates, etc. do not and cannot make one a quantitative investor. The only way to legitimately achieve that stature is to do the following:

1. Be an Investor
2. Apply Quantitative Processes to investing

Countless practitioners meet criteria #2. The ones who deserve the critique offered by Bloomberg are those who are not, and often don’t even try to fulfill criteria #1, to be investors.

Who Is A False Quant

Bloomberg got it right when it panned the practice it refers to as pattern recognition (for what it’s worth, I prefer such labels as data mining, curve fitting, or naive extrapolation). It’s incredibly widely practiced and even revered; see., e.g., the work of Eugene Fama and Kenneth French. 

Wait a minute! Aren’t those the guys who wrote so much about Value, Company Size, Quality, the impact of the Market? Isn’t that stuff real? That doesn’t sound like the work of quacks.

Such factors are, indeed, very much real. But if one merely demonstrates that they are real without understanding WHY they are real, well, that’ not a real quant. That’s merely evidence in support of the proposition that it can be better to be lucky than good.

(Am I unjustifiably lumping Fama and French in with the bad apples? Maybe. I never met them. But I did see a paper they produced dismissing Dividend Yield as a factor that badly turned me off. How could they even have bothered to look into such an obviously absurd hypothesis! Any competent investor knows, without having to stage a big research project, that higher yields exist because share prices are bid down — relative to the level of dividends — due to investor fears that the dividend will be reduced or eliminated, typically due to expected poor company performance. The more reasonable topic for research is whether lower yield (down to and including zero) is the factor to be positively associated with future equity returns given the presumption that low- and zero-yield companies tend to reinvest profits aiming at internally generated growth.)

Who Is A Bona Fide Quant; A Quant Investor?

A bona fide quant is one who acts upon the least quoted but perhaps most poignant passage in James O’Shaughnessy’s What Works on Wall Street: “If there is no sound theoretical, economic, or intuitive, common sense reason for the relationship, it’s most likely a chance occurrence.” Wall Street veteran Marc Chaikin puts it another way: Does it pass “the smell test?” A legitimate quant, upon viewing a test result that seems contrary to rational financial expectations, first reviews the way the variable has been articulated to look for the error(s), and if there are no errors, to look hard for logical explanations.

We’ve had a classic situation along these lines in 2017-18. High valuation ratios, rather than low ratios, worked. Does this mean quant is invalid? No, not at all. It’s about WHY, not WHAT?

Value doesn’t, shouldn’t, never has, and never will work because lower ratios are better than higher ratios. Value works should work, always has worked, and always will work when the ratio, whatever it turns out to be, is misaligned with what it should be given the company’s growth prospects and company-specific business risk (quality). Click here for more on this. When we go beyond Fama-French-type obsessions with WHAT and think about WHY we understand that typical value-oriented factors can point us toward good stocks when expectations are frustrated.

• When highly valued companies from which investors expect much actually deliver, then the stocks will fare well. But when highly valued companies falter, it’s look-out-below for their stocks.

• Conversely when meagerly valued companies from which investors expect little actually turn out to be better, or even less rotten, than initially presumed, their stocks fare well. But when meagerly valued companies turn out to be genuinely awful, then their stocks fare poorly leaving those who were lured by seemingly low ratios to bemoan having been caught by the “value trap.”

Value fared poorly in 2017-18 because this was a period during which companies performed pretty much in line with what the investment community had assumed when pricing the stocks. Satisfied expectations are the bane of every value seeker’s existence. The wealth of empirical data pointing to the historical superiority of low-value ratios really says nothing about the merits of high or low ratios but a lot about Wall Street’s long-term proficiency in formulating expectations about the future. If one believes that Wall Street has, indeed, conquered the prediction problem, then one should be seeing high P/E as a Buy signal and focus further research on how high a P/E should go with how strong a set of fundamentals. If, on the other hand, one thinks Wall Street will revert to the human mean, well, you know. This is the logical basis for assuming the poor performance of quant models (many of which included or were heavily weighted toward Value) in 2018 was an aberration, not a death sentence for quant investing. 

The Theory, The Economic Sense That Legitimizes Quant

Start with the understanding that every stock has an intrinsic value. It’s hard to measure — brutally and inhumanly hard. But the difficulty in measurement does not negate its existence. Instead, it calls for creative thought.

Assume that expression of a stock’s correct value can be summarized as follows:

P = D / (R – G), where 

P = Price

D = Dividend

R = Required Rate of Return

G = Expected rate of Dividend Growth

This is what’s known as the Gordon Dividend Discount Model. It’s just a model, an intellectual roadmap. Nobody should ever use this to plug in numbers and then click to place Buy orders. It’s an Ivory Tower thing, but a good Ivory Tower thing in that it shows us where to start looking for answers.

I’ve made some practical adaptations based on other measures of fundamental wealth. When I substitute E (earnings) for D and then rearrange the equation in a manner that allowed me to pass my High School Algebra Regents exam, I get this:

P/E = 1 / (R – G)

Based on this, and as further explained here, we recognize that . . . 

• As expectations call for greater earnings growth, the proper P/E rises, and/or
• As expected business risk falls (e.g., as company quality rises), the proper P/E rises

Analogous statements can be made with Price/Sales, Price/Cash Flow, Enterprise Value ratios, Price/Book, etc.

Quants that work in a framework like this, whether or not they articulate it this way (actually, nobody I know of says it like this), do “add up.” But this framework is why so many reputable quants tend to look at Value, Quality, and Growth, albeit with wide variations in how each such super-factor is expressed. (in my particular schema, Size, and Momentum, two other widely-cited factors, fall under the Quality and Growth umbrellas respectively).

But That’s Just Words: Where’s The Quant?

First things first: No investor needs to be a quant. It’s OK to use qualitative ideas sprinkled with fundamental information as one likes. The justification for Quant is that it gives one an opportunity to be better by avoiding human biases. (See generally, Chapter 2 of James O’Shaughnessy’s What Works on Wall Street.)

The task of the Quant starts with a process analogous to that of a police detective looking for clues to the identity of a criminal when no living person witnessed the act.

Can Stock Photo, Ostill

All reputable quants approach this in their own respective ways and most typically produce white papers, etc. explaining their respective processes. 

My own approach is idiosyncratic relative to those of many others in that I do not attempt to model the performance of the entire market but instead, look to screen off a tiny portion of it for purposes of portfolio construction. 

A good example of how I work is the Growth (expected future growth) super-factor. Going back to the police detective analogy, this is a phenomenon for which there are no witnesses (because we humans can’t see into the future). So putting on my detective hat, I like to examine three categories of clues:

1. The historical track record: It’s true that past performance does not assure future outcomes, but in real life, or at least in corporate life, change is often evolutionary rather than revolutionary. Although historical growth rates by themselves often don’t suffice, they do contain information that ought not to be completely ignored.
2. Investment Community Sentiment: Yes, we’re dealing with humans who are fallible. But it’s not as if we’re all random dodos. Cynics aside, there are many in the information-rich investment community of today (in contrast to the information deficient historical era that spawned Ben Graham’s iconic and manic-depressive Mr. Market) who are capable for forming intelligent judgments and assumptions. Working with data drawn from the analyst work product allows the quant to factor the Street’s collective judgment into disciplined models.
3. Momentum (including Technical Analysis): No, that’s not a Wall Street obscenity although some act as if it is. It actually goes hand in hand with Sentiment. While Sentiment quantifies what the market says, Momentum quantifies what the market does; in other words, it’s based on the collective judgment of those who put their money where their mouths are. 

Momentum and Sentiment are not direct measures of growth expectations. They’re indirect clues of the “because-they-believe” variety; i.e. the data is what it is because they believe such and such good things about the company’s future. Consider, for example, the Chaikin Money Flow technical indicator. It combines price direction and volume data to come up with a measure of money flowing into or out of a stock. Can you imagine a scenario in which investors don’t have favorable expectations about a company (and Wall Street expectations, one way or another, usually relate to future growth) yet this indicator is strongly positive? I can’t either. Yes, investors can be wrong, but it helps nobody to naively assume that. Certainties are never on the table. But clues (along with the lines of past persistence or wisdom of crowds) can be useful, especially when combined with other kinds of clues (Value, Quality).

OK. So our quant detective has a bunch of clues. But that’s not enough — unless one wants to spend a lot of time with a pen, paper and calculator and a ridiculous amount of 10-K and 10-Q downloads from EDGAR. The quant also needs to function as a language translator, from human verbiage to computer-ese. This is where the numbers come in.

Sample Idea: I’m looking for a company with good growth prospects.

Actionable Translation: I want stocks for which the data shows:

• Positive year-to-year EPS growth in the latest quarter and trailing 12 months
• The consensus EPS estimate for the current fiscal year minus what the estimate was a month ago ranks in the top 25% relative to all stocks in my investable universe
• The stock’s 50-day moving average is above the 200-day moving average

That is by no means the only plausible formulation. Here’s another:

• The stock must be exhibiting a degree of excess performance relative to the S&P 500 that is in the top 25% relative to the investable universe.
• The degree of bullish improvement in the stock’s consensus analyst rating over the past month must be in the top half of the investable universe
• The company’s five-year sales and EPS growth rates must be in the top half of the investable universe
• The company’s 5-year return on equity must not be as low as the bottom 25% of the investable universe — ditto for the debt to equity ratio (I’m looking here to eliminate the sort of low quality that might increase future earnings volatility; notice how risk and growth interact).

Which of these formulations is better? Are either good or are both bad? There are many more I can formulate. 

This is an example of how I test (with test results interpreted thoughtfully — understanding past periods of success and lack, therefore, rather than naive adoption of whatever the test spits out). I’m not testing at random. I’m looking for demonstrably credible clues to the existence of favorable growth expectations which clues will be combined with others relating to value and quality. 

Can Stock Photo, molodec

Neither I nor any other reputable quants are on a treasure hunt for factors that looked good in the past. To do it right, one needs a plan grounded in financial theory.

So This Is The Answer. It Always Works, Right?

I wish. If this could assure me positive outcomes, I wouldn’t be living in New York City. I’d own it!

If black-box pattern-matching rocket scientists slammed by Bloomberg really promised the world perennially good performance, then shame on them (and shame on regulators if any such promises found their way into asset-gathering marketing literature). We’re dealing with the future, something nobody can know. 

Quant investing is not magic. It’s the application of objectivity, discipline, and systemization to a very challenging task and its merit lies, not in any sort of performance guarantees, but in the notion of a sensible process that can better serve investors than can the sort of emotions, biases, hype, fads etc. that plague so many.

If It’s So Hard, Why Not Just Invest Passively In The Whole Market?

Good question. But that’s a topic for another day, and I will get to it.