What the Quants did next

Hey Bento, remember these guys? Interesting article here on the “new realms of science” the Quants are apparently pushing into in order to make a fast buck in stocks without looking at balance sheets or in fact doing any actual work. Mean-reverting Gaussians no more, apparently all the people who survived the Quant blowups of last August (this is what Baruch wrote about them at the time) have now moved on things like “machine learning”, “mathematical linguistics” and “agent simulations”. At least one of them is still working on a grand unified theory of finance! Good luck with that, I say.

I do think some of these ideas sound like they could be amusingly dangerous – “reinforcement” strategies sound well dodgy, for one; is it that you keep throwing more and more money into a winning trade as it keeps on winning? A signed copy of the Ethics to the first commentor who can say why that may not be a good idea!

Of all the strategies in the article I think those that use some degree of behavioural finance sound the most interesting, especially if coupled with what one Dmitri Sogoloff is talking about:

Sogoloff is wary of quants who believe the real world is obliged to conform to a mathematical model. He acknowledges the difficulty of applying scientific disciplines like genetics or chaos theory — which purports to find patterns in seemingly random data — to finance. “Quantitative work will be much more rewarding to the scientist if one concentrates on those theories or areas that attempt to describe nonstable relationships,” he says.

Sogoloff sees promise in disciplines that deal with causal relationships rather than historical ones — like mathematical linguistics, which uses models to analyze the structure of language. “These sciences did not exist five or ten years ago,” he says. “They became possible because of humongous computational improvements.”

But it does all sound dreadfully difficult. A lot of what the Quants seem to be trying now is simply tinkering around to try and find something which makes pots of money guaranteed with no risk, and unsurprisingly they haven’t found it yet. So they apply the old strategies to new markets, like commodities. Let’s see how long that works. Lots of them seem to think running different strategies simultaneously is an end in itself, much like Captain Picard found that if he timed his phasers to fire at the Borg using random frequencies it took the collective mind time to adjust the defenses to block all similar attacks, and you could generally nobble a few of them in the interim. Almost all of them (the Quants, not the Borg), however, seem to accept lower returns from their new strategies than they were used to getting with the old Gaussian ones plus leverage. Worryingly, the article never addresses the topic of leverage at all.

The thing though that all of them seem to be trying to escape from is the essentially unquantifiable nature of the market, unquantifiable in the sense that the data changes on observation. This would be true of even insights gained by behavioural finance, I imagine. Put simply, the moment I come up with the Grand Unified Finance Theory I invalidate it – my understanding of the theory, trying to make money off it by gaming the system, creates the conditions whereby I myself act in violation of its laws by becoming aware of them. And if the great unwashed (or rather, the scrubbed-clean white teeth brigade of the hedge fund hoi polloi) ever get hold of the formula it would be back to the drawing board for sure.


8 thoughts on “What the Quants did next”

  1. The advantage of quants is that they’re secretly humble. Almost everyone believes that, given enough information, they could tell you that Ford will outperform GM; quants are just looking at historical relationships between variables, so when enough people are super-confident in Ford, they’ll bet that it will converge with GM at some point in the future. This takes care of a lot of the biases that cause people to over-extrapolate from recent data, discard information that hurts their thesis, etc. Of course, quants will be blindsided when Ford really is doing better than GM, for historically unprecedented periods of time — but it would be hard to imagine a strategy whose detractors wouldn’t be vindicated at least some of the time.

    “David Shaw, who came from Morgan Stanley, where he worked under renowned trader Nunzio Tartaglia. A onetime Jesuit with a Ph.D. in astrophysics”

    I can almost guarantee that returns in any financial specialty peak when companies are being run by folks with really strange resumes. Once the ex-Jesuit PhDs get replaced by fresh MBAs, it all gets hopelessly conventional; 11% returns, plus or minus the occasional disaster.

  2. Thomas: but once the “machine” “learns” to do the one thing which works best, how can it start doing any other thing until that thing stops working? All of the capital deployed by that machine will therefore be put into that strategy, and if it does work, the capital will grow.

    By default then it will become a strategy that scales up, unless the machine has a built in “all right that’ s enough mate, I’m bored let’s think of something else” function.

    Byrne, I hear and concur with your thoughts on fresh MBAs. I was partly alluding to them with my dismissive remarks on the “white teeth brigade”. More lapsed jesuits please.

    But the problem with the mean reverting strategies is that pace Taleb, while mean reversion holds true most of the time, the times when it doesn’t, when as you put it “the detractors are vindicated”, tend to be so awfully dynamic that all at once you can lose more money than you made over the period when the means were still reverting. You have an asymmetric payoff, which will make you much more likely over a long period to blow up and lose all your money, especially if you play with leverage. That is the problem the quants faced in August. Humility was nothing to do with it.

  3. The subjective valuations of individual human participants are the cause of market movements, not mathematical relationships tortured out of the data detritus of the past.

    When they can predict how I’m going to spend my time at 12:57 pm next Saturday they may be worth watching. Otherwise the best technology for predicting human behavior remains another human mind. Data and its manipulation into cognitively meaningful form is important to calibrate and inform those predictions but without incorporating human input the clock is ticking on the next blowup.

    Applying techniques from physical science is complete folly. Hit a rock with a hammer ten times and it will behave very similarly each time; try to hit a human with a hammer and your results will vary (along with your lifespan).

  4. Baruch, have you read about Jim Simons’ early career?

    “In 1958, Simons and Mayer had celebrated their graduation by buying Lambretta motor scooters and driving to Bogota from Boston. In 1964, the three cobbled together money with Simons’s father to start a Colombian vinyl-floor-tile factory.” (http://www.bloomberg.com/apps/news?pid=newsarchive&sid=ayjImYcoCiH8).

    D. E. Shaw tries to find people with those weird resumes — they have lots of published authors, former poker champions, and the like.

    I don’t mean to imply that they set out to be humble — it’s quite arrogant to have an opinion on thousands of investment products when it’s so hard to predict the performance of even one. I just meant that quants ignore the evidence that other people overweight, they accidentally earn superior returns. Unless their models happen to mimic mania, they are less likely to be taken in by hip new technologies or awesome biotech stories, and are similarly unlikely to except that America’s industrial base will disappear any time soon. By getting rid of qualitative concerns, they get rid of the natural tendency to extrapolate heavily from qualitative data, and to overweight the value of judgments about which drugs will take off, which chips will sell well, etc.

  5. Baruch,

    I share your suspicions on the potential dangers of a “7-minute abs” approach to specu-trage. I amusingly recall Louis Navellier in one of his more charlatanical moments (after getting hammered in the tech wreck) saying something like “I’m not a growth investor or a value investor but a do-what-works investor”. Shit! Why didn’t I think of that??!?! Undoubtedly, what has (more recently) worked are mimetic strategies, which themselves are caught up in a lovely recursive positive feedback loop that has attracted yet more capital to them at the expense of the mean-reverting Gaussians (now resembling Lamarckians if P&L is any measure?) , such that front-running, order-sniffing, and trend-following oriented pursuits garner a greater relative percentage of turnover.

    Now, basking in their victory, such feedback traders, armed with more capital while market-makers and block-traders longer facilitate, and with mean-reverting Gaussians in the penalty box, or retraining for a mid-life career move, things will begin to get really interesting as the feedback traders duke it amongst themselves in an algorithmic fight to death…(or at least a pyrrhic duel to zero-alpha).

  6. cassandra: YES! that is it, you put your finger on my objection to Quants — 7 minute abs. They are trying to find some way of getting away with a doss. Yes, they are working their arses off right now, but only in the hope they find some grail which will mean they can go back to playing D&D while the money rolls in.

    Byrne, sadly I am stuck trying to work out the potential of “hip new technologies” (interestingly I saw a company that has new technology to make hips not so long ago). I am susceptible to manias, but don’t worry overly about the US’ indistrial base.

    Sadly for the Quants however, they are susceptible to “style manias” and humble or not will bet billions in capital on algorithms, rather than individual stocks — in that sense they are not so different from me, but they may not be as diversified. . .

  7. Dear Baruch

    I draw your attention to the following article in the Scientific American which you may or may not have read. In any event your readers may find it interesting. The gist being:

    “The strategy the economists used was as simple as it was absurd—they substituted economic variables for physical ones. Utility (a measure of economic well-being) took the place of energy; the sum of utility and expenditure replaced potential and kinetic energy. A number of well-known mathematicians and physicists told the economists that there was absolutely no basis for making these substitutions. But the economists ignored such criticisms and proceeded to claim that they had transformed their field of study into a rigorously mathematical scientific discipline.”

    The full article is here:

    I leave you with a fitting reminder from my hero:

    ‘Science as it exits today is agreeable through the power which it gives us of manipulating our environment and to a small but important minority because it affords intellectual satisfaction. It is disagreeable because, however we may seek to disguise the fact, it assumes a determinism which involves, theoretically, the power of predicting human actions. In this respect it seems to lessen human power.’

Comments are closed.