Investment Epistemology

Julian Sanchez | August 4, 2003

There's an interesting article cited over at PostPolitics on the epistemic roots (Popper v. Galton) of a strategic disagreement between investment contrarian Nassim Taleb and libertarian überspeculator Victor Niederhoffer, who runs (or anyway, ran when I lived there) an unorthodox Manhattan discussion group called Junto. Bonus: the author is Malcolm Gladwell, of Tipping Point fame.

Comments

Alex|8.4.03 @ 1:12AM|#

Taleb's hedge fund, Empirica, aims to profit from"outlier events" (statistically unlikely moves in a given market) rather than absolute direction (as would a conventional fund.) I have never heard Taleb explain adequately how betting on fancy statistical events (what he terms "rare event betting") is more reliable than the Niederhoffer-esque gut-instinct using blunt conventional market instruments.

Basically, the article is all about how luck is confused for skill in trading -- Taleb thinks it's all luck, Niederhoffer obviously thinks it's skill. Taleb surrounds himself with Mahler-loving mathemeticians and waxes philosophical while loading contempt on average fund managers ("middlebrows" as he calls them.) Yet Taleb makes bets just as goofy as any of theirs -- the fact that they are made using complex instruments and statistical strategy doesnt make them any more speculative, nor more likely to make actual cash.

Though in some respects startlingly astute, his "trading theory" sounds a lot like the maundering of a thwarted academic defending his lifelong involvement in a meathead-dominated industry.

Sean|8.4.03 @ 1:17AM|#

Maybe he just really liked the movie "Pi".

SM|8.4.03 @ 2:37AM|#

Does anybody here know how they (Taleb and Niederhoffer) compare to "regular" blokes like Warren Buffet in terms of return on investment, or however else they may be compared.
I am reminded of the fund run by nobel laureate economists and math geniuses which went bust a couple of years ago and had to be rescued by the taxpayer whenever talk of techniques like "stochastic calculus" applied to investing surfaces.
Anybody ?

Alex|8.4.03 @ 3:12AM|#

Niederhoffer's '97 blowout in the S&P is pretty well documented... but only the biggest ones have discoverable rates of return through trade pubs like MAR. I think it says something that he writes columns now instead of managing money..

LTCM notwithstanding, there is SOME value to fancier quant strategies-- I'm just unsure as to why the fancy stuff is exempt from Taleb's beloved efficient markets theory...does Taleb just think the other derivative guys are dumb? And as you suggest, all his strategies seem vulnerable to unquantifiable liquidity and credit risks. Taleb's [public] record is somewhat opaque, as he traded prop for various banks at not at a public fund, its impossible to tell how effective his "fat tails" strategy has worked in practice.

|8.4.03 @ 6:27AM|#

OK, I absolutely loved this article, and I believe I understand it absolutely perfectly - I might not, but I can't know if a black swan really exists until it is proven :). I will, as such, respond to the request and explain it as simply as I can - with the understanding that you'll only get the meat of the matter, not the pleasurable experience of having read it. It's like someone telling you the whole story of a movie - you can easily understand an entire movie in a few moments of explanation, but you don't get any of the pleasure or pain of having seen it.


Adding to, and disagreing with, previous analysis:

First of all, if you look through the article itself I'm quite certain you will see _no mention_ of efficient markets theory, and nor does it have anything whatsoever directly related to what Taleb is talking about or trading according to. Markets don't have to be random for them to be utterly unpredictable - predictability assumes a certain level of individual knowledge, whereas randomness implies an absolute unpredictability of outcome regardless of information.

Further, there is positively no hint of Taleb thinking the other people are dumb, or anything like that - as a matter of fact, I think that is kind of the main point of the article. The failing is due to humanness and a lack of consideration of certain possibilities and theories, not stupidity. I don't see how you could have read the article and got that...so, I'll just assume you didn't. "RTFA", as I have often learned (the hard way). Then again, you could just as well have more knowledge than I do, such as from non-article sources...but then, I can't possibly know that a priori :)

More than anything else, it seems like everyone is missing a big point, of at least the article itself, completely: Taleb's trading startegy is to buy lots of things (options) which have no potential of a massive loss - only a relatively highly likely chance of a small loss, but with a very small chance of a big payoff. To explain this, if you buy options of IBM at 30, where you will make money if IBM stock goes _down_, you will never lose more than 30 dollars per option - it just isn't possible, because stock price can never go effectively lower than $0. However, if you buy the other side of that option your potential losses are effectively infinate (not actually infinate, but if a stock can go as high as $200+ - such as Amazon stock once did - then obviously you stand to lose a whole lot more money than if you bought the other side of the option).

According to my analysis, this works (assuming it works for any reason other than luck, or that it works whatsoever) because of the psychological behavioral 'feature' of humans, in that they (we) would vastly prefer a situation where they (we) won 364 days out of 365, but on one of those days lost terribly big, than if they lost for 364 days and won once but hugely; at least you can be enjoying yourself 364 days of the year with the first strategy, even if it makes you loose more (or win less), whereas in the latter strategy it's just 364 days of constant discust, doubt, fear, and general unpleasantness. Makes me wonder if that's part of the reason why Venture Capitalists (who use something like Taleb's strategy, but in a different incarnation) tend to be perceived as unpleasant.


On a different point, I don't believe Taleb thinks he's found anything "new" - just worth writing about ("new" and "interesting" or "valuable" are not universally interchangable).


On another point, many Amazon reviewers of the book even seem to miss the point - if the point of the article is the same as the point of the book, or at least if the book includes the same point as the article - is that you cannot judge whether or not something is a good strategy (at least in financial markets) merely because someone who is successful (one who succeeded in making good money in the markets) used that strategy. You see, given a sufficiently large amount of coin flipping, you will end up with results which are absolutely astounding to humans - which is because humans simply cannot (or at least do not) 'get' statistics such as that intuitively, and provably so. This is merely one of the many proofs of how humans just don't intuitively 'get' statistics, and it might be absolutely impossible to do so intuitively - you might always have to use formal mathematical tools of statistics to do it reliably.

In financial markets, if the theory correctly applies to financial markets, you can produce Warren Buffets, and his ilk of hugely wealthy traders, merely due to pure, absolute, naked luck - with only a minimum of skill required. And this tells us nothing of that persons intelligence, and provably so - if you don't know the process is a random, or unalterably probabilistic (given limited information), you will likely go to extreme lengths to predict and analyze history and various successes to try and succeed at your own trials, and you will never ever know whether or not it was luck or skill that produced your outcome.

The way to actually test a theory is to have _many_ people try it, keep track of _all_ of them (not just the successes, as you are then falling pray to confirmation bias - you only hear the successes, like in a casino when the slots ring out loudly to announce jackpots, but hear none of the misses), and then see what the chances are of the outcome being due to pure chance. You can never know for absolute certain, but you sure as hell can have a more reasonable degree of certainty using a method like that than emulating successes.

After all, how many people have tried to be like Warren Buffet and failed? Who is going to announce that? "I tried to be like Warren Buffet and I failed miserably!" "That's because you didn't do it right, or you're an idiot - or just damn unlucky!"

But of course if you judge successes as skill and confirmation of the truth and usefulness of your methods, and ignore or explain away the failures, then of course you will absolutely convince yourself of your skill and the validity of your beliefs. It's just so damn easy to do it without even realizing it - one generally requires someone point it out and prove it, and even then, given sufficient incentive, anyone can go right on along believing what they have believed before.

This is not to say that you cannot know, or at least reasonably and reliably come to think, that certain things are due to skill, methodology, mindset, or luck, just that you cannot do it either A) in financial markets, or at least B) through looking only at what successful people do. In other words, blind imitation has it's downsides.

However, I will say that Taleb might be missing a big damn point himself - his strategy only works because not all traders use it, nor use a strategy that would produce the same results. One thing he might be missing is that all sorts of traders use the same sort of technique as he does, in that they accept lots of little losses over time - it's the people who buy options to decrease the risk of massive downsides. His strategy might be a bit unique, in that he uses lots of little losses in the hopes that, once you add in all the little chances of big wins, you'll end up with a big win eventually, but even that he seems it admit is not absolutely fool proof or riskfree - as is stated in the article, there is simply nothing stopping you from bleeding to death (loosing too much money to the "little" losses and not getting enough big wins in time to save your ass); another way to lose is if the volatility of markets decreases in the long term (which absolutely is NOT known - but it IS a theoretical possibility, a "black swan" even, in that while it hasn't so much happened so far, it might actually happen). To support that, if the volatility of markets decreases then you bought options whos price assumed more chances of a big loss than there really is, and so you at least make less money, or you just lose entirely.

What almost everyone seems to be missing is that market trading, of some sort, will always work (it will always produce a profit) because of a seemingly inalterable truth: Different People Value Things Differently. So I might be perfectly willing to pay you $1 to reduce my chance of loosing too much money by 1 out of 1,000, and if you have a different tolerance of pain you can as such make $1. That's it - we just value things differently, at different times and in different places, and that's how real traders (and merchants!) make real money. That's the absolute fundamentals of what wealth and value and money IS, and if you forget it you can come to all kinds of faulty conclusions.


On a final, and somewhat academic, point, the world does not appear to be random on any non-quantum level - it is utterly deterministc. This doesn't mean we can predict all things, in keeping with the Heisenburg Uncertainty Principle, just that unpredictable != random. While true randomness is unpredictable, unpredictability is not neccessarily random.


Oh - and sorry for using the word "point" so many times. I hope it wasn't too painful.

So...point point point point point point point point point point, point, point, and point. There - I feel better now.

Alex|8.4.03 @ 7:41AM|#

Well, it's far from certain Taleb makes money... amassing out of the money options at a discount to market value would be one thing, but taleb[presumably] pays market price. Humans might not understand distributions, but option market makers surely do; since their prices incorporate the likelihood of outlier events, Im not sure Taleb isn't simply betting red just like every other schmoe at the roulette wheel. and in zero-sum games like derivatives, only one side will ever make money on every last trade.

Alex|8.4.03 @ 7:53AM|#

Re: efficient markets .... the entire BlackScholes methodolgy presupposes a random walk of the underlier... anyone using a model with a stochastic variable (all option models) is directly applying random walk theory.

Hovig John Heghinian|8.4.03 @ 8:02AM|#

Forgive me, but I think I can boil this article down to one line: Taleb is basically a glorified short-trader.

P.S. To "Confused" at 12:34 pm - The "Perpendicular Pronoun" is the ninth letter of the alphabet. (Think about it.)

Alex|8.4.03 @ 10:58AM|#

Taleb is a bright guy, but he sure does love the 'perpendicular pronoun.' And beneath the layers and layers of intellectual frosting and inscrutable stochastic calculus, Taleb's strategy seems no less speculative than Niederhoffer's.

|8.4.03 @ 10:58AM|#

Hovig - ...actually, yes. That pretty much boils it down to one line.

Good show, that. Really. :o

confused|8.4.03 @ 12:34PM|#

ok, both the original posting and the first comment make absolutely no sense to me. Can we rewrite this in layman's terms?

Douglas Fletcher|8.5.03 @ 6:25AM|#

God I'm glad I dropped that class in college.