I just finished reading Nassim Taleb's treatise on the nature of randomness, Fooled by Randomness
. In it, Taleb argues that, while many in the financial industries accept that the market is random, their approach to working in it completely ignores this. A central theme of the book is how randomn events are frequently mistaken for skill. Taleb highlights various patterns that occur when people do not appreciate the randomness inherent in the market.
The Black Swan Problem
One of the central themes in the book is how people deal with (or characterize) rare events. The Black Swan problem is as follows:
- No number of observed white swans will prove the assertion that all swans are white
- The observation of a single black swan will be sufficient to disprove the assertion
Mutual funds usually brush this under the rug with the "Past performance is no guarantee of future performance" (wink, wink, nudge nudge - look how well we did last year!) but the Tech crash of 2001 would be a good example of a black swan.
Survivor bias is introduced as a way of explaining why traders, in Taleb's world, who have had several good years may appear to be successful but are really just lucky since the vast number of trader's who fail are no longer in the business. Given a set of N equally competent traders in a random market a certain percentage will have years of success, many will have some good years and some bad and others will be all bad. How many of each depends on the size of the original set. Survivor bias indicates that may of the lower performing cases will be pruned leaving only the "best". However in this case, the "best" only arises from statistical distribution rather than skill. If the set of N is small then the successful traders might truly be skillful but it is not, hundreds of thousands of people working in the security industry and millions have washed out of it. The same principal applies to mutual funds - given the large number of funds available today, it should not be suprising that some have had many good years but it is not necessarily as a result of skill.
Taleb notes that people typically attribute their success to skill and their failures to bad luck or rare events. In addition to the actual outcome, there are (potentially infinite) alternative outcomes or histories with an associated cost/benefit associated with them. Of course, to have real value, this analysis should be done prior to committing to a course of action. But analyzing decisions that have been made is a good way to keep yourself humble.
Taleb uses the term Asymmetric Odds to describe distributions where the probability of each outcome is not equivelent. He notes that people tend to focus only on the probability and not (correctly) on the product of the probability and the expected return from the event.
The Rare Event Fallacy
Much of the book deals with the "Rare Event" - it is common to call these "Unforeseen events" either because they have no historical precident and have not been considered or because they are are simply ignored. Because of this reticence to deal with rare events, Taleb trades on the basis that they are not fairly valued and the rarer the event, the less fairly valued it is. The idea is that the frequency of the event is not the only factor - the product of the event frequency and the magnitude of result that event do. Taleb is content to lost small amounts of money over a long period of time provided the occurrence of a rare event will be sufficiently profitable. Taleb higlights the blow up of LTCM which used trading models developed by Nobel Laureates but failed spectacularily when the Russian banks failed. It should be noted that one can only afford to lose money over a definite period of time (the human life span at the very most) but it seems that the sort of events that Taleb is capitalizing on occur on the order of a decade.
Finding patterns in randomness
Taleb observes that humans are geared to finding patterns in noise regardless of whether there is a pattern to be found. Our brains seem to be better geared for dealing with thing analytically than statistically. This leads to people confusing correlation with causality - a phenomenon that anyone who reads of the latest medical discovery is certainly familiar with. In Taleb's world, the most common example of this is traders who develop trading strategies that are designed to work well on historical data and then trading on that basis. True randomness means that coincidences will occur and this has to be taken into account. Simple put, people are not built to deal with probabilities and to act rationally means ignoring the emotional or gut reaction. Taleb describes some measures he goes through to prevent his emotions from second-guessing - watching CNBC but with the sounds muted, not looking at his daily wins / loses. He feels that people will only focus on the noise (which plays out over the short term) and the signal which occurs over a longer time period.
The Sandpile Effect
Another failing of the human mind is that, if left unchecked, people will assume that all relationships are linear. The simple fact is that most things in life are non-linear or appear linear over some restricted domain but are not. Taleb focuses on a model (Polya processes) where the odds of winning/losing is non-linear and increase with the amount of winning/losing. The statistical noise that picks the winners under the Survivor bias becomes much more important in this situation where winning early on causes an insurmountable advantage over other competitors. This particular non-linearity magnifies random noise to the point where it is more important than any signal (for example, skill).
I really enjoyed reading the book. Many of the ideas in it were familiar to me from studying physics but there was a fair amount that was new or presented in a new way to keep me reading. The book uses various people (seemingly people Taleb has met in his career) as archetypes of various ways of thinking. Of course, Taleb himself is the archetype of the rational investor taking randomness into account - the book does come off as arogant in places - but there is enough self deprication in it to suggest he doesn't take it too seriously. The net effect of the book was to probably dissuade me from trading stocks the way I did in the late 90's and at least to behave more rationally about investment (for example, if you wouldn't buy a stock at the current price, you should sell it - easy to say, hard to do). I most definitely will not be trading options waiting for the next stock market collapse as Taleb is now doing - I read the book out of interest and to learn something. I suspect the book will be read on Wall street and largely ignored because it requires too much discipline and most investors have short time scales depsite what they may say to the contrary. Taking small loses for an extended period of time, waiting for a rare event with a large payoff may strike some as being akin to a lottery ticket. On the other hand, the only difference between the venture capital industry and a lottery ticket is the odds.