Monday, May 10, 2010

Stochastic and epistemic uncertainty

Are there two kinds of uncertainty, one a property of systems and the other a property of our knowledge of systems? The first, known alternatively as stochastic uncertainty, aleatory uncertainty, and randomness, resides in the real world and is irreducible. The second, epistemic uncertainty, is a function of our lack of knowledge about the world and is reducible by acquiring more or better information.

Terje Aven in Misconceptions of Risk takes an unabashedly idealist position, contending that all uncertainties are epistemic. He quotes R. L. Winkler: “Consider the tossing of a coin. If we all agree that the coin is fair, then we would agree that the probability that it lands heads the next time it is tossed is one-half. At first glance, our uncertainty about how it lands might be thought of as aleatory, or irreducible. Suppose, however, that the set of available information changes. In principle, if we knew all of the conditions surrounding the toss (the initial side facing up; the height, initial velocity, and angle of the coin; the wind; the nature of the surface on which the coin will land; and so on), we could use the laws of physics to predict with certainty or near certainty whether the coin will land heads or tails. Thus, in principle, the uncertainty is not irreducible, but is a function of the state of knowledge (and hence is epistemic).” (p. 146)

How should we understand market uncertainty? And how does the model that we choose inform our trading strategies?

If we assume that markets move randomly or at least with enough randomness that we cannot predict a single sequence or do better than 50-50 calling a series of sequences, we should abandon all attempts to gain an edge with our entries. No amount of work on our part will improve our entries over those of the dart thrower. Either we buy an index fund and trust that over time markets will rise or we become really savvy risk managers, protecting our assets while steering our trades toward gains and counting on that inevitable outlier somewhere down the pike.

If we assume that market uncertainty is epistemic, we can gain an edge by increasing our knowledge of market movements. Analysts churn out reports on individual companies, economists provide data and every kind of opinion you’d ever want to hear, chartists study how the past can portend the future. There are enough pockets of success for us to believe that market uncertainty is at least in part epistemic and that knowledge can make a real difference.

Beginners are true believers in epistemic uncertainty; professionals tend to have more respect for the randomness of markets. As Don Kaufman said in a recent Think or Swim chat on the Greeks and expiration (one that I highly recommend for anyone with even the most marginal interest in options), “Most retail traders are good at finding trades, bad at managing trades. Most professional traders are good at managing risk, less good at finding trades.”

Even though I believe that market uncertainty is at least in part epistemic, I think we would all be well served to work under the hypothesis that market uncertainty is stochastic. That hypothesis forces us to manage risk in ways that can be both imaginative and effective. Whatever we earn over and above that from our stock picking skills or whatever other means we use to enter a position is pure gravy!

No comments:

Post a Comment