I am reading Critical Mass, Philip Ball. It talks about patterns in nature and in market and how they are scientifically linked. The author does a tremendous job showcasing order in a presupposed random world. In one of the chapters the author illustrates how financial crashes and statistical mechanics overlap.
The log periodic behaviors in markets have a distinctive signature. The system is prone to oscillatory periodic fluctuations. In an economic context this would be analogous to periodic business cycles. But log-periodic variations are not like the regular oscillations of a light way or a turning fork. Instead the peaks and troughs of the waves get closer together. At the critical point itself they pile up on top of one another.
“French mathematical physicist Didier Sornette, based at the University of California at Los Angeles has research extensively on the subject and is convinced that market crashes are log periodic.
Despite much proof Sornette admits with some chagrin that trying to make a genuine prediction of a crash is a thankless task. There are, he says, at least three possible outcomes.
1: No one believes the prediction, but the market crashes anyway. Then critics will say it is a just a lone, lucky correlation with no statistical significance. Besides, what good is a warning if it fails to avert a crash. 2: Many investors believe the prediction, get triggered into panic buying a selling, and thereby cause a crash - that is, the prediction becomes self fulfilling. 3: Many investors believe the prediction and take careful compensatory action so that a crash is averted - that is, the prediction becomes self defeating.
That's the problem with the dream of predictability in economics: future market behavior depends on what traders and investors believe that behavior will be, so the act of predicting the future (if it is taken at all seriously) is likely to change it.”
I agree with Sornette, when it comes to the lucky correlation, most technicians must have experienced that at some point of time in their life. The second point Sornette gives a causal explanation of how internal market dynamics causes non equilibrium (or crash) sate. The third aspect is Sornette's way of escaping a bad forecast, as much before Sornette and log periodic behavior, Benner laid down his predictive tool. The clock predicted most lows and highs on popular averages 100 years after its construction. Now as we head into Benner 2011 lows, the forecast would become hindcast months from now. It would be interesting to see Benner going wrong here in 2011. In any case, right or wrong, we would not elevate Benner to a statistician or thinker who got it wrong. He would remain a part of an archaic history of counter-intuitive cycles.