I have two Ph.D.: one in Power Electronics (1979) and the other one in Artificial Intelligence and Cybernetics (1985). I also have a master's degree in Mathematical Psychology (1984).
Statistical Arbitrage (SA) is based on finding Stable Distributions and capitalizing on them. Placing bets on Fat Tails, for example, is one of the methods of SA. It is quit opposite to the mean reversion, wouldnât you say? On another hand, mean reversion is one of the patterns that one might use in SA. In this respect you are correct.
Our average holding period varies. It could be as short as a few hours and as long as 2 months. The average, I would say, is 20 days.
Isn't arbitrage between the S&P cash and futures one of the "safest" forms of arbitrage ? What other forms of arbitrage are more risky ?
Yes it is. The spread between the Base and its Derivative is always the safest, but in the same time you are only making the interest rate on your cash over the time that is left in your future contract. If you want to make more than that you need to use other forms of SA such as Options Constructs. Constructing different synthetics out of variety of options gives you a better environment to do SA.
One can't assume away extremes as outliers. Not in the financial markets. Of course, that's just my opinion. I could be wrong.
I am not sure what you mean. I was talking about the skew in the Stable Distribution pattern that has more probability to produce outliers than a Gaussian distribution.
I am another one of those quants that has been around for nearly 20 years! I trade Nasdaq100 futures primarily but with my intraday System have been able to gain more than 20,000 points since 2001 with <60pts maximum drawdown.. on a $10,000 account! [by the way trading no more than 3 to 5 contracts max -both long +short -NOTHING held overnight!] I no longer manage money for others because, quite frankly, I don't need the headache of meeting all the myriad regulatory requirements. Besides, I like my privacy! My approach is derived from thermodynamics.. create a boundary condition problem, determine key inputs/forces -the "true drivers" -acting on the environment (they are very specific but variable in magnitude). I consider rumors, news, info/reports of any type to be like a "mist" that can have an effect on the "true drivers" of a market: Buying +Selling! How the "mist" will affect the "true drivers" is UNKNOWN! As a scientist I can measure these "true drivers" and create a series of "tools" which when assembled properly can reliably predict a specific pattern consistently over time [there are many different patterns -hence multiple "Models" w/n the "System"]. Imagine a box as the MARKET with a general "mist" surrounding it. There are 2 vectors (buyers, sellers) acting on the box with 5 specific measures of these vectors: (!) Price, (2) Time, (3) a f(volume(weighted)), (4) ProprietaryA +(5) ProprietaryB.. the tools I develop are derived from these 5 measures. I must still manage my capital properly.. for this we can turn to the methods poker players employ when they are winning vs. when they are losing. I aggressively bet when I have the best of it and pare back when losses build. Most importantly, the MARKET determines my every move. I do not fear market turbulence.. I plan for it, expect it +THRIVE on it! This is what we do, isn't it.. we are all traders! We live on the proper MANAGEMENT OF RISK!! We must understand the methods we employ +their limitations.. otherwise any of us could fall victim to an outlier event.. a DISCONTINUITY, in the parlance of the mathematician.. Mh