My grandfather was a gambler.
Nice. Smart. But he thought he was smarter than everybody else in the room.
I haven't gambled much, as a direct consequence of what happened decades before I was born. Going to the bookies is popular in Ireland but unheard of in our family.
I started playing Poker when I moved to New York. When I look across the poker table, I just see the backs of cards. I can't spot tells. Perhaps the risk analyst in me respects chance too much (not good for poker) there are patterns in the noise but I don't try to parse them. I am not smarter than the cards.
My grandfather however believed in hidden order - even after he couldn't gamble anymore he spotted patterns in the Irish lottery (turned out he was right on that one!).
Yesterday, I trashed one of my beloved strategies. 'Skewerage' predicted when to go long and short based on the movements of the CBOE's skew index.
It looked decent, but very small changes in initial conditions causes its performance to drop significantly.
Now, another for the rubbish heap.
A while back I spotted that, ironically, momentum indicies generate a lot of mean reversion. And mean reversion strategies applied to them seemed profitable.
But alas, when I calculate Big 'O' Sharpe (simply, start the backtests from different days and report the lowest result) the Sharpe ratios drop significantly; which means the results I reported previously were 'lucky'.
Specifically, testing on the S&P 500 Tilt over the last 6 years, has the originally tested Sharpe of ~1 dropping to 0.35 with Big 'O' Sharpe.
In another example, the 'Lucky' Sharpe Trajectory shorts early just as iShare Momentum (MTUM) dips and also early in 2014. 'Unlucky' bets big on a drop when MTUM rises. The decision making process is exactly the same, just spaced 10 days apart.
Lucky has a Sharpe of ~2, and Unlucky 0.5 (i.e. the actual Big 'O' Sharpe figure).
Not only that, but the strategies do worse than the original momentum indicies to begin with (Sharpe of 1).
There's still a gold nugget to be extracted from momentum indicies' generation of mean reversion, but we can safely nix my first two attempts!
Big 'O' Sharpe goes a long way to removing the noise and revealing reliable patterns. Next up Steady Volatility.
The thing I took from playing poker is how unforgiving chance is. Your opponents make mistakes but the cards don't. You might be the smartest guy in the room but the cards are inscrutable.
Same with markets, which is why eschewing luck is important.