I agree completely. I believe it's possible to develop a system, then distribute it either to improve one's winnings or to fade the effect of others trading it. I know two traders who license signals to institutions, and have had the problem of a few of their clients trading too much size and spoiling the system for other clients. This makes an excellent case for keeping systems private or restricting usage.
I would just like to hear one story, just one example, of a secret mechanical system that was pulling in cash, was revealed (or better yet, discovered by a "keen market participant"), and then had its entries erode to the point where it was no longer useful.
Hi T-Rex, I'd like to add my two cents worth of knowledge about common known systems - In the Journal of Finance in 1992, Brock, Laknishok and LeBaron published simple trading rules for the stock market which had been working from 1897 til 1986. Simply put, they had discovered the 200day-moving-average, buy when above, short when below. In the Journal of Finance 1997 Sullivan, Timmermann and Whie re-tested those rules for the period 1987 til 1996 - the results had been turned around. Out-of-sample the rules didn't hold up, the edge in predicting direction was gone only the second momentum (volatility) was kept intact. The distortions and the stress put to the rules in 1987 til 1996 was bigger than any-time in the previous 100 years. Data-Snooping, Luck or knowledge of the rules? Regards, Oliver
Good question. I know a trader who did very well in the late 90âs through â01 trading extreme daily deviations of pairs like QQQ and INTC. Doesnât work so well now. You might say the market has changed since then. The participants are more sophisticated. Systems are proliferating. Arbs are using more automation to find opportunities. Mutual funds and large players are more careful now with their entries and exits to avoid getting chewed up by traders. The public is more risk adverse, and more and more are dabbling into short positions... and getting squeezed. I have some of my own âsecret mechanical systemsâ that stopped working. How does one tell if the âmarket changedâ or others discovered my system? Probably a combination of both. Maybe itâs the same phenomenon. How can you tell? Lets go ask every trader what system they are trading.
First I see what you are saying now about purely mechanical systems-- if everyone tries to do exactly the same thing, on the same instrument, the entries erode. Big discretionary funds face the same problem, that's why they set up fake order desks to blow smoke. A big poker game. Now to your experience with your systems. It should be easy to tell what happened by looking at the entries over time. Did they deteriorate?
Hi Prophet, when I do a system, I do a 10-year-Monte-Carlo-Simulation of expected MaxDrawDowns. I draw and put back into the sample daily-returns of my single market / portfolio equity-curve. By drawing 260 returns for a year and standardizing the length of the MC to 10 years, I get comparable numbers for what to expect and how much draw down to sustain. As we all know, MDD increases over time and this fact is reflected in the MC. If a market takes out my 95% confidence interval of MDDs, I start paying attention and when the 97% interval is hit, I stop the system. Neither the 95% nor 97% case has happened since 1998 but you never know
Yes. The purity of the equity charts deteriorated over time, not uniformly, but in an unpredictable way. Risk increased. Perhaps this is because the market "discovered" related patterns/systems, or because my systems simply weren't robust enough, or had overfit the past character of market. It's hard to tell. It's not like other participants suddenly discover a new system and start trading it with large size exactly as it was traded in private. It's very hard to pin the blame, except that the system is not good anymore in general.
Very interesting Olintner, May I ask you what kind of systems you test; I mean daytrading ES or so; how many trades average a day? Thank you for commenting, nononsense