Friday 28 November 2008

System Development Update

I just found some time to take a closer look at David Aronson's excellent book Evidence-Based Technical Analysis and found some important things to consider about my trading setups based on the weekly Commitments of Traders reports. One is testing the setups based on detrended price data. My results and choice of best setups are based on the raw price data, but it's clear from Aronson's book that detrending the data is an important step. I tried that last night on the best few dozen setups I've found for the S&P 500 and can report that most of them still produce excellent results - on balance, even better than on the raw data. The problem is the results are different from the ones I used to choose my best setup for trading. So I need to look back at all of them to find the one I like best. Same for all the markets I'm trading.

The second major step Aronson talks about is a bootstrap or Monte Carlo test in order to verify if the setup was obtained by luck. Aronson offers code for a variation of the Monte Carlo test on his website here. I think he makes an excellent argument for the importance of this kind of testing, and I plan to do that too.

The question is where does all this leave my existing setups. Until I've done all this testing, I don't think I have much choice but to include a bit of discretionary technical analysis in the picture when I'm trading these signals. I might, for example, consider trading some signals with a tight stop - for example, a two-percent drawdown stop. Trading the system until now the way I've done has been an important education on the value and challenges of mechanical trading - more so than any paper trading could have been. And I think the signals still have great value in guiding trading decisions. But until I go through the important processes above, I feel like I'm taking on too much risk. On the other hand, the tests above will hopefully lead me to a super-robust trading system. Until then, I will still post the signals weekly and give my take on what the data is saying, as well as give updates on my system development progress. See you later today or sometime this weekend with an update on this afternoon's COT data.

2 comments:

Anonymous said...

Alex another thing you may wish to look at the use of CFTC COT supplemental reports which removes the swap dealers contributions from the true commercial positions.

best regrds, Ben.

Alex Roslin said...

Hi Ben - thanks for your message. Those reports do have interesting information, but not enough data historically to have much usefulness for trading yet.

Regards,
Alex