Summary of a small disaster

Discussion in 'Automated Trading' started by travis, Feb 19, 2009.

  1. travis


    I am going to write it here for future memory and reference, and in case someone would like to share similar experiences.

    I thought I had it all figured out. I had built about 20 different trading systems, that I thought were quite different among each other, all of them triggered trades that lasted a few hours and all of them about 60% of wins.

    Starting in November, I went from 8000 dollars to 26000 dollars in one month. Then, at the end of November, I took out 8000 dollars to repay the loan from the bank, and had 18000 left. Then I went up and down for two months, but always in the range between 13000 and 18000. Then in the past few weeks, also by adding a few losing discretionary trades, I went from 18000 down to 3000. Now I am being patient, very very careful, but obviously I am pretty much screwed since I have such a low capital.

    Nothing else, just wanted to share this terrible experience. I thought with so many systems and about 18000 dollars, I would be ok, with no risk of big drawdowns. Instead here I am, with 3000. It all happened in just 2 weeks.
  2. bidask


    how many trades do you amek a month with those 20 systems?
  3. chvid


    Sounds like poor money management and overleverage.
  4. travis


    Yeah, poor money management and poor everything, no doubt. Each system makes about one trade per day, on average. But of course with 18k I could only trade 5 or 6 systems per day. And now I can trade only a couple of systems per day (but none overnight). We'll see.
  5. Every systems will go through a period of major drawn down, if this drawn down still inline with historical tested behavior, then stick to system rules and those systems will recover in the long run.
  6. Corey


    How much correlation (more importantly, covariance) analysis did you do between strategies?

    In my eyes, there are three primary ways to create diversification: time frame, strategy, and instrument.

    Mean-reversion trading currency over 5 minute intervals should be non-correlated with trend-following equities over multiple weeks.

    Go through each strategy and try to 'box' this diversification.

    If all your strategies run on the same instrument over the same time-frame, you may be more correlated than you think...
  7. travis


    The systems all have the same time frames and same concepts (one method for intraday and one method for overnight trading), but they trade 9 different futures (these two methods I use works equally well on all such markets).

    I am not skilled enough to calculate covariance and similar concepts.

    Also, for the question before, the systems didn't do worse than in backtesting. I lost money and almost went broke because of having so little capital I guess. The problem is that being so undercapitalized I could not trade 20 but only about 10 systems at once and I also had to hope that they didn't all fail at once, which to some extent they did, and it happened more easily because I was only trading 10 and not 20, and then 9, 8, 7... and now only 2.

    I may or may not have made a mistake in repaying back my debt when I had 26k (made from just 8k in one month). With that capital, I was able to trade all 20 systems, but a friend insisted that I repay my debt so to not regret it later, and he convinced me. In some sense it was wise, in other another sense it wasn't. I don't know what would have happened otherwise. Maybe I'd be broke, but with a debt of 8k. Or maybe I'd have 50k.
  8. That should be relatively easy to figure out, just look at what your systems would have done had you kept the 8K.

    If they preformed similarly to what you backtested, it sounds like you were under capitalized and just hoping that you would have a really good period.

    Back-testing isn't perfect, you should always expect your biggest drawdown to be in the future. It actually sounds like you have some decent systems, you should figure out how much you need to survive the worst possible drawdown (and then some) and still be able to implement the systems and not trade them again until you have accumulated that much capital.

    Sorry to hear about your experience.

  9. travis


    Thanks all, for the information and the advice.
  10. bespoke


    I think you have to ask yourself this - do your systems actually work? How long have they been tested for? Years? 1 month of positive results is not what I could consider a working strategy.

    I bet most system traders only have 1 or 2 profitable systems. If we had 20 we'd be laughing all the way to the bank everyday.

    The market is much different in November than it is today as well. For a dull market like it is this year you should test back a few years.

    It isn't much capital you've got to work with but anything is possible. You did it before, you can do it again. Good luck.

    edit: did you mean 2 systems across 10 different instruments?
    #10     Feb 19, 2009