System drawdown

Discussion in 'Strategy Development' started by clarodina, Jun 17, 2008.

  1. Hi all veteran system traders,

    How would u decide whether a system drawdown is another normal equity dip or busting? And How do u cope with the drawdown? Ride them out, reoptimise, cease trading...etc?
     
  2. Corey

    Corey

    First of all, back-testing over a long (10-20 year) period with many, many trades (tens of thousands) would give me a good idea as to what the distribution of my results should look like -- so that I can determine whether any draw down is within reason.

    Then, I would look at the trade logs, identify why your system decided the trade had the proper risk/reward characteristics, and see if those metrics are still appropriate in the current market.

    Without knowing what your system is, or how it works, it is hard to say...
     
  3. Paper trade till you have a strategy with a high probability of success, i.e,. 80% or greater. Then begin trading only 100 share lots till you are satisfied that it is working.

    Draw downs are for suckers.
     
  4. I$land

    I$land

    "Draw downs are for suckers."

    Are you saying you don't have any drawdowns ?
     
  5. eagle

    eagle

    Excellent question. One who can effectively identify the range of wave (some call it as noise) will be very very rich. It depends for every stock. I have noticed that NASD stocks are very volatile than NYSE stocks and therefore drawdown for NASD stocks must be set lower (greater negative number) than NYSE stocks. When it comes to a particular stock, even within the same exchange or industry, I believe its wave length is different as well. Screen time is the best method for trying to identify what is its normal wave (noise) and outside that range might be a new trend.
     
  6. hausse

    hausse

    I can tell you from experience long-term trading a portfolio of many commodities and other futures that drawdowns do increase over time; seems to be a normal occurence. They have especially done so during the last couple of years with the increase in volatility in many markets. This lead to a deterioration of the MAR but returns are still healthy. The ride became rougher.

    Whether a system is in a normal drawdown or goes bust I don't think one can tell in advance. There is no security. I have yet to see a system that goes bust for good. Eventually they all make it back but it can take longer than the user can stand. Would like to know a system that dependably loses money.
     
  7. Don't really agree that system eventual return profitably before going negative. If the edge is no longer there probably the way is down to negative value in equity if the trader keeps trading the system. The thing is often difficult if not impossible to know whether the edge is temp out or persistent. Do u trade the system even your equity down below 50% in dd? How do u decide whether your system edge is no longer there while in dd?
     
  8. hausse

    hausse

    I can't say anything about day trading systems. I do long-term trend following, a different game. If you are day trading you may need different answers. In the past I went through drawdowns of >40% and kept following the signals but it was gut wrenching and the temptation to second-guess the signals was very much present, nah extreme. Back then I was willing to take bigger risks to get high returns. So it is when building up something.

    Markets can move such that the profitability of any trend following system I can imagine suffers. This usually means rugged or absent trendiness and that in seemingly unrelated markets at the same time.

    Once I have dropped a system because it's equity curve had a very strong rise (equity increased 7x from previous drawdown low) and initial risks for new trades became so high that I considered it unreasonable and I didn't want to give back too much of the equity build up. The system then went into a new max drawdown and it took it years to recover. But recover it did. One wouldn't do that strictly by the book, however.

    Like I said earlier, I don't think one can tell in advance if a system stays profitable or not. Whether you want to continue to trade a system in a drawdown is a very individual decision and depends on your confidence in the system, financial and general situation, risk willingness. Following trends as a principle I believe is a valid approach and as such is something I am betting money on in spite of increased risks due to higher volatility and deeper retracements and there is nothing that I know of suggesting that there is a limit to adverse conditions one may encounter. Implementing several systems and trading more markets tends to smooth the equity curve but is no guarantee. There aren't any, it is speculation after all.

    I would continue to trade if the funds allocated to a system are down 50%. That is because I consider trend following a valid principle and have made up my mind in advance that I can live with that kind of variance. I wouldn't start trading a system when the drawdown in testing is very high. When I encounter it in real trading I'd keep going until either the allocated funds are depleted or until the next profit wave; then I'd reconsider. Wouldn't bet the farm, however. Just what I allocated to it.
     
  9. Bo_D_

    Bo_D_

    the chances of a system failing- will eventually reach certainty.

    statistically anyway.


    my systems produced great profit until 2 very nasty trades from 2 different systems in the last week. basically lost 50% of my portfolio, extremely nasty trades. my system profits from price action. but it cant tell when bad announcements come out. basically i ditched one system as it hit a much larger DD than historically tested. and i still trade the other one as it is a good system.

    you need to know why it went into DD. analyse it, and decide whether or not it can still be profitable.

    midnight here in australia so i apologize if none of that makes sense.
     
  10. Hi,

    my method is to run first the normal historical backtest to get the max historical DD (5 to 10 years may be sufficient for EOD system, if all principal market scenarios are included) and then add a monte carlo simulation system test, which simulates 10.000s of trading years.

    Description here:
    http://www.zentrader.de/mcs_article_traders2007.pdf

    So you get a better feeling of potential system-inherent DDs of your system and a better decision base (should I stay or should I go...).

    bye,
    zentrader
     
    #10     Jun 19, 2008