Anyone actually trade mechanically?

Discussion in 'Strategy Development' started by MYDemaray, Jan 7, 2003.

  1. If you trade a 100% mechanical system, I would really like to hear how you have done with it. How do actual results compare to the backtested results. Has slippage been an issue? (I'm particularly interested in ES). What are the major challenges?

    The reason I ask is, I'm thinking about going live w/ a few strategies in TS...first manually and then automated. Wondering how much I should expect my results to differ from backtests. (Obviously, a large part of this will be due to the robustness of the system, but I'm curious more from an operational -- slippage, etc. -- standpoint.)

    Thanks in advance to all....
  2. I'm using a 100% automated and mechanical system on ES and NQ. On an initial account balance starting last July of $10,000, and a closing balance Dec 31 of $11,729.94, I had $175.90 difference between actual results and backtesting results. I'm happy with that.

    This system is a swing trading system (trades 1 to 5 days in length, several times per month). I use stop orders for entry, and either a stop or limit for exit.

    If you are going to trade a high volume of trades, intraday, the slippage factor will increase. I've traded several 100% mechanical systems in the past on ES that were daytrading systems, but found that slippage was too big of a factor, cut too deep into the profits, and the end result was not successful. .25 point slippage on each side of the trade is not too big a problem when the average trade gains or loses 15 points. However, it is a problem if you are only looking for 1 to 3 points.
  3. Thanks for the feedback. If you don't mind my asking, what was your max drawdown over that time period? How did it compare to what you expected?

    Thanks again. Anyone else?
  4. My max drawdown was $2,884. It was less than the backtested max drawdown of around $4,500.
  5. Thanks very much! Anyone else...?
  6. maxpi


    Tradestation reported that their average account gained 7% in 2000. I believe that most of those accounts are automatically traded, not real sure about that.

  7. Tradestation doesn't yet support automatic trading with futures (they will in version 7), so their estimates don't really apply in this context, since it only includes equities.

    Also, there are so many ways to approach mechanical trading that broad numbers are meaningless. I think to be relevent the studies would need to be around specific types of systems.
  8. I trade 100% mechanically in futures and options. The results are in line with historical testing and I keep a close eye on slippage. In the big futures indexes the average slippage for the sp is $26 and for the nd is $58. I use different entry techniques to mitigate slippage in different environments.

    good trading.
  9. acrary


    I develop systems using TS for SP and ES and automate them. However when it comes to trading them, I use my own discretion.
    I do so for a variety of reasons:

    1). I can see things that I would find too difficult to automate such as if a entry/exit point is indicated near the previous days high/low then I adjust for a earlier entry knowing these places are where hedge funds like to put their stops. Likewise if the market gaps open I try to avoid putting stops near the previous days close. I suppose the analysis could be automated, but it would take up allot of code.

    2). Globex doesn't accept MOC orders. If I have a winning trade going, I like to hold it till the close. Because of this I've had to do my buy/sell at market orders a few seconds before the close. (One thing to note - keep your system clock in sync. with your brokers or else you'll have orders getting rejected before they get submitted to Globex).

    3). If the arbitrage guys are taking a break, the spread between the ES and SP can be 1 pt or more. I've seen this many times. In these cases it's better to execute in the pit because the ES tends to lead the SP.

    4). ES tends to overshoot the SP at the extremes of the day so if the market is trading at new highs/lows it's usually better to execute in the pit.

    A couple of things to note when automating intraday:

    1). Forget using buy/sell next bar on open and exit this bar on close. Those were designed for daily strategies and can't be realistically duplicated with intraday data.

    2). If you're using buy/sell stops or limit orders make sure the strategy doesn't need split second execution. For instance, if you have a buy at the highest of the last 2 bars on 1 min. data, it could be hard to setup and enter the stop or limit order before the market trades through the price.

    3). Try to avoid market orders on ES. My own experience has been that I always have 1/2 pt slippage on each side. My stop and limit orders when hit on ES have never had any slippage so if you want the same results as the model, avoid market orders.
  10. Foz


    Have you tried the e-minis? I found that the reduced slippage more than made up for the 5x commissions.

    What we really need is a large, electronically traded equity index. Can anyone say DAX?
    #10     Jan 8, 2003