Thoughts on this strategy development...

Discussion in 'Strategy Development' started by frostengine, Aug 23, 2006.

  1. I have developed strategies before that have proven to be successful, these strategies were developed using 1 minute bars, which I could then change into 5 minute bars, 10 minute.. whatever was needed for the strategy... that sort of data was easy to come by.

    My newest strategies are based off of tick type charts. And my strategies dont just involve the actual futures contract i'm trading but several other things for which I record tick data for. All of this data together, which represents nearly a 15 mb file PER trading day, is used to determine buy and sell signals. I currently only have 10 trading days worth of this type of data captured. In the past I always felt I needed at least a year or 2 years to prove how successfull a intraday strategy is. In fact I normally used up to about 5 years of intraday data.

    My question is, how much reliabilty can be put into a method that has only had 10 days of data? Even if that method has peformed over 100 trades in that time period? I would love to get more data to back test a promising new method, except the type of data I require I dont believe can be easiy found or purchased.....

    This leaves me with only a few options, wait a long time capturing data and see if the statistics hold up and then trade this live. OR capture just a few more weeks, if it holds up start trading live and watch key statistics such as win rate, and expectancy to see if they begin deviating from what I expect.....

    I have performed all of my usual analysis to prove to myself the method is NOT curve fitted to this amount of data, so I feel confident in that, now the only thing is, was this 10 days worth of data an anomoly.

    All figures take into acount $5 round trip commisions, which is slightly higher than my true comisions. All figures trades were made at either the bid or ask, with a random 1 pt slippage.

    The current statistics of this method are: Trading 1 YM contract

    Wins: 30 Total: $1645 Avg Win: +$54.8
    Losses: 84 Total: $1240 Avg Loss: +$14.7

    Win Percent: 26%
    Loss Percent: 73%

    Reward/Risk Ratio: 3.7/1

    Expectancy: = $3.55

    Net for 10 days: $405

    The same method was used to trade ES, ER2, and NQ. All of which showed a pretty consistant win/loss percentage, but the average win and average loss made the NQ and ES both have negative expectancy.

    unlike previous methods I have designed, this one has a VERY low draw down..... but it also has a very low winning percentage.

    Out of the 10 days of data, the strategy made money 7 of those days. The largets loss on any one trade was only $25, largest gain on any one trade was $200

    Is the best course of action to simply wait and capture more data for a long while?