Re: Success Stories
I like that approach a lot ATS! Also please add to your FF thread! I read over it several times and find it very informative!
I for one have several different approaches and theory’s about data mining strategies,
I have built strategies using over 200,000+ bars of data with 20% OOS with the OOS results not varying more than 5% r squared and balance line stability than IS. Those particular strategies do not trade often however they are looking promising in live testing however there is not enough results to establish a result just as yet.
My latest method is to create a portfolio of 50+ strategies using 10,000 bars of data all IS on M30 timeframe using strict Monte Carlo testing to weed out the over fitted strategies. I have just started testing these in the last two weeks with some good results so far.
This goes against everything I believe should work however I am very mindful of the fact that the market doesn’t care about what I think should work which is why I keep a log and all the settings used to create different portfolios so over time I can come to some sort of conclusion as to what works well and what doesn’t.
On every strategy however I run an entry time indicator as a preset to eliminate the strategy from trading around rollover (midnight for brokers on the international Timezone standard for forex GMT+3 incl daylight savings I believe)
I also put a spread protection in which i refer to my brokers average spread and adjust accordingly.
I also like to turn my strategies off around high impact news events as these can have a huge impact of over all results.