How brute should be execution of brute force algorithm?
2 replies
Rom
8 years ago #114200
The data: daily oil price contract since 2003; several easy to spot “trends”.
The setup somewhat restrictive, but nothing prohibitive, I think.
The result: after >6.5 million tests not a single strategy popped.
I modified settings removing, for example, the stops (system all the time in the market) hoping it will produce a few strategies, but nothing materialized. I examined the data (seems ok), changed the data to gold contract, without result.
I wander how typical is my experience ?
geektrader
8 years ago #132639
Brute Force will take you years to find something useful if you combine a lot of parameters. Go with genetic optimization several times, figure out a good range for each parameter of the strategy, then you can run genetic optimization again in just that range and narrow it down. Brute force, except if you have a low amount of runs, does lead to nothing without a NASA computer:)
Rom
8 years ago #132647
Thanks !
Viewing 2 replies - 1 through 2 (of 2 total)