Using my basic Flo algo, I optimised Light Crude with many inputs fixed, leaving the necessary dynamic ones to be optimised. The regular 133 day optimisation shows a profit of about $3500 per contract with a high win rate and low drawdown. I don't believe that this is overly curve fitted even though the win rate is "abnormally" high. I've based it on how I trade.
Friday, 6 July 2012
Successful Trading with Algos Part 5
As I have said, the Walk Forward Analysis is a critical part of the algo trading process. The pics below show a good example of this.
Using my basic Flo algo, I optimised Light Crude with many inputs fixed, leaving the necessary dynamic ones to be optimised. The regular 133 day optimisation shows a profit of about $3500 per contract with a high win rate and low drawdown. I don't believe that this is overly curve fitted even though the win rate is "abnormally" high. I've based it on how I trade.
But look what happens if I do a weekly re-optimisation of those dynamic parameters. If you add up the Out Of Sample (OOS) profits outlined by the red in the second pic, the profits add up to over $6000 per contract - and this is on data UNSEEN by the algo.
Using my basic Flo algo, I optimised Light Crude with many inputs fixed, leaving the necessary dynamic ones to be optimised. The regular 133 day optimisation shows a profit of about $3500 per contract with a high win rate and low drawdown. I don't believe that this is overly curve fitted even though the win rate is "abnormally" high. I've based it on how I trade.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment