To me, an outlier is a sign of lack of consistency. Going off piste can work for or against you - have a big win or big loss. I don't want to count on luck or hope. I want to base my trading on the probabilities and the way to know the probabilities is to test and test until I can data mine the information I need.
A big part of what I do, both for discretionary and autotrading, is Walk Forward Analysis. To put it in its simplest terms, the way I use WFA is that after a period of trading I look at what worked. For example, I may see that a maximum Stop Loss of, say, $300 was profitable. I then want to see what happens using that Stop Loss if I roll the data forward and use that Stop Loss. Is my methodology still profitable? If yes, then I have robustness with my methodology.
Now I know that was a very simple example and in fact I do my WFA on many variables but I want to see robustness or otherwise I have a big risk that I have just curve fit my methodology to the past data.
Below is an example of a WFA report. It looks back about 123 days and chops the data into 4 week periods and then runs the methodology forward for 5 days.
As you can see, each 4 week period is profitable. I then drill down to make sure I don't have any one trade being more than 1% to 3% of the profit (depending upon the amount of each In Sample data block). No outliers.
Using WFA is equally useful with discretionary trading as with auto and hybrid trading. Any one or more parameters can be tested.
To be CP I need Consistency and the only way I can get it is to know what works.
The morning in London was quiet. The DAX continued to provide reasonable trades although my second entry had a lot of slippage due to the pop in the price. The RED and CYAN arrows are the entries and the WHITE arrows are the exits.
No comments:
Post a Comment