In a recent post, I shared my “Proven Winner” optimization methodology. The concept in brief is to only allow the walk-forward optimization engine to select strategies that worked over a long period of time for WFO consideration. It is possible to do this in Tradestation Easylanguage or Multicharts Powerlanguage by optimizing a single master switch which is used to select the sub-parameters of the strategy.
One potential benefit is that it sets up the ability to do “logical adaptation” or optimization by “bounding” or “constraining” the optimization potentials. For example, a strategy could have variations that are more suited to specific market regimes. By bounding the optimization to choosing among the logical variants, it reduces any risk that strategy logic is altered into something else, i.e. buying strength instead of weakness.
As per example, you may have strategy that is optimized for range markets such as one that uses a profit target and another that uses an EOD exit for trend markets (i.e. simply set the profit target to a huge value). The developer can encode a few variants of each. By virtue of constraining the possibility space, the optimizer is forced to do something logical in optimizing for range markets or trending markets compared to finding a very specific optimal. If the optimizer has fewer choices to consider, it may mean the choice decided upon is less random.
I do not suggest this approach will always work better or produce better results or even good results. But, the ability to setup more logical decisions does make some sense to me.
The author is passionate about markets. He has developed top ranked futures strategies. His core focus is (1) applying machine learning and developing systematic strategies, and (2) solving the toughest problems of discretionary trading by applying quantitative tools, machine learning, and performance discipline. You can contact the author at email@example.com.
Please log in again. The login page will open in a new tab. After logging in you can close it and return to this page.