I create my algos by going through a fairly easy process. It's not easy to create a good algo, but having a process makes it easier.
I start with an idea. Eyeball the charts. Then I make a simple strategy to run in TradeStation or MultiCharts. If it looks promising, I refine it. I then run an optimisation. Below is one such result.
In addition to the rows of numbers, I draw your attention to the comment circled in red that the algo is a good candidate for Walk Forward Analysis. This is a TradeStation result and TS can produce WFO files to do the WF. After running the WFO, TS examines the result and creates a test. This is what it's result looks like.
Looking further into the WFO results, I can look at lots of stats including the ones below.
Once I am happy and I lock in the inputs to use based on the WFO, I can look at the stats for the historical performance of the chosen inputs.
Looking at the Equity Curve, I can see a couple of dips.
As this algo will run as part of a portfolio of markets, I'm not concerned about the dips as I have accepted the draw downs flagged by the results.
In the coming year, I will be relying more and more on my fully auto algos. I can get some good diversification with proper money management and believe that I'll work less and earn more.
Hi, what minimum level of Profit factor would you consider the system usable, assuming all other statistics are okay?
ReplyDeleteHi EL,
ReplyDeleteI thought you liked high win ratios (70% or better). These algos have win ratios less than 50%. What changed?
Both great questions! I'm going to answer both these questions in a post tomorrow, Thursday as a short answer here wouldn't be enough.
ReplyDelete