The biggest issue is the window size. If you mean a cross-over system, for how I trade at least the signal is way late. If just for the trend, I think if you take the middle value between the high and low as a series it is close enough with the benefit of no moving window. Beyond that you should just use a time series model. My guru for time series analysis is Aileen Nielsen. She is a lawyer from Yale with a M.S. in Applied Physics but she writes books about time series analysis on the side lol. Her 2016 lecture from a python conference is gold. Aileen has nothing on the geniuses on this forum though.
Instead of cross over, monitor the number of days a stock can stay above a MA. Do a frequency plot of the results at various MA period.
Also look at the distances and rate of change. It is pretty easy to see a cross over BEFORE it happens based on the rate of change of the delta between the two.
%% Good question, but no such thing as ''the best ma''; thats why some use a 50 + 200dma........................................................... barchart .com has done what you are asking, but they also use a 100dma which i seldom use. High tide maybe forecasted, but on MO orMS river, dont need that also. Hurricanes may mess up hi tide\LOL Surf fishers may differ; those that dont like moving averages help insure ma wisdom. Markets dont have ALL people people doing the same thing
I did an extensive moving average backtest using variables a long time ago. No edge found. There might be value with moving averages along with other inputs or as part of a "Trade logic" consideration for certain trading systems.
%% Moving averages really don't work; we have to do the work. Speed bumps or speed limit signs dont work all the time also ; on private property I try to avoid a speed bump, but drive under 200mph LOL+ true. Some one that cant find a profitable way to use a 200 dma should not try to use it. Price ''slamming thru'' a 200dma tell me a lot; but no predictions
Here's a clean and simple way to use lines like moving averages and pivots that you don't even have to draw yourself. Relax and let the chart draw them. Super easy to quick check a dozen cases to determine if of interest for further benchmarking. https://www.elitetrader.com/et/thre...t-right-here-baby.335635/page-29#post-5646944