Moving Averages with Machine Learning Boosting

Hi all

For my master's theses, I'm studying some technical indicators to use it later and compare it with the simple buy and hold strategy (B&H). I started with the crossing moving averages, that will give me signals to buy or sell when the short moving avg crosses the long moving avg. Currently, I don't have any logical way to choose the dimension os these averages. From what I have seen, people seem to choose these lengths without any complex rational thought. "Just choose like 50 and 150...." I was wondering if there is actually a more precise way to do it? I saw a very small article with some results of comparing the B&H strategy with the Moving avg but with a learning method called Boosting to give better results.

http://cs229.stanford.edu/proj2008/...cationInVariableMovingAverageTradingRules.pdf

I know next to nothing regarding Machine Learning since my background is from Mathematics. Can you provide me with some ideas or where can I find more about this.

Thanks a lot!
Tomás

 

Have you tried hyperparameter tuning via gridsearch?

ML(machine learning) is one of the areas that I work in professionally. Not saying that it will necessarily work in trading. But you should try it out.

 

Price action curves have consistency for specific derivative over time. Curve fitting for optimal results works.

 

Where is traderGOD when you need him lol.

 

Thks a lot. I will definitely look at this! Although one question. As I was looking for some information I saw that it can also be done by via "random" and not gridsearch. The gridsearch is probably more accurate but it will take more time. What do you think about the "random" process

 

Thanks a lot! I had already heard about this but never really consider to actually use it. I will start looking now for more information about this!
Once again many thanks

 

yah man where is he (or she! I don't judge)

 

Are you familiar with Excel and Excel VBA ... or Excel Solver?

 

I don't know what field you're studying for your master's, but you might want to check with your adviser to see if there is value in putting a lot of work into this. Because studies of "buy and hold" versus market timing, indicators, technical analysis, etc., etc. have been published many times.

 

Who uses MA crossovers? People that don’t like money?