trading algo using random forest classifier. I am using a tree depth of 60 and number of trees as 500. Each time I back test with same parameters and with same time frame, I get different result. Why so, if the parameters and history tick values are same, why will I get different return each time?
Also, I read that tree depth should be normally square root of features and more number of trees are better. What is the number of features in a algo trading, how o I find that out ? So, how do I use an optimal tree depth?