A 40 Trading Day Model for the S&P 500

I ran my largest dataset ($SPY) through a random forest machine learning model to identify which metrics, and metric combinations could best be used to predict SPY 0.00%↑ price in XX number of days. I used the same model to determine the optimal number of days such useful metrics could predict. It landed on 40 days as the optimal prediction period using a 5-perio…