More on the accuracy measure because you’re using classification, if significantly more stocks in the universe went up than down during the sample period then the model can predict 1 or True most of the time and still achieve a high True positive rate and thus high accuracy. But the model won’t generalise when the universe is falling or the number of ups and downs are balanced

Researcher | Investor | Data Scientist | Curious Observer. Thoughts and insights from the confluence of investing and machine learning.