A momentum trading strategy is based on the premise that past winners are likely to keep winning, and past losers are likely to keep losing. In other words, a momentum trader looks at past performance to decide which investment to buy and sell. This is not unlike looking at historic price charts and deciding if the price will go up or down in the near future. This is an exercise of pattern recognition and learning and applying that learning to predicting among all possible candidates how likely each ones will provide the most returns. Sounds like a good problem for a machine learning algorithm?

Winning stock predictions on this website are generated by training a neural network classifier on historical daily closing prices of more than 1000 US stocks from 1995 to present day. The model outputs probability of each stock having a superior return 1 month into the future. The list is ranked by the number of days out of the past month that the model predicts will have superior return. The theory is that the stocks that exhibit the winning patterns more often should be a stronger signal than others.

All content on this website is for informational purposes only. The stock picks and listing do not constitute investment advice. A computer algorithm or model will always give an answer based on the inputs provided to it, but it is by no means an accurate prediction or guarantee. Investing in stocks involve risk of loss. Loss of principal is possible. A security’s or a firm’s past investment performance is not a guarantee or predictor of future investment performance.