Portfolio Strategy Based on Gaussian Reversal Predictions

As the stock market boom continues, stock forecasting has gained more attention, and trading strategies based on statistics, mathematics, and other principles have gained greater credibility. This paper introduces stock prediction and portfolio strategy selection models based on reasonable assumptions, and proposes a novel online passive-aggressive algorithm stemming from the Kelly criterion. The Portfolio Strategy based on Gaussian Reversal Predictions is obtained by combining the prediction model with the innovative online learning passive-aggressive algorithm. Comparison with typical forecasting methods such as LSTM and PAMR shows that the proposed method is applicable to stock forecasting and works better than both models overall, with final gains surpassing the LSTM model by $553K and the PAMR model by $654K.