/RNN-moneymaker

Deep learning to forecast stock price

Primary LanguageJupyter Notebook

RNN-moneymaker

Utilizes a parallel convolutional-LSTM architecture to predict closing price on NASDAQ stocks. Trained and tested on historical 20-year time series data (price, range, volume) from 3000+ NASDAQ stocks to achieve a MSE loss of ~ 0.73, doubling the performance of a vanilla LSTM and simple one-step lag predictor.