Traditional time-series techniques, such as auto-regressive and moving average models, can have difficulties when applied to stock data due to the randomness inherent to the markets. In this study, Long Short-Term Memory Recurrent Neural Networks, or LSTMs, have been applied to pricing data along with sentiment scores derived from web sources such as Twitter and other financial media outlets. The project team utilized this approach to complement the technical indicators observed at the end of each trading day for three stocks from the NASDAQ stock exchange over a 12-year span. A common benchmark to assess model performance on time series data is using the prior day’s closing price of a given stock to predict the next day’s closing value, which is a naive, but surprisingly accurate method when calculating the mean absolute error. The main objective of the paper is to use predictions from the various models assembled for the research, and then calculate whether the next day’s closing price will rise or fall when compared against the last predicted value. All models showed on average a roughly 2% accuracy improvement over the largely balanced up and down movements for the tickers used in the study.
Burgess, Michael; Javed, Faizan; Okpara, Nnenna; and Robinson, Chance
"Stock Forecasts with LSTM and Web Sentiment,"
SMU Data Science Review: Vol. 6:
2, Article 10.
Available at: https://scholar.smu.edu/datasciencereview/vol6/iss2/10
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License