Improvement of classification accuracy by large-scale language generation model

Classification analysis of TOPIX Core30 companies through LSTM

Abstract

News has great impacts on asset prices in the financial markets. Many attempts have been reported to ascertain how news influences stock prices. However, the limitations in the number of available data sets usually become the hurdle for the model accuracy. In this study, Reuter's news texts were classified based on each return through LSTM models. As a result of the analysis, we found the prediction accuracy could be improved by using texts generated by GPT-2. Further examination of the classification model will be planned.