An Empirical Examination on Forecasting VN30 Short-Term Uptrend Stocks using LSTM along with the Ichimoku Cloud Trading Strategy

Abstract

Stock market forecasting is a highly difficult time-series problem due to its extreme volatility and dynamic. This paper proposes a Long-short term memory (LSTM) model that predicts the probability to outperform the market of all of the VN30-Index constituents by using historical price changes along with features calculated from the Ichimoku Cloud trading strategy. After acquiring the pro-posed model’s outputs, we buy three stocks with the best probability to sell them ten days later. We then reinvest the money on the next day using the same strategy. The yearly returns of the above trading scheme are used as the empirical results. This study is conducted in a period of 9 years – from the VN30-Index’s establishment in 2012 to the end of 2020. On average, the adoption of the Ichimoku Cloud features in the LSTM model makes our trading strategy go from an annual loss of 2.86% to a profit of 14.29%.