An Enhanced Stock Index Forecasting Scheme Embedded Lag Effects of Price Movement Velocity

It is always challenging for investors to extract and use the information behind the raw data to make both accurate and fast predictive models. Most methods use autocorrelation variables to construct technical indicators or implement signal transform to extract patterns from different feature spaces. This study proposes an enhanced predictive scheme named OS-SVR, OS-MARS, and OS-ELM, combined with the lag value of the open price and the lag value of the first-order derivative from displacement vector and time to predict the next day's open price. The empirical experiment proves that the scheme improves the algorithm's performance, and OS-ELM outperforms other methods for improvement rate and accuracy.