Prognosticate of the financial market utilizing ensemble-based conglomerate model with technical indicators

The financial exchange is known for its outrageous multifaceted nature and instability, and individuals are continually searching for a precise and successful approach to control stock trading. The exact expectation of stock cost could assume a significant job in helping speculators improve stock returns. The motivation behind this paper is to precisely foresee the pattern of stock costs, giving a reference model to the pattern of financial exchange and the following technique for stock value expectation, and offer some incentive reference for a look into the estimating model of securities exchange and speculators investment choice. Research using a hybrid model to anticipate the stock market will have a fundamental improvement that stood out from using a single model to forecast the same. The proposed model is a regression-based ensemble model where specialized technical indicators are used as input features to anticipate the 1-day future opening cost of individual stocks. In this paper, we had taken five regression-based machine learning algorithms (Lasso, SVR, RidgeCv, K-NN, random forest) to make ensemble-based hybrid models that are a combination of them. The combined ensemble models are Lasso + SVR + RidgeCv, Lasso + SVR + RandomForest, K-NN + SVR + RidgeCv, K-NN + SVR + RandomForest, K-NN + Lasso + RandomForest, Lasso + Random Forest + RidgeCv, Lasso + K-NN + RidgeCv, Lasso + K-NN + SVR. We have selected three stock indexes for our experiment, i.e., S&P 500, DJIA, HSI. During our research, we found that the combined predictive model "Lasso + SVR + RidgeCv" is a generalized model that can perform best on all three datasets. The proposed model gives the best MSE, MAE, RMSE values are 0.0308, 0.1348, 0.1756. Our proposed model can bring some direction and reference for some investors and market controllers. Our study offers down to earth experiences and possibly helpful headings for additional examination concerning how machine learning can be viably utilized for stock market investigation and expectation.