Since the “high stock dividend” of A-share companies in China often leads to the short-term stock price increase, this phenomenon’s prediction has been widely concerned by academia and industry. In this study, a new multi-layer stacking ensemble algorithm is proposed. Unlike the classic stacking ensemble algorithm that focused on the differentiation of base models, this paper used the equal weight comprehensive feature evaluation method to select features before predicting the base model and used a genetic algorithm to match the optimal feature subset for each base model. After the base model’s output prediction, the LightGBM (LGB) model was added to the algorithm as a secondary information extraction layer. Finally, the algorithm inputs the extracted information into the Logistic Regression (LR) model to complete the prediction of the “high stock dividend” phenomenon. Using the A-share market data from 2010 to 2019 for simulation and evaluation, the proposed model improves the AUC (Area Under Curve) and F1 score by 0.173 and 0.303, respectively, compared to the baseline model. The prediction results shed light on event-driven investment strategies.
Publicly available retirement planning tools are publicized to aid households in their retirement planning efforts, but households are likely overestimating tool effectiveness. The authors conclude that the advice provided from a majority of these tools is extremely misleading to households and propose a more systematic approach to tool development by improved choice of input variables. Analyzing theory and professional advisor opinion, critical input variables are identified and recommended. A case scenario is developed incorporating these critical variables and used to test the efficacy of 41 publicly available retirement planning tools.
There are challenges to analyzing huge volumes of data in the financial sector. How to handle big financial data intelligently is among one of the most important topics faced by researchers and practitioners. The stock market data are too large or complex to be dealt with by traditional data-processing application software. In this research, we propose using the orthogonal array to systematically generate pairs of input data fields for the Machine Learning model developed in our previous works. Trials in the automated wealth management industry (e.g. Robo-Advisors) have increased with the introduction of newer data analysis tools and technology applications. This has resulted in new methods, variables, and ideations being considered for optimal predictive analysis in the stock, bond, and cryptocurrency markets. Large data sets used in conjunction with machine learning are telling and predictive for different points in time. Our research attempts to understand which input factors will affect the stock market the most. As a result, we are expecting to reduce the volume of data needed to supply our machine learning model.