The fast development of artificial intelligence (AI) and machine learning techniques can help develop precise and trustworthy forecasting models with precision to predict the volatile and irregular stock behavior in the market. The purpose of this paper is to develop a robust predictive model that has the potential to use AI in enhancing the accuracy of stock market profit assessment. This study evaluates the effectiveness of a long short-term memory (LSTM) model for predicting the stock market behavior by incorporating key economic factors.
This study constructs an LSTM model that incorporates economic factors influencing stock market direction, including interest rates, inflation, GDP and unemployment rates. These indicators are selected based on their theoretical relevance and empirical significance. A multivariate LSTM model is developed and validated against traditional autoregressive integrated moving average (ARIMA) and machine learning-based random forest (RF) models. Model performance is evaluated using root mean squared error, mean absolute error and directional accuracy.
The experimental results show that the performance of LSTM model varies across four different cross-validation folds, with the fourth fold appearing to have better generalization performance than the first fold, as both the mean training (0.0450687) and validation (0.01293) values are lower in the fourth fold than in the previous folds. This study highlights that the recurring neural network-based LSTM model, when combined with a holistic approach provides accurate and interpretable predictions for stock market indices of highly volatile markets. The results demonstrate that the LSTM model significantly outperforms ARIMA and RF models in terms of predictive accuracy. The LSTM effectively captures nonlinear patterns in financial time series and shows strong generalization capabilities.
This study suggests that AI-driven techniques can be adapted to predict stock behavior with precision. In addition, such models may be effective for other market indices displaying similar behavioral patterns, aiding stakeholders in making informed investment decisions.
This study contributes to the body of knowledge by proposing a practical AI-based stock market prediction model that incorporates economic factors directly impacting stock market movements. This integral approach helps develop more accurate prediction models, providing insight into various elements of the stock market. The inclusion of baseline model comparisons and detailed residual analysis enhances the methodological rigor and applicability of the approach.
