Application of machine learning algorithms in the domain of financial engineering

Abstract

Financial engineering is crucial for effectively combining finance with quantitative approaches. This study aims to forecast the performance of the Nasdaq stock market by considering numerous factors like wind, hydro, thermal, gas, and nuclear variables. To accomplish this, we utilize sophisticated predictive models, namely adaptive lasso (ALasso), elastic net (Enet), artificial neural network (ANN), convolutional neural network (CNN), and long short-term memory (LSTM). By using these advanced methods, our goal is to offer perceptive and precise predictions, which will enhance comprehension of the complex dynamics within the financial markets. The evidence suggests that the LSTM model has demonstrated superior accuracy in predicting changes in the Nasdaq stock market when compared to ALasso, Enet, ANN, and CNN. While ALasso, Enet, ANN, and CNN exhibit comparable RMSE and MAE values, their performance is slightly less competitive than that of the LSTM model. The marginal differences in RMSE (ALasso: 0.319, Enet: 0.317, ANN: 0.3, CNN: 0.32) and MAE (ALasso: 0.277, Enet: 0.276, ANN: 0.252, CNN: 0.278) emphasize the comparable effectiveness of various methods, but they somewhat drop below the LSTM model in terms of precision. The findings showed the significance of well-known and advanced ML techniques, particularly LSTM, for enhanced accuracy in financial market predictions.

Keywords

• Financial engineering
• Stock market
• Machine learning models
• Forecasting