| |
The study explores the application of singular value decomposition (SVD) as a tool for analyzing the linear dependence of the predicted stock price on a set of market indicators. Singular value decomposition is employed to decompose the feature matrix, enabling the identification of the most significant components that describe data variability and determining the degree of linear dependence [1] between the indicators and the target variable—the stock price. This approach facilitates dimensionality reduction, mitigates multicollinearity, and highlights key factors influencing the price.
Additionally, the study addresses the regression task for predicting stock prices in the short term based on the extracted features. Machine learning methods, such as linear regression, regularized models (e.g., Ridge or Lasso), or more complex algorithms like gradient boosting applied depending on the data characteristics. The analysis includes data preprocessing steps, such as normalization, overseeing missing values, and feature selection. Predictions evaluated using metrics such as mean squared error (MSE), mean absolute error (MAE), or the coefficient of determination (R²) [2], allowing for a quantitative assessment of the model's accuracy.
The study also discusses the limitations of the approach, including assumptions of linear dependence, market volatility, and the impact of external factors not captured in the dataset. To improve prediction accuracy, the integration of additional data, such as news feeds, macroeconomic indicators, or market sentiment analysis. The results obtained can utilized for developing trading strategies or supporting investment decision-making in the context of short-term trading.
Keywords:singular value decomposition, linear regression, linear dependence, stocks, indicators.
|
LEGAL
INFORMATION:
