Stock market prices are intrinsically dynamic, volatile, highly sensitive, nonparametric, nonlinear and chaotic in nature, as they are influenced by a myriad of interrelated factors. As such, stock market time series prediction is complex and challenging. Many researchers have been attempting to predict stock market price movements using various techniques and different methodological approaches. Recent literature confirms that hybrid models, integrating linear and non-linear functions or statistical and learning models, are better suited for training, prediction and generalisation performance of stock market prices. The purpose of this review is to investigate different techniques applied in stock market price prediction with special emphasis on hybrid models. This review paper classifies the literature pertaining to hybrid models applied to stock market price prediction in accordance with their input characteristics, makes comparisons between hybridised models, and exhibits the performance evaluation measures used. It summarises the salient characteristics of the contemporary models applied in the stock market price and index prediction. The surveyed papers show that hybrid models are widely used for stock market prediction.
Dassanayake, W., Jayawardena, C., Ardekani. I., & Sharifzadeh, H. (2019). Models Applied in Stock Market Prediction: A Literature Survey. Unitec ePress Occasional and Discussion Papers Series (2019:1).
About this series:
Unitec ePress periodically publishes occasional and discussion papers that discuss current and ongoing research authored by members of staff and their research associates. All papers are blind reviewed. For more papers in this series please visit: https://www.unitec.ac.nz/epress/index.php/category/publications/epress-series/discussions-and-occasional-papers/