Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
An Intelligent Market Cycle Detection System
The detection of stock market cycles has attracted the attention of finance scholars and market practitioners. Accurately identifying the direction of a market can significantly increase the returns of investors. Despite this importance, conventional methodologies in the literature have predominantl...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
AUC Knowledge Fountain
2021
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| Summary: | The detection of stock market cycles has attracted the attention of finance scholars and market practitioners. Accurately identifying the direction of a market can significantly increase the returns of investors. Despite this importance, conventional methodologies in the literature have predominantly attempted to evaluate the effect of subsets of factors as precedents to stock market cycles and with little agreement on what constitutes critical factors. There seems to be a lack in the literature for a comprehensive study that examines a multitude of factors at the same time on the S&P500 as the laboratory. Factors are categorized into: political events, economic factors, market fundamental indicators and technical trading rules. The main objective of this research is to detect a 5-tier market cycle and conduct a comprehensive analysis of the above factors by employing different machine learning and deep learning classification techniques. Our findings were that the Extreme Gradient Boosting method was of the highest accuracy in identifying the cycles, which led to returns that beat the buy-and-hold strategy by 88% versus 37% between 01- 2018 and 12-2020. ARIMA, Fourier Cycles, and Remaining days to the next presidential elections together were found to be of the highest importance in detecting the 5-tiers. |
|---|