The aim of this project is to use machine learning to combine technical analysis and fundamental analysis in stock trading. Technical analysis and fundamental analysis are two commonly employed investment tools for timing the market and select stocks. Technical analysis often relies on daily price or trading volume, and is usually applied for short-term trading, while fundamental analysis focuses on financial ratios and more suitable for long-term investing. Hedge fund managers have argued that, to achieve best trading performance, one should combine both tools.
Yet an important but difficult problem remains -- there are a large number of technical and financial indicators. While each of them can be used for predicting future returns, it is ex-ante unclear which indicators should be used, and how to combine them to improve trading performance. The high-dimensional nature of machine learning methods is well suited for such a challenging problem. Existing literature has used machine learning for combining different technical indicators or fundamental ratios. Much less is explored to combine both methods. This project attempts to fill the void. It may also shed some lights on the questions such as when ML-combined strategy works and for what types of stocks it works better.
This project is particularly useful to those students who intend to proceed to PhD study in algorithmic trading and machine learning, it will help you lay a foundation for understanding latest advances in relevant research of machine learning and stock trading.
Indicative background references:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3233119 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3741015 https://ieeexplore.ieee.org/abstract/document/8623000 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3706532
You may also see a related Kaggle competition at https://www.kaggle.com/c/ubiquant-market-prediction.