The Revitalization of Old School Trading with Machine Learning and Technical Analysis

Shen Wang, Fordham University


The Efficient Market Hypothesis (EMH), initially proposed by Fama (1970), is one of the cornerstone theories of today’s Financial Economics. By far, the most frequently quoted version of the EMH is the weak form or its more ambitious offspring, Random Walk. Besides a very well-reasoned illustration of the original theories, Fama (1970) points out that “the existence of a strategy that consistently outperforms the “buy-and-hold” is what’s needed to bring counterargument to the EMH”. In this paper, we try to establish such a trading system with a modified Extremely Randomized Trees algorithm as the baseline “engine”, and stock-specific trading rules from traditional technical analysis for performance enhancement. Our results show that such a trading system can be feasible for most of the non-arbitrage S&P 500 member stocks. Meanwhile, with a micro-level strategy in place for each member stock, we can almost certainly find some hedging effects in an initially equally weighed portfolio and curb the volatility of that portfolio to 25% of the level of the buy-and-hold strategy. Granted that the current model has enormous room for improvement, we can verify that traditional technical analysis and active management strategies still have their raison d’etre in today’s US stock market, and our proposed trading strategy can be a successful application of them. Furthermore, several stock selection methods are found to be capable of generating positive future alphas in our experiments, including a volatility-based approach, a momentum-based approach, ones from financial metrics such as EV/EBITDA and ROE, and ones from the “decileVote” and “momentVote” algorithms. A monthly VIX directional prediction model is also proposed in a bid to assist with the timing of regime switch for the stock-specific enhancement modules, with two options for the variable selection for the VIX forecasting model proposed and verified.

Subject Area


Recommended Citation

Wang, Shen, "The Revitalization of Old School Trading with Machine Learning and Technical Analysis" (2022). ETD Collection for Fordham University. AAI29328088.