Week 8: X-Value
Xander S -
I spent this week formulating and validating a composite rating system to evaluate the performance of the backtested strategy-context triple combinations. I have named it the “X-Value,” and it is designed to aggregate and normalize a variety of performance metrics into one, easily comprehensible score. Most people assume the return of a strategy is enough information to evaluate its effectiveness, but return is highly variable and does not paint a full picture. The goal with the X-Value is to identify strategies that score highly across multiple metrics, as that should suggest structural superiority. The X-Value is a number between 0 and 10, calculated using the following formula (simplified):
2.5 x Quality Score (also created – Bayesian-adjusted Expectancy x square root of # of trades) + 1.5 x Kelly Criterion + 2 x Sharpe Ratio + .5 x Calmar Ratio + .5 x Profit Factor + Alpha + 2 x Realized Return
Each metric is capped and normalized on a scale of 0 to 1. However, creating this formula was not enough. I needed to validate that it actually meant something. I decided the best way to do this would be to analyze the correlation (or lack thereof) of an X-Values to whether or not it’s associated return beat simply buying and holding. Essentially, this is comparing active trading with passive ownership of equity – the crux of the debate around the legitimacy of technical trading. I fitted a two logistic regression curves around training and test windows to see whether they would resemble each other or diverge.
As you can see, a higher x value correlates with a higher probability of beating buy and hold. Since the curves match, X-Value has some predictive power, albeit modest (AUC of ~ .535). I am in the process of fitting logistic regressions around individual companies in the ETFs to see if the correlation holds up with specific stocks. Validating the X-Value with out-of-sample data allows me to use it as the overall performance metric for which I can answer my original research questions. I will expand on this next week. See you then!
Comments:
All viewpoints are welcome but profane, threatening, disrespectful, or harassing comments will not be tolerated and are subject to moderation up to, and including, full deletion.