The trading literature deals mostly with the design and development of trading strategies. There is very little discussion on the topic of how to properly validate them, and most of the system validation techniques have been developed under a traditional statistical framework. Article [1], however, argued that using a traditional single testing framework is not enough, and provided some guidance on how to properly evaluate trading strategies using a multiple testing framework. It concluded,
In summary, the message of our research is simple. Researchers in finance, whether practitioners or academics, need to realize that they will find seemingly successful trading strategies by chance. We can no longer use the traditional tools of statistical analysis that assume that no one has looked at the data before and there is only a single strategy tried. A multiple testing framework offers help in reducing the number of false strategies adapted by firms. Two sigma is no longer an appropriate benchmark for evaluating trading strategies.
We agree that evaluating a trading strategy under a single testing framework is not enough, as the strategy might happen to be a winner by chance. However, we believe that the approach proposed by the authors is not practical, i.e. it’s not scalable, intuitive, and cannot be easily implemented.
The article then pointed out that the Sharpe ratio is not an appropriate measure of risk-adjusted return. This is true, especially with options trading strategies whose PnLs are not normally distributed and are often heavily skewed.
Our work has two important limitations. First, for a number of applications the Sharpe Ratio is not appropriate because the distribution of the strategy returns is not Normal. For example, two trading strategies might have identical Sharpe Ratios but one of them might be preferred because it has less severe downside risk.…
The article also discussed the importance of adding a low-correlated trading strategy as a diversifier to a portfolio.
…our work focuses on individual strategies. In actual practice, the investment manager needs to examine how the proposed strategy interacts with the current collection of strategies. For example, a strategy with a lower Sharpe might be preferred because the strategy is relatively uncorrelated with current strategies. The denominator in the Sharpe Ratio is simply the strategy volatility and does not measure the contribution of the strategy to the portfolio volatility. The strategy portfolio problem, i.e. adding a new strategy to a portfolio of existing strategies is the topic of Harvey and Liu (2014c).
For more discussion on the benefit of adding a diversifier to a portfolio, read Impact of a Low Correlation Trading Strategy
References
[1] Harvey, Campbell R. and Liu, Yan, Evaluating Trading Strategies, https://ssrn.com/abstract=2474755, 2014
Further questions
What's your question? Ask it in the discussion forum
Have an answer to the questions below? Post it here or in the forum
I made some mistakes while visiting Sicily for the first time. Next time I go to Southern Italy, I'd golf and skip Taormina from "The White Lotus."
