Are you only using the Sharpe ratio?

Introduction

The Sharpe ratio is a useful metric to evaluate strategies. But by using it in isolation you risk missing out in two main ways:

• you deploy strategies that don’t perform well in real-life because of errors in accurate modeling of realities such as transaction costs, slippage or black-swan events

• you discard promising strategies that you could have used in conjunction with others, as part of a portfolio, simply because it didn’t meet some arbitrary threshold of Sharpe ratio.

In this article, we will explore various performance evaluation statistics beyond the Sharpe ratio, shedding light on their significance and how they can contribute to a more robust assessment of trading strategies.

Why is the Sharpe ratio a limited metric?

Like any other metric, the Sharpe ratio measures just one aspect of a strategy. This is clear from the formula of the Sharpe ratio:

Sharpe Ratio = (Expected return - Risk free rate)/standard deviation

The Sharpe ratio measures the return per unit volatility (or standard deviation of returns). It’s in this implicit definition of what is considered “risk” where the differences between different metrics lie.

Other metrics with different perspectives on risk

So which one should I use?

Choosing the right metric is difficult. In fact, you should not choose just one but a basket of metrics.

What matters more is the correlation between the metrics you choose. In the table above, each metric is highly correlated with each other. So, there is little additional information to be found in adding multiple metrics from the above table to your basket. Perhaps a more nuanced way to approach choosing between the metrics in the table above is to realise that just like beauty, risk is somewhat subjective and lies in the eye of the beholder. Investors must decide in advance which measures of return and risk align best with their preferences and select a combination of ratios that reflects those preferences.

And then, we suggest combining your preferred metric with one of the metrics below that are less correlated with the Sharpe ratio.

Metrics with lower correlation with Sharpe Ratio

What about time?

You will have noticed that so far we have not spoken about how any of these metrics change over time. Exploring how your strategy’s returns & metrics evolve over time is crucial to understanding why your strategy works.

There are two ways in which focusing on average metrics, without taking into account time, can lead to poor decisions.

Ignoring time in market

Consider a scenario where we are trying to choose between two strategies:

• Strategy A gives a CAGR of 10% and is invested in the market 5% of the time

• Strategy B gives a CAGR of 20% and is invested in the market 90% of the time.

Assuming equal leverage, it might be wise to choose strategy A, despite the lower CAGR. The reason for that is that while strategy B has a higher CAGR, it will also occupy a lot of “space” - since it is almost always invested in the market, no other position can be opened during that time without increasing leverage, incurring a significant opportunity cost.

On the other hand, if we choose strategies that rarely take a position in the market, we can afford to deploy more of them at the same time and yield superior returns.

Being exposed to the market for a small fraction of the time may also be desirable as the chance of being affected by an unforeseeable black swan event would also go down proportionately.

Therefore, adjusting for time spent in the market can be very useful. One way of doing this could be to divide the (risk-adjusted) returns by the strategy’s average exposure or time spent in the market expressed as a percentage.

Ignoring concentration of results

When scrutinising a trading strategy’s distribution of returns, the last thing you want is for the bulk of returns to be concentrated either in a very short timespan (which may indicate overfitting) or in a low number of trades (which would diminish the statistical significance of the edge). Ideally, besides from a high risk-adjusted return metric, a trading strategy will exhibit the following characteristics:

• A high number of trades per year;

• A high percentage of winning trades;

• A return distribution with low kurtosis (i.e. no fat left and right tails);

• Returns that are not concentrated in time.

Therefore plotting risk-adjusted metrics against time may reveal hidden information behind when & why your strategy performs well or poorly.

A note on portfolios of strategies

Thus far, we have focused on evaluating single strategies against other strategies in a “winner-take-all” type of world.

Of course, in reality, traders often choose to deploy multiple strategies as part of a portfolio of strategies.

While all of the metrics above can be calculated at a portfolio level as well, a paramount and arguably underrated metric when evaluating a trading strategy is its correlation to your existing portfolio of strategies. The intuition is very simple: if a strategy’s correlation to a portfolio is very high, its addition will not be accretive to the portfolio-level risk adjusted returns. On the other hand, if strategy level returns are promising and the correlation to the portfolio equity curve is low, then the strategy in question is a valuable diversifier and can help improve the portfolio-level risk adjusted returns.

We recommend understanding why a strategy in your portfolio underperforms when it does and then aim to develop complementary strategies that exhibit strength during those same periods. Thinking about your portfolio as pairs or combinations of complementary strategies is a very powerful heuristic that can pave the way to high overall portfolio risk-adjusted returns.

In fact, it is very conceivable to have a collection of highly complementary strategies with mediocre Sharpe ratios that can yield enviable portfolio-level Sharpe ratios. This realisation is very much in line with two very common adages within the trading world:

1. Keep your strategies as simple as possible

2. Diversification is the only free lunch in trading.

Conclusion

In the world of trading strategy evaluation, the Sharpe ratio has long been hailed as a popular metric for assessing risk-adjusted returns. However, it is important to recognize that relying solely on a single metric may not provide a comprehensive understanding of a trading strategy's effectiveness.

To form a more complete and nuanced picture, we discussed a range of uncorrelated metrics that capture different aspects of a strategy's success. Ultimately, you want to use a basket of metrics to assess your strategies or portfolio of strategies to maximise the chances of success.

There are a few other concepts when it comes to evaluating strategies that were not covered in the article, for example, capacity - measuring the highest AUM that can deliver a target risk-adjusted performance without suffering from performance decay. We will publish a future article covering capacity in detail.