Performance Metrics: The Sharpe Ratio and the Sortino Ratio

For a systematic trader, effectively managing risk in a trading system is crucial to the long-term viability of the system. The most obvious performance metric for testing a strategy is its profitability— we all want a winning system. But alone, the historical returns of a strategy don’t provide enough information. For example, say you are trying to decide between two strategies with historical returns of 7% and 5%, respectively. Which would you choose? Without any other information, the easy answer is the first strategy because it has a higher historical return. However, what if you then found out the first strategy had a drawdown of 50% while the second strategy had a drawdown of 20%? Knowing this, your answer may change.

The Sharpe Ratio

One way to measure  a strategy’s risk compared to its reward is to calculate its Sharpe Ratio. Initially termed the reward-to-variability ratio by its namesake William F. Sharpe ¹, the Sharpe Ratio indicates the average return per unit of risk in excess of the risk-free rate of return. A high Sharpe Ratio is generally more attractive because it indicates a higher risk-adjusted return. The Sharpe Ratio can be calculated using the following formula ²:

Sharpe Ratio = (r – r_f) / σ, where:

r = expected return of strategy x
r_f = risk-free rate of return
σ = standard deviation of expected return of strategy x

Generally speaking, higher values are considered ‘good’ and low or negative values are considered ‘bad’. A high Sharpe Ratio indicates the strategy produces returns in excess of the risk-free rate, while a negative number indicates the strategy can be expected to perform worse than a risk-free investment, or if the risk-free rate is 0, is expected to produce negative returns. What is considered high or low may vary considerably by asset class and market conditions, so the Sharpe Ratio is often used as a comparative measure between two or more strategies.

While this metric can help evaluate risk when building and testing a trading system, the Sharpe ratio is not perfect. For one, it assumes returns are normally distributed and does not account for skew or kurtosis which may occur in a real market scenario. Additionally it is a measure of historical performance and therefore assumes that future market conditions would be similar to past conditions which we know is not always the case.

The Sharpe Ratio calculates the return per unit of risk, using standard deviation as the risk measure. However, because standard deviation is agnostic when it comes to direction, large positive returns will increase the standard deviation the same way as large negative returns. This can make the Sharpe Ratio misleading when comparing strategies with asymmetric distributions, i.e. a strategy where most of the volatility is positive vs. a strategy with volatility in both directions. Because higher standard deviations will “penalize” a strategy in the context of a Sharpe Ratio, this is an important consideration.

The Sortino Ratio

An investor may not wish to “penalize” a strategy for positive volatility. The Sortino Ratio attempts to correct for this by considering the standard deviation of negative returns, thus only penalizing the strategy for negative volatility/outcomes.

The calculation of downside volatility is a topic of some debate ³. By definition, standard deviation measures the dispersion of data around its mean. The typical standard deviation definition, when applied to only downside returns, measures dispersion around the average downside return. However, it can be more meaningful to modify the calculation of downside deviation so that it measures dispersion around a different target, like 0; in this way, we measure each return’s “distance” from 0 instead of from the average negative return.

This is the idea behind the research note from Rollinger and Hoffman (2013). In this note, Rollinger and Hoffman explain a calculation for the downside deviation measures which the dispersion of data below a selected center ⁴. Because volatility to the upside means profits, a trader may be happy to accept positive volatility and measure only negative volatility. The downside deviation allows for this by isolating only returns below the target return of the strategy. The formula is below:

Sortino Ratio = (r – r_{f )} / σ_d, where:

r = expected return of strategy x
r_f = risk-free rate of return
σ_d = downside deviation of expected return of strategy x

As with the Sharpe Ratio, the Sortino Ratio is most helpful when it is compared to that of another strategy or system. However it is important to use a constant target return in the calculation when comparing Sortino Ratios to get an accurate comparison. Whether you are using the Sharpe Ratio to calculate overall volatility or the Sortino ratio to measure downside volatility, these measurements can be added to your strategy as another means of evaluation.