Dalal Street Blogs provides information for educational purposes only and does not constitute investment advice. Please consult a SEBI-registered investment adviser before making any investment decisions.
CAUTION: Dalal Street Blogs publishes market news and educational content only. We do not promote, recommend or endorse any stock, and nothing here is a solicitation to buy or sell securities. Please consult a SEBI-registered adviser before investing.
Trading Strategy

Sterling Ratio: Judging Returns Against Average Drawdowns, With an Example

4003
By Dalal Street Blogs EditorialUpdated 9 min read

Two strategies can post the same return and still feel completely different to live with. One grinds higher with shallow dips. The other gives you a heart-stopping fall every year before it recovers. The sterling ratio is one of the older attempts to put a number on that difference: it divides return by drawdown, and in its original form it uses the average of yearly drawdowns rather than the single worst one.

That one design choice is the whole point of the ratio, and it is also where most of the confusion starts. Different textbooks, spreadsheets and software packages compute something they all call “Sterling”, and they don’t always agree. So we’ll walk through what the ratio is trying to measure, the versions you’ll actually run into, and a worked example where the choice of version flips which strategy looks better.

In this article
  1. What the sterling ratio measures
  2. The original formula and that odd 10%
  3. The versions you’ll actually meet
  4. How it relates to the Calmar ratio
  5. A worked example: two strategies, three years
  6. Using the sterling ratio in your own backtests
  7. Where the sterling ratio falls short
  8. A sensible way to use it

What the sterling ratio measures

At heart, it’s a reward-to-pain number. On top you put the return the strategy or portfolio earned, usually compounded and annualised. Underneath you put a measure of how deep it fell from its highs along the way. A higher ratio means you got more return for each unit of drawdown you had to sit through.

The Wikipedia entry on the Sterling ratio describes it as a measure of risk-adjusted return that looks at return over average drawdown, in contrast to the more commonly used maximum drawdown. The method it describes is simple enough to do by hand: for each calendar year, find the worst peak-to-trough fall inside that year, then average those yearly figures.

Why bother averaging? Because a single maximum drawdown is one data point. It can come from a freak month, a crash that hit every strategy at once, or a single bad trade. Averaging the yearly worst falls gives you a sense of how much pain is routine for the strategy, not just how bad its worst day on record was.

The original formula and that odd 10%

The version usually given as the original, credited on Wikipedia to Deane Sterling Jones, looks like this:

Sterling ratio = Compound annual return ÷ (Average annual drawdown + 10%)

The drawdown is entered as a positive number, the 10% is added to it, and the ratio comes out positive. That 10% looks arbitrary until you hear the reasoning. According to the same entry, the ratio was devised in 1981, when Treasury bills were yielding 10%. T-bills don’t have drawdowns, so a T-bill would score exactly 1.0 on this formula (10% return divided by 0% drawdown plus 10%). Anything scoring above 1.0, the thinking went, was offering a better risk-reward trade-off than parking money in bills.

We should be honest about how solid this history is. The Wikipedia page itself is a short stub and flags most of these statements as needing citations. Treat the 1981 backstory as the commonly told explanation, not settled fact. What matters for you as a user of the number is the mechanics: the 10% is a fixed cushion added to the denominator, and it was tied to a US interest rate from four decades ago.

The 10% in the formula isn’t a law of nature. It’s a cushion borrowed from a 1981 T-bill yield, and you should decide consciously whether to keep it.

The versions you’ll actually meet

Here’s where it gets slippery. At least three things travel under the same name.

1. Average yearly drawdown plus 10%

The original form above. It rewards consistency because one terrible year gets diluted by the calmer ones.

2. Maximum drawdown plus 10%

The R package PerformanceAnalytics defines it differently. Its documentation for CalmarRatio and SterlingRatio says both ratios are annualised return over the absolute value of the maximum drawdown, and that the Sterling version adds an “excess” amount to the maximum drawdown, traditionally and by default 10%. So if you run SterlingRatio() in R, you are not getting the average-drawdown version.

3. Excess return over average drawdown

Wikipedia also gives a Sharpe-style adjustment: (annual portfolio return minus the annual risk-free rate) divided by the average largest drawdown. Here the 10% disappears from the denominator, and a risk-free rate is subtracted from the top instead. For an Indian investor, that’s arguably the more sensible version, because you can plug in a rate that’s relevant to you rather than a 1981 US figure.

None of these is “wrong”. But two reports quoting a Sterling ratio of 0.8 might be measuring different things. Before you compare numbers from different sources, ask which drawdown (average or maximum), what cushion, and over how many years.

How it relates to the Calmar ratio

If you’ve read our explainer on the Calmar ratio, version 2 above will look familiar. The PerformanceAnalytics documentation puts them side by side: Calmar is annualised return over maximum drawdown, and Sterling (in that package) is the same thing with 10% added to the drawdown. The same page also notes it is traditional to compute these on a three-year return series, though the functions themselves don’t enforce any length.

The practical difference between the original Sterling and Calmar is the one we started with. Calmar asks, “How bad was the worst fall?” Original Sterling asks, “How bad is a typical bad patch?” Neither is complete on its own.

A worked example: two strategies, three years

Let’s make this concrete. The numbers below are hypothetical, made up purely to show the arithmetic. They don’t describe any real fund, index or strategy.

Suppose you’re comparing two swing-trading systems, each started with ₹10,00,000 and run for three years.

  • Strategy A compounds at 18% a year. Its worst peak-to-trough falls in each of the three years were 12%, 20% and 7%.
  • Strategy B compounds at 22% a year. Its yearly worst falls were 9%, 31% and 8%.

First, the ending values. A at 18% a year for three years turns ₹10,00,000 into about ₹16,43,032. B at 22% gets to about ₹18,15,848. On raw return, B wins comfortably.

Quick reminder on how one of those yearly drawdowns is measured. If B’s account hit a high of ₹12,50,000 during a year and then slid to ₹10,00,000 before recovering, that’s a fall of ₹2,50,000 from the peak, or 20% of ₹12,50,000. You measure from the highest point reached, not from where the year started.

Step 1: average and maximum drawdowns

  • A: average of 12, 20 and 7 is 13%. Maximum is 20%.
  • B: average of 9, 31 and 8 is 16%. Maximum is 31%.

Step 2: run each version

Measure (hypothetical data)Strategy AStrategy BLooks better
Original Sterling: return ÷ (average DD + 10%)18 ÷ 23 = 0.7822 ÷ 26 = 0.85B
Calmar: return ÷ maximum DD18 ÷ 20 = 0.9022 ÷ 31 = 0.71A
Sterling, max-DD form: return ÷ (maximum DD + 10%)18 ÷ 30 = 0.6022 ÷ 41 = 0.54A
Excess-return form, assumed 6.5% risk-free rate: (return − 6.5) ÷ average DD11.5 ÷ 13 = 0.8815.5 ÷ 16 = 0.97B

The 6.5% risk-free rate in the last row is an assumption for the example, not a quoted market rate. Swap in whatever rate you actually use.

Step 3: read what just happened

Look at that last column. The two average-drawdown versions prefer B. The two maximum-drawdown versions prefer A. Same data, opposite verdicts.

The reason is B’s middle year. A 31% fall is brutal, and any measure built on the worst drawdown punishes it hard. The average-based measures spread that one bad year across three, and B’s higher return carries it over the line.

Which verdict is right depends on you. If you know you’d abandon a system after a 30% drop (most of us would, honestly), the maximum-based numbers are telling you something important that the original Sterling ratio smooths over. If you’re judging a strategy over a long run and you’re confident you can sit through an ugly year, the average-based view is fairer to B.

When two versions of the same ratio disagree, the disagreement is the insight: it tells you one bad year is driving the story.

Using the sterling ratio in your own backtests

A few habits make the number far more useful.

Say which version you used. Write it next to the result, every time. “Sterling 0.85 (average yearly DD + 10%, 3 years)” is a usable figure. “Sterling 0.85” on its own is not.

Use enough years. With only one or two calendar years, the “average” drawdown is barely an average at all. The three-year convention mentioned in the PerformanceAnalytics notes is a sensible floor. More is better if your data allows it.

Watch partial years. If your backtest starts in October, that first “year” only has three months to produce a drawdown, so it will usually look shallow and flatter the average. Either use full calendar years or be aware of the bias.

Keep the frequency consistent. Drawdowns measured on daily closing values will be deeper than ones measured on month-end values, because month-end data hides intra-month dips. Compare strategies only on the same frequency. The R documentation notes the usual annualisation scales of 252 for daily data and 12 for monthly, which is a reminder that your input frequency shapes the output.

Think about the cushion. If you keep the 10%, you’re implicitly benchmarking against a 10% riskless return. Whether that’s a high or low bar depends on the rate environment you’re in. The excess-return version lets you choose the benchmark explicitly, which is cleaner.

Where the sterling ratio falls short

It’s a blunt tool, and it helps to know its blind spots.

  • It ignores how long drawdowns last. A 15% fall that recovers in three weeks scores the same as a 15% fall that takes two years to claw back. If time spent underwater matters to you, look at a drawdown measure like the Ulcer Index alongside it.
  • Calendar years are arbitrary. A single drawdown that starts in November and ends in March gets split across two years, and each piece looks smaller than the whole.
  • It says nothing about day-to-day volatility. For that you’d look at the Sharpe ratio or the Sortino ratio.
  • Even its software authors aren’t fans. The PerformanceAnalytics documentation states plainly that its author considers newer measures, such as Sortino’s upside potential ratio or Favre’s modified Sharpe ratio, to be better and preferable to Calmar or Sterling.

That last point is worth sitting with. The Sterling ratio is easy to compute and easy to explain, which is why it survives. It’s a fine first filter. It’s a poor final judge.

A sensible way to use it

Here’s how we’d fit it into a strategy review. Compute the original Sterling ratio and the Calmar ratio together. If they agree, you’ve got a reasonably clear picture. If they disagree, as in our example, go look at the year that’s causing it: what happened, was it a market-wide event, and would you genuinely have stuck with the system through it?

Then pair it with position sizing. A strategy whose typical yearly drawdown is 16% can still be run at half size so the account-level pain is smaller. Our guide to risk management in trading covers how sizing and stop-losses change the drawdowns you actually experience, which matters more than any ratio printed in a backtest report.

Used that way, the sterling ratio does its job: it pushes you to think about the routine pain a strategy will put you through, not just the return it promises.

Disclaimer: This article is for educational purposes only and is not investment advice. The strategies and figures in the worked example are hypothetical. Past performance and backtested results do not guarantee future returns. Please consult a SEBI-registered investment adviser before making investment decisions.

Leave a Reply

Join the conversation

Commenting is open to registered readers. Sign up with just your email — no password needed, and it takes under a minute.

Sign up to comment Log in