Two strategies can make the same money and still feel completely different to live with. One bleeds a little every few weeks. The other runs smoothly for months and then drops 15% in one ugly stretch. The Burke ratio is a risk-adjusted return measure built for exactly this gap: it penalises deep drawdowns far more than shallow ones, so a single big hole in the equity curve hurts the score more than a string of small dips.
If you already track the Sharpe or Calmar numbers for your trading system, this is one more lens worth adding. Here is how it works, where it came from, and a worked example that shows it flipping a ranking that another popular ratio gets wrong.
In this article
What the Burke Ratio Measures
The idea came from Gibbons Burke, who suggested it in a 1994 article titled “A Sharper Sharpe Ratio”. The Sharpe ratio divides a portfolio’s return above the risk-free rate by its standard deviation. Burke kept the numerator and swapped the denominator. Instead of volatility, he used drawdowns, and specifically the square root of the sum of the squares of each drawdown.
In his paper “How Sharp Is the Sharpe Ratio?”, performance measurement specialist Carl Bacon describes the purpose plainly: squaring each drawdown penalises major drawdowns as opposed to many mild ones. That squaring step is the whole trick. A 2% dip contributes 4 to the sum. A 15% fall contributes 225. The big one dominates.
The formula, as written in the documentation for the BurkeRatio function in the R package PerformanceAnalytics, is:
Burke ratio = (rP − rF) ÷ √(D1² + D2² + … + Dd²)
- rP is the portfolio or strategy return over the period.
- rF is the risk-free rate over the same period.
- Dt is the t-th drawdown, measured as a positive percentage fall from a peak.
- d is the number of drawdowns you include.
Higher is better. A higher Burke ratio means more excess return per unit of drawdown pain, with the deepest drawdowns weighted the most.
Why Drawdowns Instead of Volatility
Standard deviation treats an unexpected 4% up day the same way it treats a 4% down day. Traders don’t. Nobody loses sleep over a surprise rally. What breaks people, and what breaks accounts, is the peak-to-trough fall in capital. Bacon makes the same point about hedge funds: it is easy to see why investors there might prefer the maximum loss from peak to valley as their measure of risk. Calmar, Sterling and Burke all start from that idea.
Each of these ratios reads drawdowns differently, though. Bacon’s paper lays out the family:
- Calmar ratio, suggested by Terry Young in 1991, uses only the single maximum drawdown.
- Sterling ratio replaces the maximum drawdown with the average drawdown.
- Burke ratio uses the square root of the sum of squared drawdowns.
- Martin ratio (also called the Ulcer Performance Index) is similar to Burke but uses the Ulcer Index, which also cares about how long drawdowns last.
Calmar looks at one data point and ignores everything else. Sterling looks at all drawdowns but averages them, and averaging has a nasty side effect: lots of tiny dips can dilute one big fall until it almost disappears. Burke sits in between. It uses every drawdown you feed it, but the squaring keeps the deep ones loud.
Averaging drawdowns lets seven small dips hide one 15% hole. Squaring them doesn’t.
Worked Example: Two Strategies, Same Return
Let’s make this concrete with two hypothetical swing trading strategies on Indian large-caps. These numbers are illustrative assumptions, not market data. Both strategies return 18% over a year, and we’ll assume a risk-free rate of 6% for the same year, so the excess return for each is 12 percentage points.
Strategy A has six moderate drawdowns during the year: 4%, 5%, 3%, 6%, 4% and 5%.
Strategy B has eight drawdowns: seven small dips of 1% each and one deep fall of 15%.
Step 1: The Sterling view (average drawdown)
Strategy A’s drawdowns add up to 27, so the average across six is 4.5%. Its Sterling ratio is 12 ÷ 4.5 = 2.67.
Strategy B’s drawdowns add up to 22 (seven 1s plus 15), so the average across eight is 2.75%. Its Sterling ratio is 12 ÷ 2.75 = 4.36.
On this measure, B looks clearly better. Its average dip is smaller. But that average is flattering, because the seven tiny dips drag it down and hide the 15% hit.
Step 2: The Burke view (square root of summed squares)
For Strategy A, square each drawdown: 16, 25, 9, 36, 16 and 25. The sum is 127. The square root of 127 is about 11.27. So the Burke ratio is 12 ÷ 11.27 = 1.065.
For Strategy B, the seven 1% dips square to 1 each, so they add 7. The 15% fall squares to 225. The total is 232, and its square root is about 15.23. The Burke ratio is 12 ÷ 15.23 = 0.788.
The ranking flips. On Burke, A beats B by a decent margin, because the 15% drawdown on its own contributes 225 of B’s 232 total. The small dips barely move the needle, which is how it should be.
Step 3: Cross-check with Calmar
Calmar only looks at the worst drawdown. For A, that’s 6%, so Calmar is 12 ÷ 6 = 2.0. For B, it’s 15%, so Calmar is 12 ÷ 15 = 0.8.
Calmar agrees with Burke here. The difference is that Calmar would have given B exactly the same score if it had only one drawdown, the 15% one, and no small dips at all. Burke would not: B’s score would barely change, but a strategy with two 15% falls would score noticeably worse than one with a single 15% fall, while Calmar would rate them identically.
Restricting the Burke Ratio to the Largest Drawdowns
Bacon notes that, as with the Sterling ratio, you can restrict the Burke calculation to a set number of the largest drawdowns. Doing so stops the long tail of tiny dips from mattering at all.
Using only the three largest drawdowns in our example:
- Strategy A: 6%, 5% and 5%. Squares add to 36 + 25 + 25 = 86. The square root is about 9.27, so Burke is 12 ÷ 9.27 = 1.294.
- Strategy B: 15%, 1% and 1%. Squares add to 225 + 1 + 1 = 227. The square root is about 15.07, so Burke is 12 ÷ 15.07 = 0.796.
A’s score rises because we dropped three of its drawdowns. B’s barely changes, because nearly all of its denominator was the one deep fall anyway. The gap widens. If you compare systems this way, use the same number of drawdowns for every strategy, or the comparison is meaningless.
The Modified Burke Ratio
The PerformanceAnalytics documentation also describes a modified version. Instead of the plain sum of squared drawdowns, the denominator is the square root of that sum divided by n, the number of observations in the whole return series. In practice this equals the Burke ratio multiplied by the square root of n.
Say both our strategies were measured on 12 monthly observations. The modified Burke ratio for A would be 1.065 × √12, or about 3.69. For B, 0.788 × √12, or about 2.73. The ranking does not change between two strategies measured over the same number of periods. Where the modified version helps is when you compare series of different lengths, because a longer record naturally collects more drawdowns and would otherwise look worse.
The Burke ratio doesn’t ask how bumpy the ride was. It asks how deep the potholes were.
How to Calculate It for Your Own Trading
You don’t need special software. A spreadsheet with your daily or weekly equity values is enough.
- Build a running peak. In a column next to your equity value, keep the highest equity reached so far.
- Mark each drawdown. A drawdown starts when equity falls below the running peak and ends when a new peak is made. Record the deepest percentage fall within each such episode.
- Square and sum. Square each drawdown percentage and add them up. Decide up front whether you’re using all drawdowns or only the largest few.
- Take the square root. That is your denominator.
- Compute excess return. Subtract the risk-free return for the same period from your strategy’s return. For Indian traders, a Treasury bill yield for the same period is one option. Whatever you pick, be consistent.
- Divide. Excess return over the denominator gives the Burke ratio.
One practical caution: how you define a drawdown changes the answer. If you measure on daily closes, you’ll find more and shallower episodes than on weekly closes. Pick one convention and stick with it across every system you compare.
Where the Burke Ratio Falls Short
It is not a magic number, and it has real blind spots.
- It ignores duration. A 10% drawdown that recovers in a week and one that takes a year to recover contribute exactly the same 100 to the sum. If time underwater worries you more, the Martin ratio, built on the Ulcer Index, is the better fit.
- It needs history. A system that hasn’t been through a bad market yet can show a flattering Burke ratio simply because its deep drawdown hasn’t happened. Short backtests are the usual culprit.
- Definitions vary. All drawdowns, top three, top five, modified or not. Two people quoting a “Burke ratio” may not be quoting the same thing. Always ask how it was calculated.
- It may not change your ranking much. Bacon points to a 2006 study by Eling and Schuhmacher, “Does the Choice of Performance Measure Influence the Evaluation of Hedge Funds”, which concluded that most of these measures are highly correlated and do not lead to significantly different rankings. Our example was built to show a flip; in real portfolios, the ratios often agree.
Bacon’s own take is useful here. He argues the real question is not which ratio is “best” in the abstract, but which measure of risk reflects what the investor actually cares about. If a single deep fall is what would make you abandon a strategy, the Burke ratio speaks directly to that fear.
Using It Alongside Other Risk Measures
We’d treat the Burke ratio as one column in a small dashboard rather than a verdict. Pair it with your maximum drawdown, which tells you the worst single fall, and the Sterling ratio, which tells you about typical dips. When Sterling looks great but Burke looks weak, that is your signal that one or two deep drawdowns are being averaged away. Dig into those episodes before you size the strategy up.
And whatever ratio you use, it only describes the past. Position sizing and stop losses are what actually limit the next drawdown, which is why the broader framework in our guide to risk management in trading matters more than any one statistic.
Sources: PerformanceAnalytics R package documentation, BurkeRatio (CRAN); Carl Bacon, “How Sharp Is the Sharpe Ratio? Risk-Adjusted Performance Measures” (PDF). The worked example uses illustrative assumptions, not market data.
Disclaimer: This article is for information and education only and is not investment advice. Please do your own research or consult a SEBI-registered investment adviser before making any investment decision. Past performance ratios do not guarantee future results.





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