wilmott.com

I have a very basic problem for performance measure usually used in finance.The very basic performance measure is :(measure expected return-benchmark return)/measure of riskThe problem I have is how you distinguish low risk-low return from high risk-high return with numbers such as sharpe ratio ?For me when a guy disclaimes his sharpe ratio for his strategy it provides not enough info.

Yeah, even apart from the question of whether the past has any implication for the future, performance measurement is quite an involved subject, certainly never captured by a single number.There are various ways of measuring returns: time-weighted vs. internal rate of returnThere are issues of annualization, compounding, averaging, and selection of time horizons. There are various ways of measuring risk: std. deviation, semi-deviation, etc -- of course, you should look at the entire distribution of returns. There are issues of suitable benchmarks, universe comparisons, perfomance attribution, and sector analysis.There are problems with survivor bias, selective disclosure, closed accounts, incubating funds, etc.So, yeah, by itself the sharpe ratio tells you virtually nothing.Even after a lengthy 20 page performance analysis report, laboriously discussing all the items in my list above,you still can know very little if you don't know the details and the rationale of the strategy. Even if you know the details and the rationale of the strategy, how much was really luck? A good marketing staff can invent a good story that goes with anything. Welcome to the world of investment management!

I don't completely disagree, but you can use a version of the classical Modigliani-Miller argument to say that any positive Sharpe ratio strategy can always be levered up or down to the level of risk you can tolerate (or the level of expected excess return you need).

Sure but what do people prefer ?Invest a lot in low risk low return strategy or invest a little in high risk high return strategy ? Moreover I believe that for large amount of investment the return decreases and the risk increases (market feedback effect).

a High Sharpe ratio says you can take those returns and leverage and given an assumption of normality then its fine. However, we all know that returns are anything but normal.A good way of measuring and ranking performance is the Omega ratio. It effectively is a measure of the upside to the downside and captures the who distribution of returns - assuming you have an empirical distribution.

In case anyone is interested, there's an Excel spreadsheet to calculate the Omega Ratio here: Omega Ratio in Excel.I prefer the Omega Ratio to other performance measures because it captures all the information in the empirical returns distribution, without place any assumptions on the data.

the Omega ratio can even evaluate lottery tickets as was explained by Shadwick and Keating.

the only thing I don't understand about either Sharpe or Omega ratio is that they are indifferent as to how many data points are used in estimating it. 2 trades or 2 million, 12 monthly returns or 250 daily returns - which one would be more trustworthy?forgive my trading illiteracy...

QuoteOriginally posted by: frenchXI have a very basic problem for performance measure usually used in finance.The very basic performance measure is :(measure expected return-benchmark return)/measure of riskThe problem I have is how you distinguish low risk-low return from high risk-high return with numbers such as sharpe ratio ?For me when a guy disclaimes his sharpe ratio for his strategy it provides not enough info.When you do sensitivity tests, most reward-risk measures point to the same areas. Choose whatever you like (I assume you are a trader/researcher, not an academic searching for some ultimate truth which does not exist).