About the Sharpe Ratio and Expectancy in the analyze of intraday trading systems

Sharpe Ratio is one of the most important ways to look at the quality of an investment strategy. It can be used for any kind of time frame since it uses annualised data.

The key idea is to combine return and risk. Return is measurd as annualised return (which can be derived out of monthly, daily or minute data) minus risk free rate of interest (=short term US rates), the whole divided by the annualised standard deviation of returns.

The higher the Sharpe ratio the better. Good readings on daily data are above 1. Good readings on intraday can exceed that by far. Best I've seen, is about 5.0.

 

simple excel sheet.

 

We were expected to keep the RRR above 10. It was computed daily and was reviewed monthly.

Ex. Jan.

profit + 2.0%
Max DD -2.0%

RRR = annualized return / Max DD
RRR = 2%*12 / 2.0
RRR = 12

Feb.

Profit + 2.0
Cumm. Profit +4.0
Max DD -2.5 from beginning of year to present

RRR = 4%*6/2.5
RRR = 9.6 and we'd get a visit from a risk control guy.

 

Sharpe ratio is very misleading as it penalises upside volatility - i.e. it treats upside and downside variance equally. Yet, upside volatility is actually good. Thus, a much useful tool, which is underused - largely because people don't understand it...is the Sortino ratio which uses semivariance (or bad/downside volatility)

 

Interesting. I had always heard this called the "semi-Sharpe ratio". Thanks for defining the Sortino ratio, Globaltrader.

 

Globaltrader
you are perfectly right IMO. from my experience with hedge fund equity curves i would say they are rather seldom significantly different in their upside and their downside vola. usually what you get is what you risk. meaning that what you can expect to gain within a period is seldom much higher than what you should be prepared to loose for the same period.
of course there are exemptions, but as I said, they are rare. (of course this can be different for very short equity curves.) thus altough sortino ratio is clearly the more precise tool it will usually not tell a different story than sharpe.

IMO the most important thing is to use any kind of risk measure to evaluate a return figure. it could as well be a MAR ratio, where you divide return by max draw down or an average of the highest x draw downs.


peace