Sharpe ratio

Is there an industry standard as to what is a good, bad, average and great number for the sharpe ratio?


I would defer to Science Trader or Jules Ellis for a more scientific answer but from my own experience and as a general rule I would say a Sharpe Ratio of 2 is highly desirable and an industry standard benchmark. I would also highlight though that such a ratio is only relevant and comparable across different systems if it is achieved over a statistically significant time period, which again although subjective I would speculate should be at a minimum one year, preferably two or more.

I believe Jules did an excellent piece recently on Sharpe’s tendency to flatter young systems and depreciate with age which you should check out for further reference.

Hope this helps.


These threads might be of your interest:

In general, I would consider a Sharpe ratio of 2 over a multi-year period quite good–in particular if it was for a 100% realistic system offered to the public on C2. The Grid shows for active systems with more than 2 years of history:

- 2 systems with Sharpe ratio > 2 (extreme-os, Tango)

- 2 systems with Sharpe ratio between 1.5 and 2 (Turbo Trader, Weekend Trader)

I once read LTCM had a Sharpe ratio of 4 before it collapsed… It’s a good example that the Sharpe ratio should always be understood in the context of the trading strategy. Similarly, it is possible to obtain high Sharpe ratios by selling options or averaging down, until your account blows out on a bad day.

Another problem you should be aware of, is that Sharpe ratios estimated on daily data with even a few hundred observations can still fluctuate a lot. A good example is Weekend Trader, which recently saw its Sharpe ratio jump within a week from 0.9 to 1.3 even though it was estimated over almost 500 trading days (see:

Finally, it seems a problem with many high-Sharpe trading strategies is that they are not very scalable. So, a large fund might prefer a scalable system with smaller Sharpe ratio to a system with high Sharpe ratio but limited scalability.

Thank you ST, I appreciate the time you took to assist me on this question.


There is no single answer to your question. It depends on capacity,

nature of the system, reputation of the manager, and a host of other


For a system with $1 billion or more in capacity run by a manager

who is a known quantity, a Sharpe Ratio over one is superb.

For a daytrading system of limited capacity run by a trader with a

history of blow-ups, even a nine Sharpe Ratio won’t attract any

real money.

However, if you can maintain your current performance, your

system will have more money than it can handle within the next

twelve months. Good luck.

Don’t take this as criticism, I will almost certainly subscribe to

your system at the end of the week.

You’re welcome. This might be of your interest as well:


In addition to what the others said, it makes a difference whether the Sharpe ratio is computed on monthly data or on daily data, and whether it is computed on the return rates or on the logarithm of the return rates. (All four versions are computed in the advanced statistics.) Moreover, many people compute it simply the wrong way, namely on portfolio values instead of portfolio returns. I think it would be inappropriate to apply the same standard to all versions. This being said, I do not know an “industry standard”, but Investopedia says

"To give you some insight, a ratio of 1 or better is considered good, 2 and better is very good, and 3 and better is considered excellent."

Please note that a Sharpe ratio larger than 0 reflects a profitable system. I think that it is wise to require that the Sharpe ratio is significantly positive. “Significant” in this context means that there are enough data to be reasonable sure that it is indeed positive, and that the positive value is not just a matter of luck. You can infer that from the lower bound of the confidence interval, which should be positive. I don’t believe that the assumptions (such as normality and independence) underlying the confidence interval are satisfied, but at least it gives you some rough indication of how unreliable the present estimate can be. A small but positive Sharpe can be significant if there are enough data. Indeed, I believe it may be better to have a 3-year old system with a Sharpe of 0.60 that is significantly positive than a 3-week old system with a Sharpe of 2 that is not significantly positive.

To offer a different perspective: The Sharpe ratio is also used in social sciences, where it is called “effect size” (Cohen’s d). The following conventions are used:

0.20 = small

0.50 = medium

0.80 = large

Now you might argue “here we are talking about fincance, not social sciences”, but on the other hand, is trading not just a kind of applied psychology? :slight_smile: Anyway, to give you an impression, consider this example from the book of Aron & Aron: An effect size of 0.50 corresponds about to the difference in heights between 14- and 18-year old girls.

Assuming a normal distribution, you can also relate the Shapre ratio to the probability of a loss. Sharpe ratio’s are usually annualized, i.e. they pertain to the annual return. These are the probabilities that the annual return is smaller than the return of the risk free asset, given various values of the Sharpe ratio:

Sharpe, probability

0.20, 0.42

0.50, 0.31

0.80, 0.21

1.00, 0.16

1.50, 0.07

2.00, 0.02

Hope these examples give you some feeling for the numbers.