Talk:Sharpe ratio/Archives/2012

some topics that confuse
1) many database entities use the geometric return when it should be the average...which leads to 2) If you have 10,20, or 30 years of data, why would you annualize the monthly numbers. Nobody has talked about what the appropriate number of completed years of performance that is necessary to average for return, and calculate the standard deviation on rather than using monthly numbers. Finally, is it not imperitive to use the geometric historical (equal to length of complete years of returns) tbill rate instead of the current rate? For very very low average rates of return, the sharpe ratio can become negative when it should be positive. For example, currently databases are using 5% as the RF rate. However, over the last 13 years it has averaged below 4%. If you are evaluating a return stream that has earned 4.8% then historically, you have outperformed the risk free rate but the sharpe would be negative. How can we accept the use of the current rate when it distortes history. —The preceding unsigned comment was added by 216.47.169.143 (talk) 19:44, 19 March 2007 (UTC).

from the horses mouth
http://www.stanford.edu/~wfsharpe/art/sr/sr.htm —The preceding unsigned comment was added by 216.47.169.143 (talk) 18:18, 27 March 2007 (UTC).

"Constant" vs. "Deterministic"
The note about the 1994 revision of the Sharpe ratio makes no sense. There are two Sharpe ratios -- one based on previous performance and one based on predictions of future performance in the next SINGLE cycle. It's wrong to combine them like this. Also, there is no note about the risk minimization of the Sharpe ratio. If returns are normally distributed (there are other distributions that will work too), the Sharpe ratio will minimize risk (the Sharpe ratio is effectively a z-score of a normal distribution). --TedPavlic 22:15, 4 May 2007 (UTC)

Reference Quality Issue
The cited article by Scholz is from a pretty low quality journal. Why was this particular article chosen? Is this a self-cite?

137.142.126.126 (talk) 21:02, 6 March 2008 (UTC)

Example problem?
In the example, the standard deviation of the S&P return is used to compute the ratio $$\sigma (R_{SP}) $$, but it should be the standard deviation of the S&P return minus the risk free return $$\sigma (R_{SP}-R_f)$$, right?Gpeilon (talk) 10:46, 21 May 2009 (UTC)
 * This seems to be a mistake in the formula and not the example. The denominator should be $$\sigma (R_{SP}) $$. The article specifically mentions the correct method.  "(this is often confused with the excess return over the benchmark return; the Sharpe ratio utilizes the asset standard deviation whereas the information ratio utilizes standard deviation of excess return over the benchmark, i.e. the tracking error, as the denominator.)."  But then proceeds to show the incorrect formula.

Does anyone disagree with fixing this?

Excess Returns
This is misleading, as one actually wants mean and stdev of excess returns, which means the subtraction of risk free comparators should be inside the operators.
 * Good catch, I had never looked that closely into the definition, but I just checked the paper that is the external link and it confirms you are correct. There may be a clearer way to word that than I did, so feel free to fix it if you can. - Taxman Talk 21:43, 15 February 2006 (UTC)
 * Actually it is not a good catch. The current formula is equivalent to the previous formula within the usual assumptions
 *  - C == , where V is some value, and C is a constant
 * --Gene s 05:58, 16 February 2006 (UTC)
 * The return on benchmark asset is not necessary a constant, i.e. the benchmark asset may or may not be risk free. The catch was good, making the formula more general..
 * It's my understanding the Sharpe ratio is supposed to be about a theoretical "risk-free" asset. The post-hoc statistical analysis that you do to determine what errors occurred as a result of reliance on the Sharpe ratio is aside from the theory. 68.96.49.125 (talk) 17:49, 17 December 2012 (UTC)

Lucidity
I do not understand the article, and there is NO practical information in it for me. The article does not even say if HIGH Sharpe Ratio is a reccommendation for an investor, or is it more encouraging to find a LOW Sarpe ratio asset. Indeed, I find the article in its present form completely useless. The article should be re-nemed, e.g. 'Theories of constructing the Sharpe Ratio' or something like that; the advantage would be that nobody would waste their time trying to find info in it which, apparently, is beyond its interest.110.164.241.105 (talk) 01:30, 3 April 2010 (UTC)
 * Wikipedia is not a financial advisor. this article describes the Sharpe Ratio for what it is. If there is source material showing a reason to prefer certain ranges of its values, it should be referenced and summarized. 68.96.49.125 (talk) 17:41, 17 December 2012 (UTC)

Information Ratio
This page says the Sharpe ratio is different from the information ratio because it's the information ratio applied to finance. But if you click through to the page for information ratio, it's all about finance, too, and gives a different distinction between information ratio and Sharpe ratio (one which seems to be the same as the distinction between the original and new Sharpe ratio). so the whole matter of how the Sharpe ratio is different from the information ratio is confused. that page needs to be edited to be about information theory with perhaps a section about the financial-jargon use of the term, and this one needs to be edited to state simply that the Sharpe ratio is the information ratio with particular financial parameters plugged into the formula. maybe. if i knew for sure i'd do it. someone who's dealt with this jargon directly will need to straighten it out. 68.96.49.125 (talk) 17:38, 17 December 2012 (UTC)