Craig: First, I wouldn’t complain about a Sharpe over 2.0 Win% is irrelevant to Sharpe, just the “shape” of the equity curve. One deficiency in Sharpe is that your very recent run up can hurt your Sharpe, as it will tend to increase the std dev of your returns. Having said all of that, be aware that if we are indeed using monthly stats for Sharpe computation, the relatively short time the system has been on C2 can potentially give erratic results. But, again, don’t be disappointed with a Sharpe over 2.0
Hans.
MK, I strongly prefer Sharpe ratio calculated on daily equity changes.
The main reason is that by calculating on a monthly basis, any
system with a track record less than three months in length will
not show up on a Sharpe Ratio sort.
MK:
when you says Sharpe is now calculated monthly (instead of daily), what is meant by that? That all the systems on the same day each month have Sharpe calculated? In that case it can become highly inaccurate until the next monthly calculation. Or is it a rolling one-month calculation (calculated daily). I suspect the first scenario is true, because (for example) Million Dollar ATM has one of the higher ones, a Sharpe of over 6, but has a closed trade negative balance. Should any system that has lost money have a positive Sharpe? I think not.
Sharpe is calculated each day. But the numbers that are analyzed are monthly returns.
If the sharpe ratio was of any real value there would be no controversy here.
Everyone would get it.
Million Dollar ATM has one of the higher ones, a Sharpe of over 6, but has a closed trade negative balance. Should any system that has lost money have a positive Sharpe? I think not.
Why not? This system is very loyal in achieving those negative results. SR does not distinguish between whether a system has positive or negative results. It just measures how loyal it is to those principles that lead it to achieving those results, regardless of whether they are negative or positive.
Of course, whether this is an example of virtue in a system should now be crystal clear to all and the fallacy of using (or misusing) this measure to compare across all systems regardless of their virtue instead of using it to compare two similar virtuous systems should also be clear (I hope).
SR = (profit - risk-free profit) / SD, so it should be negative when the profit is negative (assuming the risk-free profit is not negative).
Jules
You are right. I forgot the fact that StDev cannot be negative. So if that is the case then, one could argue that find a typical investment that makes money with perfect consistency (for eg., a CD that returns slightly above risk free return that would have a SR of near infinity) and just increase the amount invested (position size) proportionally to get uncharacteristic returns. But, that would not be practical, because investors typically do not have large sums to invest.
Given two portfolios with the same average arithmetic return, the one with the lower StDev (or volatility) will have a higher compounded rate of return. SR facilitates only in comparing these two systems that have the same average arithmetic return and because of the volatility of the portfolio, the portfolio with lower volatility will have a higher geometric average rate of return.
What one needs is an investment that returns far above the risk free return (uncharacteristic returns), but that does not require large sums to invest. SR does not facilitate in identifying such an uncharacteristic investment and so it is not a suitable measure to compare (regardless of their returns) across systems that have those characteristics.
ps: Still, I do not have an explanation as to why the SR is positive for this particular system.
Please note I accidentally posted this as a new thread, not a reply…my apologies.
I thought I would chime in with a couple of things related to system statistics. Please forgive me if what I say is repeated elsewhere on these boards; this thread was recommended to me and I have not visited this site is many months.
First, I enjoyed Jules’s post earlier on system statistics; I can only assume you are very well respected around here.
With regards to the normality arguement, I wholeheartedly agree with avoiding the assumption of normal distributions whenever possible, especially when discussing returns in financial markets. In fact, when discussing common global financial markets, returns are most certainly not normal. There is usually very significant leptokurtosis on high-frequency data; as the frequency increases, the kurtosis usually does as well. Also, stock market returns usually have significant skewness.
As an example, I did a very quick and crude test on weekly spot SP 500 data. The data showed kurtosis of 10 and skewness of -.33, with a Jarque-Berra normality statistic of 82 (1% critical value is 9.21).
Trading systems can be all over the spectrum with regards to return distributions. A very consistent mean-reversion strategy may have very strong kurtosis. A long term trend system, on the other hand, may have less kurtosis but very strong skewness. Either way they’re not normal.
I really think Sortino ratios are a far better choice. The point of using a Sharpe ratio is to assess the return/risk tradeoff; any application of the ratio when the distribution of returns is not symmetrical is misleading.
