Longer than 2 years and no short option. Sortino does not exist as I have not had any single negative month.
"Usually" is no mathematic argument.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Expectancy above .50??
- Thread starter shanoballs
- Start date
Longer than 2 years and no short option. Sortino does not exist as I have not had any single negative month.
"Usually" is no mathematic argument.
Quote from science_trader:
Longer than 2 years and no short option. Sortino does not exist as I have not had any single negative month.
"Usually" is no mathematic argument.
why do you need downmonths? i assume you had downDAYS so you can calc an annualised semideviation ...
i am used to discuss sharpe ratio with people who come up with a constructed chart that shows how flawed the concept is, yet this construction never happens in real life. and especially sharpes north of 3 show not too much difference in deviation. at least that has been my experience. and sortino always suffers in my opinion for getting flawed by relative few observations. tends to fit to specific trades ... i remember i backtested lumber in the nineties (what a market!) and there were a handful of trades you had to catch .. that was it.
Quote from man:
well, sharpe tells much more about an equity curve than does profit factor. and if you don't want to penalize upward vola then
use sortino. nevertheless, profit factor IS flawed.
Profit factor only seems flawed if you don't understand what it measures. That's like saying a hammer is flawed because you can't drive screws with it.
The Sortino Ratio fails a basic test case for the effectiveness of performance measures BTW... see paper at: http://www.edge-fund.com/Shadwick2004.pdf
Quote from Trader666:
Profit factor only seems flawed if you don't understand what it measures. That's like saying a hammer is flawed because you can't drive screws with it.
The Sortino Ratio fails a basic test case for the effectiveness of performance measures BTW... see paper at: http://www.edge-fund.com/Shadwick2004.pdf
well, while i agree that omega is a very reasonable concept, i tend to disagree that the lottery test is a reasonable proxy for trading, thus i doubt it is by any means a "basic test case for the effectiveness of performance measures". yet it is great marketing, i mean, if it cannot even get lottery right ...
Quote from man:
well, while i agree that omega is a very reasonable concept, i tend to disagree that the lottery test is a reasonable proxy for trading, thus i doubt it is by any means a "basic test case for the effectiveness of performance measures". yet it is great marketing, i mean, if it cannot even get lottery right ...
Obviously you either didn't take the time to read the paper, or you didn't grasp it. So let me put it another way.
Sortino is flawed because it can fail at what it was designed to do.
PF is not flawed because it is simply total profits/total losses and doesn't try to be an entire distribution in a can. Clearly PF needs to be looked at in conjunction with other things, like number of trades and outliers.
Quote from man:
actually my point is that profit factor i neither a hammer nor a screwdriver but just ... a word. and you can't build anything with words alone.
You have no point because you never gave any reason at all for PF being flawed. You just said it was. I, on the other hand, pointed you to an objective paper that mathematically showed why Sortino IS flawed but obviously it went over your head. Perhaps some day you'll realize that usually the simple things work best. Highbrow stuff is more useful for trying to impress the unwary on message boards.
The Sharpe ratio does indeed have it's limitations. Because it is a ratioed measure of average return to the standard deviation of the return, the sharpe ratio captures the statistics of the returns only as accurately as the standard deviation can be meaningfully applied to the data.
Let it be known that standard deviation is a measure of dispersion around the mean for NORMALIZED data. If the distributions of returns contain a skew or kurtosis, the standard deviation does not completely accurately capture the statistics of the return.
If you have 9 losing trades followed by 1 big winner that absorbs the loss and yields profit, the Sharpe ratio does not provide a good measure of evaluating the return. The data is not normally distributed, so another metric of evaluation is required.
RoughTrader
Let it be known that standard deviation is a measure of dispersion around the mean for NORMALIZED data. If the distributions of returns contain a skew or kurtosis, the standard deviation does not completely accurately capture the statistics of the return.
If you have 9 losing trades followed by 1 big winner that absorbs the loss and yields profit, the Sharpe ratio does not provide a good measure of evaluating the return. The data is not normally distributed, so another metric of evaluation is required.
RoughTrader
Quote from man:
why do you need downmonths? i assume you had downDAYS so you can calc an annualised semideviation ...
i am used to discuss sharpe ratio with people who come up with a constructed chart that shows how flawed the concept is, yet this construction never happens in real life. and especially sharpes north of 3 show not too much difference in deviation. at least that has been my experience. and sortino always suffers in my opinion for getting flawed by relative few observations. tends to fit to specific trades ... i remember i backtested lumber in the nineties (what a market!) and there were a handful of trades you had to catch .. that was it.
If you enjoy accounting so much to compute your daily returns... I prefer concentrating on trading ideas and automatic systems design.