Probabilistic Sharpe ratio: Probability of skill (Excel)
HTML-код
- Опубликовано: 12 сен 2022
- How to test the significance of Sharpe ratio outperformance when returns are not normally distributed? The Probabilistic Sharpe Ratio (PSR) developed by Bailey and Lopez de Prado (2012) is a very concise and powerful tool which is robust to non-normality and includes return skewness and kurtosis into its calculations. Today we are discussing the concepts behind the PSR and apply it in Excel.
Don't forget to subscribe to NEDL and give this video a thumbs up for more videos in Investment!
Please consider supporting NEDL on Patreon: / nedleducation
You can find the spreadsheets for this video and some additional materials here: drive.google.com/drive/folders/1sP40IW0p0w5IETCgo464uhDFfdyR6rh7
Please consider supporting NEDL on Patreon: www.patreon.com/NEDLeducation
Excellent Neidl! Takes into account Skewness and Kurtosis as well as factors out fatter tails and peaked excess Kurtosis (investment in high beta funds, activating specific sectors via active share measures selecting equities which outperform the overall benchmark return. All of which were core issues with the first (infamous) Sharpe Ratio. Portfolio Managers would Minimize denominator and maximize numerator. This can be factored as well into different attribution analysis via regression and even a probit regression making the probabilistic Sharpe Ratio the dependent variable.
Awesome video!
Suggestion: Deflated Sharpe Ratio, but I have a feeling you're already on it 😎
Your intuition is indeed correct, stay tuned for this video :)
You came through!! 🙏👌🤌🎉😎
Excellent video, awesome work Savva!
One question. If there is no tartget Sharpe ratio Than PSP is 50%? To be more accurate, if the tartget and the fund Sharpe ratio is the same Than PSP is 50%?
Hi Ferenc, and glad you liked the video! Yes, it is equal to 50%, as if you perform on par with the market there is equal chance you underperform and outperform skill-wise.
By the way, I'm quite new at stats, but doesn't excess kurtosis start at 3? (Above 3 == excess, I mean)
Hi Wilton, and thanks for the question! This is one of the most annoying things about statistical notations in my opinion :) Excess kurtosis is kurtosis minus three as three is the value of kurtosis for the normal distribution. So 0 is the value for the normal distribution that Lo (2002) implicitly assumes. The Excel KURT function gives excess kurtosis automatically, hence the formula :)
@@NEDLeducation ahhh I see, I did not know that. Thank you!
@@NEDLeducation I see. That explains it. But how about python scipy's kurtosis function?
I noticed you used kurtosis + 2 here which I don't understand. The paper uses kurtosis - 1
Hi Robert, and thanks for the excellent question! It depends on whether you are using kurtosis (then -1) or excess kurtosis (then +2). As Excel KURT function returns excess kurtosis, I adapted the formula for it this way.