Search
× Search
Thursday, January 9, 2025

Archived Discussions

Recent member discussions

The Algorithmic Traders' Association prides itself on providing a forum for the publication and dissemination of its members' white papers, research, reflections, works in progress, and other contributions. Please Note that archive searches and some of our members' publications are reserved for members only, so please log in or sign up to gain the most from our members' contributions.

The Predictive Power of Sharpe Ratios

photo

 Greg Kapoustin, Principal at AlphaBetaWorks

 Thursday, September 8, 2016

To select superior strategies one must first abandon popular non-predictive investment performance metrics:


Print

9 comments on article "The Predictive Power of Sharpe Ratios"

photo

 Alex Krishtop, Consultant at Edgesense Solutions. Mentor at Algorithmic Traders Association

 Friday, September 9, 2016



What predictive power is in returns divided by their standard deviation? Is it greater than the predictive power of, say, standard error? Or maybe variance? Or simply mean?


photo

 Kilian Mie, Vice President at Goldman Sachs

 Sunday, September 11, 2016



If the question is 'How good will my trading strategy perform out of sample?', this might be of interest: https://blog.quantopian.com/using-machine-learning-to-predict-out-of-sample-performance-of-trading-algorithms/


photo

 Marc Verleysen, founder at TSA-Europe -systematic trading and money management

 Monday, September 12, 2016



and what have we learnt ? Absolutely nothing. Sharpe has no predictive value (not even as a contra indicator). Sometimes, people write "scientific" papers just to "write scientific papers". Or .. the disconnect between statisticians/scientists and the real world illustrated


photo

 Greg Kapoustin, Principal at AlphaBetaWorks

 Monday, September 12, 2016



To Alex's point, the correlograms of Sharpe Ratios and nominal returns (charts 1 and 2) do look very similar. Though this result may be intuitive to many in this group, it is stunning how many continue to rely on these and similar dangerous metrics.

Since Sharpe ratios show similar reversion to nominal returns, they did appear to offer a contra indicator in our test. You can see a distinct AR signature in the correlogram/autocorrelation plot 1.


photo

 Alex Krishtop, Consultant at Edgesense Solutions. Mentor at Algorithmic Traders Association

 Tuesday, September 13, 2016



Kilian, to answer your question it is absolutely insufficient to employ only numeric methods. I tried to suggest an idea of the methodology which could give at least a rough idea about the future performance here: http://www.futuresmag.com/2016/01/25/building-robust-strategies


photo

 Chang Min (Leo) Chu, Quantitative Associate at Symphony Asset Management

 Wednesday, September 14, 2016



What is working then?


photo

 Zubair Badar, TopFundManagers.com | Quant Researcher & Developer

 Monday, September 19, 2016



Sharpe ratio indicates authenticity of your strategy logic, when logic fails, Sharpe ratio will automatically fail, so why not focus and work on primary root which is strategy logic instead of Sharpe ratio of strategy logic?


photo

 Zubair Badar, TopFundManagers.com | Quant Researcher & Developer

 Monday, September 19, 2016



If tree branches are effected then check if there is something wrong with root where branches depend.


photo

 Greg Kapoustin, Principal at AlphaBetaWorks

 Tuesday, September 20, 2016



If you have to rely on a popular ratio, Information ratio is a decent choice. Unlike the standard Sharpe ratio is has positive predictive value: http://abwcharts.com/2016/08/17/the-predictive-power-of-information-ratios/ Plus, the better you specify a benchmark, the better information ratio gets at measuring active performance.

Note that after its original publication even William Sharpe advised modifying Sharpe ratio to a metric similar to information ratio: http://web.stanford.edu/~wfsharpe/art/sr/sr.htm

Please login or register to post comments.

TRADING FUTURES AND OPTIONS INVOLVES SUBSTANTIAL RISK OF LOSS AND IS NOT SUITABLE FOR ALL INVESTORS
Terms Of UsePrivacy StatementCopyright 2018 Algorithmic Traders Association