PT - JOURNAL ARTICLE AU - Campbell R. Harvey AU - Yan Liu ED - Moore, Howard TI - Practical Applications of Backtesting AID - 10.3905/pa.2016.3.4.143 DP - 2016 Apr 30 TA - Practical Applications PG - 1--4 VI - 3 IP - 4 4099 - https://pm-research.com/content/3/4/1.2.short 4100 - https://pm-research.com/content/3/4/1.2.full AB - Backtesting Campbell R Harvey Yan Liu The claimed performance of new trading strategies often looks too good to be true—and indeed, in many cases, the good performance is a result of data mining. When implementing the strategy in the real world, practitioners routinely make some corrections to the backtests by haircutting the Sharpe ratio by 50%.If a large number of strategies have been tested and a modest Sharpe ratio resulted, one should haircut the result to zero. But if a strategy is truly outstanding, why decrease the Sharpe ratio by a full 50%? “In that case, it seems more reasonable to take just a little off the top,” Cam Harvey says in an interview with Institutional Investor Journals .TOPICS: Statistical methods, portfolio management/multi-asset allocation