CEU Electronic Theses and Dissertations, 2020
| Author | Szemán, Krisztián |
|---|---|
| Title | A Statistical Test to Control Excessive Parameter Fitting of Trading Strategies |
| Summary | Abstract. Checking the past performance of many variants of a trading strategy with different parameters and selecting the one with the highest observed Sharpe Ratio is a common, nevertheless ill-advised practice when optimizing trading strategies. While the portfolio with the highest Sharpe Ratio might be viewed as the best choice of a risk-averse investor (e.g. according to Modern Porfolio Theory), true Sharpe Ratios cannot be deducted from backtests. Observed Sharpe Ratios can only estimate them through an error. Among several similar trading strategies, having one with the highest observed Sharpe Ratio is inevitable, but whether it is due to randomness or actually enhanced performance needs investigation. In my thesis, I propose a hypothesis test that helps determine the answer to this question. This test is applied to a dataset containing the performance of 18 of iShares used by previous authors for two different timeframes. I conclude that the iShare with the highest realized Sharpe Ratio does not exhibit a significantly better performance than the average iShare in neither of those periods. |
| Supervisor | Ferenci, Tamás Dr. |
| Department | Mathematics MSc |
| Full text | https://www.etd.ceu.edu/2020/szeman_krisztian.pdf |
Visit the CEU Library.
© 2007-2021, Central European University