2019/05/01 by Adriano Koshiyama, Koshiyama, Adriano, Nick Firoozye +1
Decision Sciences · Economics, Econometrics and Finance · #60G10 #62E15 #62F99 #62P05 #91G70 #91G80 #Complex Systems and Time Series Analysis #FOS: Economics and business #Forecasting Techniques and Applications #Portfolio Management (q-fin.PM) #Risk Management (q-fin.RM) #Stock Market Forecasting Methods #Trading and Market Microstructure (q-fin.TR)
paper · pdf · doi:10.48550/arxiv.1905.05023
openalex publication_date 2019/05/01 · openalex created_date 2021/02/01 · openalex updated_date 2026/07/28
Systematic trading strategies are rule-based procedures which choose\nportfolios and allocate assets. In order to attain certain desired return\nprofiles, quantitative strategists must determine a large array of trading\nparameters. Backtesting, the attempt to identify the appropriate parameters\nusing historical data available, has been highly criticized due to the\nabundance of misleading results. Hence, there is an increasing interest in\ndevising procedures for the assessment and comparison of strategies, that is,\ndevising schemes for preventing what is known as backtesting overfitting. So\nfar, many financial researchers have proposed different ways to tackle this\nproblem that can be broadly categorised in three types: Data Snooping,\nOverestimated Performance, and Cross-Validation Evaluation. In this paper, we\npropose a new approach to dealing with financial overfitting, a\nCovariance-Penalty Correction, in which a risk metric is lowered given the\nnumber of parameters and data used to underpins a trading strategy. We outlined\nthe foundation and main results behind the Covariance-Penalty correction for\ntrading strategies. After that, we pursue an empirical investigation, comparing\nits performance with some other approaches in the realm of Covariance-Penalties\nacross more than 1300 assets, using Ordinary and Total Least Squares. Our\nresults suggest that Covariance-Penalties are a suitable procedure to avoid\nBacktesting Overfitting, and Total Least Squares provides superior performance\nwhen compared to Ordinary Least Squares.\n