Technical Analysis, Spread Trading and Data Snooping Control
International Journal of Forecasting, Volume 39, Issue 1, pp 178-191, January–March 2023, DOI: 10.1016/j.ijforecast.2021.10.002
70 Pages Posted: 8 Mar 2018 Last revised: 13 Jan 2023
Date Written: June 17, 2020
Abstract
This paper utilizes a large universe of 18,410 technical trading rules (TTRs) and adopts a technique that controls for false discoveries to evaluate the performance of frequently traded spreads using daily data over 1990–2016. For the first time, the paper applies an excessive out-of-sample analysis in different subperiods across all TTRs examined. For commodity spreads, the evidence of significant predictability appears much stronger compared to equity and currency spreads. Out-of-sample performance of portfolios of significant rules typically exceeds transaction cost estimates and generates a Sharpe ratio of 3.67 in 2016. In general, we reject previous studies’ evidence of a uniformly monotonic downward trend in the selection of predictive TTRs over 1990–2016.
Keywords: Technical Trading Rules; Spread Trading Predictability; False Discovery Rate; Bootstrap Test; Portfolio Performance
JEL Classification: C12, C53, G11, G14, G15
Suggested Citation: Suggested Citation