Anomalies and Multiple Hypothesis Testing
UNIVERSITY OF TECHNOLOGY SYDNEY
Finance Discipline Group
Research Seminars in Finance
Topic: Anomalies and Multiple Hypothesis Testing: Evidence from Two Million Trading Strategies
Speaker: Amit Goyal, University of Lausanne
Abstract:
We use a large laboratory of over two million trading strategies to provide thresholds for statistical significance based on multiple hypothesis testing as well as estimates of the cost (in terms of falsely rejecting a null) of evaluating hypothesis based on classical hypothesis testing. We refer to the latter as the proportion of lucky rejections (i.e., strategies that survive classical hypothesis testing but fail multiple hypothesis testing thresholds). We estimate the proportion of lucky discoveries to be around 60% and find that our estimate of optimal multiple hypothesis thresholds is higher than previously reported.
Date: Wednesday, 5th September 2018
Time: 12.00 – 1.00 p.m.
Venue: University of Technology, Sydney
Building 8, Room 8.002, Dr Chau Chak Wing Building
Dr Chau Chak Wing Building
Dr Chau Chak Wing Building
14 - 28 Ultimo Road, Ultimo
Enquiries: Mala Kapahi (Ph: +61 2 9514 7777)