Jul

23

Are the S&P returns different for different days of the week? According to a vast published literature on the Day of The Week Effect, the answer is yes. (And the worst return is generally on Monday, according to French (1980)).

To make money from such a regularity, one could determine the best/worst weekdays to invest during a certain period, then try to take advantage of that during the next period. The results then determine what days to bet on during the following period, and so forth.

In my statistical study here, every non-overlapping 100 trading day interval in SPY was checked for return, by day of week, back two or three bull markets ago 5/03. Here are the T-scores (test for significance vs zero) for each of the weekdays over time:

Values outside the range (-1.96,1.96) indicate that a statistically significant anomaly (at 5% level) has been detected in the given period for the given weekday.

Alex Castaldo adds:

Mr. Bonferroni would be pleased that out of 95 reported coefficients there are 5 that are statistically significant: 2.03 (Mon), 2.85 (Wed), 2.76 (Tue), 3.25 (Wed again), and 3.36 (Mon again).


Comments

Name

Email

Website

Speak your mind

1 Comment so far

  1. Pete Fry on July 24, 2010 8:46 am

    Can you tell me a little more about what this is?
    Thanks

Archives

Resources & Links

Search