A Curious Case of the Mondays
I’m sure there’s a perfectly reasonable explanation for this, and I’m sure it’s been discussed by someone somewhere before, but for the life of me I’m stumped (sometimes you miss the most obvious details).
The graph below shows the result of an investor only investing in the S&P 500 index on MONDAYS (from the previous close) since 1930:
[logarithmically-scaled, growth of $10,000, frictionless]
From at least 1930 until right around the October, 1987 crash, Mondays were incredibly bad for the stock market. Since the crash? Mondays have been right around where we’d expect them.
So fellow market geeks, what gives?
Happy Trading,
ms
Geek note: I’m using dividend-adjusted data.
. . . . .
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: Time-based | 10 Comments
This is the so-called “weekend effect” which was first written about in academic journals in the 1980s. Supposedly the weekend effect shifted back to Friday when people started to play it. Now, it’s a bit more random.
There are other “day-of-the-week” effects as well, but these have supposedly not been tradeable since the late 80s.
RE to HG: thanks…got link? michael
Here’s a paper from the 80s that talks about day-of the-week effects.
http://www.jstor.org/pss/2327747
What Buy/Sell rule did you use for this Monday backtest?
RE to blink: buy at the close of whatever day preceeded Monday (usually Saturday prior to 1952 and Friday after), and sell at the close on Monday.
Michael,
there are other weekday effects as well.
Since 1990 Tuesday is THE key reversal (short-term mean reversion) day of the week (utilizing the SPY):
A higher close on a Monday is more often followed by a lower close on a Tuesday (246 higher and 276 lower closes), while all (!) other weekdays show an above-average probability (> 50%) of some follow-through of the previous session’s gains (means a higher close is more often followed by another higher close on all other weekdays).
And a lower close on a Monday is more often followed by a higher close on Tuesday’s session. With 56.67%, Tuesday’s show the highest probability (of all weekdays) for a higher close if the SPY closed lower on the previous session (but only slightly better than all other weekdays), but by far the best odds (cumulative returns and geometirc growth rate per trade).
Best,
Frank alias tTradingThe Odds
RE to Frank: always good to have you around sir. Interesting numbers. I agree with the numbers as averages over the whole sample, but I don’t know if I agree with the conclusion.
Daily MR has only really been effective since the late 1990′s, so let’s look at just the numbers from 2000. Yes, looking purely at average return, Tuesdays have trumped any other day. But looking at a graph of five hypothetical portfolios (each only trading daily MR on a single day of the week) we see that almost all of those gains came from a brief spike between mid-2007 and mid-2009. Further, from this portfolio view over the whole decade we see that (in IMHO) no day has been consistently superior enough to say it’s a key reversal day.
In short, looking purely at averages I agree with you. But drilling down into the data and I don’t think the observation is significant enough to be tradable in today’s market. Just my $0.02. michael
Hi Michael,
(For SPY) I’ve noticed that since Jul/09, 28/37 Mondays close +ve, and 5 of the 9 that failed to do so, still closed above previous session close. I used Jul/09 because it was around the time where the 50d EMA crossed above the 200d SMA.
I guess what i’m getting at is that do Mondays perform better in bull markets ?
RE to Phileo: I wouldn’t say that at all. Mondays have been very strong in recent history, but I think that’s more random chance than anything particularly useful.
Further reading:
http://marketsci.wordpress.com/2010/02/02/i%e2%80%99m-serious-this-time-enough-with-the-mondays/
michael