Market Performance on the Last Day of Quarter
This is a test (inspired by Bespoke) of the market’s performance on the last day of the quarter versus other months. The question is whether the end of the quarter behaves differently as a result of the popular belief in EOQ window-dressing.
Yes, I realize it would have been nice to have written this before today’s poor performance (but better late than never).
[logarithmically-scaled, growth of $10,000]
As I’ve talked about before, the last day of the month has always been a significant day for the markets.
The graph above shows the results of going long the S&P 500 on the last day of every month since 1930.
From the mid-1930’s through about early-90’s (50+ years), the last day of the month was consistently very bullish, but since then, consistently flat to bearish.
The next graph makes this more clear. Here I’ve calculated the rolling 10-year average return of the S&P 500 on the last day of the month.
Note how abruptly the observation came and then went (as market inefficiencies tend to do).
The question at hand though is whether the last day of the quarter behaves differently. So below I’ve run the same test, splitting out the last day of the quarter (red) versus the last day of all other months (blue).
Here we see that both have been bullish/bearish over more or less the same periods, but returns on the last day of the quarter (red) have consistently underperformed.
Note that we’re dealing with a relatively small number of observations here, so take these results with extra caution.
In Summary
Returns on the last day of the month have been flat/bearish since the early-90′s. Based on a limited number of observations, the last day of the quarter (like today) may in fact be even more bearish than the last day of other months.
Any belief in window-dressing having a positive affect on the market’s is no longer applicable today. If anything, the opposite is true; the end of the quarter is, in today’s market, a particularly tough time for stocks.
Happy Trading,
ms
Geek notes: returns are frictionless (i.e. ignore transaction costs, slippage, etc.) and dividend-adjusted (dividends interpolated from quarterly 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 | 3 Comments
Can we speculate it is due to baby boomers’ cashing-out?
Hello haoqi001 – unfortunately, I don’t do attribution, just numbers. I’m a geek, not a pundit =) michael
Thanks for reply. I understand. The only reason for me to “find” an attribution is to have something to logically “hang” there.
Okay, whatever the attribution, it may be especially true on Q3, the end of SEP. Could you please help get the number for the last weeks of SEP vs. the last weeks of all other quarters.
My “bad” habit tends to imagine further the pensioners need a lot of money for their Winter spending…just kidding.