Analysts and researchers have documented all kinds of "calendar effects," looking at the breakdown of days, weeks, months, seasons, years, presidential cycles and on and on. It is often hard to distinguish statistically between a true effect and a random blip. And even when an effect has been proven to exist, future traders in a more efficient market can eliminate the expected gains over time.
The Big Crunch looked at S&P 500 median daily returns in a given month since 1980, and found evidence for consistent outperformance on the first and 16th of the month. The median return for all trading days is about 3 basis points, or 0.03 percent. But look at the chart above, and median returns for the first and 16th of a month are almost 0.25 percent per day. (And because those are median returns, they are not skewed by large individual outliers). The data suggest consistent outperformance on those two days, month after month.
The "turn of the month" effect was first publicized in a 1988 paper, which found that the first four days of the month accounted for all of the positive return in the Dow from 1897 to 1986. Another paper found a similar effect from 1926 to 2005. Our data suggest a similar effect from 1980 through 2017. Even when we focused on the more recent time period since 2008, we saw similar gains.
The cause of the effect isn't clear, but in the 2008 paper, the researchers found the early-month gains exist in other sets of trading data. It wasn't limited to small cap or cheap stocks, it wasn't caused by higher volatility at month-end, and it occurs in most countries. The effect also doesn't seem to be caused by higher trading volume or mutual fund flows.