Last week we explored the “best-day-of-the-week” theory and found no significant evidence to suggest that there is an optimal day of the week to trade. This week I’m going to test the theory that there is an optimal time of month to trade, and as you shall soon learn, things begin to get a lot more interesting.
Testing the Theory
The first test I ran simply computed the average daily return for liquid JSE listed equities for each day in a month. The results from the test follow in the bar chart below.
One obvious pattern in the data is the lumping of positive returns toward the end of the month and into the new month: the 25th to the 2nd of each month on average generate positive returns. This translates into nine consecutive days of positive returns. Is this the result of a low probability streak in a random series or is there something more going on. To answer this question I tested the average five day returns associated with each day of the month. Basically, I iterated through each day in a month, bought stocks and held them for five days and then computed the average which I assigned to the day of entry. The result of this test can be seen in the chart below.
We find the same pattern in the data above. It appears that there is a strong tendency for the market to rally into the end of the month: average five day returns during the final five days of the month are more than twice as large as the average five day returns associated with the rest of the month. The pattern is persistent and looks to be fairly robust. Considering the sample size, or number of stocks and trades in the backtest, and the length of the backtest, it appears that this is a statistically significant pattern. But what is the cause of this anomaly?
Evidence of possible cause
Whenever I uncover a pattern that seems to contradict randomness I always attempt to understand the possible cause. If one can arrive at a logical and intuitive explanation for the patterns existence, then one greatly increases the robustness of such a pattern. I did a little research and came up with the following explanation:
In 1990, the scholarly Journal of Finance published a paper by Joseph P. Ogden of State University of New York. He offered an explanation of why this monthly seasonality might take place.
Professor Ogden’s research discovered that 45% of all common stock dividends, 65% of all preferred stock dividends, 70% of interest and principal payments on corporate bonds, and 90% of the interest and payments on municipal bonds, is paid to investors on the first or last business day of each month. All this is in addition to the month-end contributions into 401(k) and other retirement savings accounts. Dr. Ogden concluded that this “regularity of payments” was responsible for the seasonality phenomena.
We could further argue that people are generally paid near the end or start of a month. It then follows that money is likely to flow into the market during those periods because funds are readily available to the investing public.