Blog - Investing Notes

August 24, 2006 - Taking the Summer Doldrums Out of the Turn-of-the-Month Effect

Reader Dennis Page of Beverly Hills MI suggests that we might enhance turn-of-the-month trading returns by combining the Turn-of-the-Month Effect with the seasonal stock market behavior suggested by our Trading Calendar. Specifically, Dennis suggests excluding the weak market months between trading days 135 and 198 of the year from a monthly trading routine. To test his hypothesis, we use the data from our blog entry of 8/9/06 and train two Turn-of-the-Month (TOTM) monkeys, as follows:

Aug-Oct TOTM Monkey buys the S&P 500 index at the close six trading days before the ends of July, August and September and sells at the close on the fifth trading days of August, September and October. The rest of the year, this monkey stays in cash.

Nov-Jul TOTM Monkey takes a mirror-image strategy, trading the turn-of-the-month the rest of the year and staying in cash when Aug-Oct TOTM Monkey is active.

Starting both monkeys with $100,000 on 12/21/89 and charging them $20 for each trading round-trip through 8/7/06 (paying no interest on cash balances), we find that:

The following chart shows the account balances for Aug-Oct TOTM Monkey, Nov-Jul TOTM Monkey and All Year (trading all 12 turns of the month annually) TOTM Monkey over the test period. As Dennis suspected, Aug-Oct TOTM Monkey does not fare well, and Nov-Jul TOTM Monkey outperforms All Year TOTM Monkey (at lower volatility).

Summary monkey trading statistics are:

All Year TOTM Monkey executes 200 trades for an average +0.7% gain per trade, with standard deviation 2.8%;

Aug-Oct TOTM Monkey executes 49 trades for an average -0.1% loss per trade, with standard deviation 3.5%; and,

Nov-Jul TOTM Monkey executes 151 trades for an average +1.0% gain per trade, with standard deviation 2.5%.

We also trained a Sell-in-May TOTM Monkey that stays out of the market at the ends of May, June, July, August and September. This monkey matches the average gain and standard deviation of Nov-Jul TOTM Monkey, but underperforms both Nov-Jul TOTM Monkey and All Year TOTM Monkey economically due to fewer trades. In other words, the end-of-May and end-of-June trades tend to be good ones over the test period.

Note that as we slice and dice the TOTM strategy, sample size and confidence in results decline. Viewed as a monthly effect, the above sample sizes are moderate. Viewed as an annual effect, the above sample sizes are very small. The advantage of Nov-Jul TOTM Monkey over All Year TOTM Monkey may not be reliable.

In summary, excluding the summer doldrums from a Turn-of-the-Month Effect strategy may enhance returns.

For related research, see Blog Synthesis: Calendar Effects and the Trading Calendar. See especially our blog entries of 7/20/06 on the Turn-of-the-Month Effect study and 8/9/06 testing the conclusion of this study.