First Day Funk

January has a distinguished place in market lore. The first is the January Effect, which is the tendency for market returns during January to exceed the average for all other months. The second is the January Barometer, which is the tendency for market returns during January to predict the rest of the year’s return.

These January-related tidbits have been one of the most common pieces of market trivia that most market players take them for granted. One possible reason is that the time-frame used to take advantage of the returns (i.e. monthly/yearly trading) are too impractical for the January insights to have any benefit to the investor.

Good news: we won’t really be talking about January here.

However, the concept behind the January biases may have some merit. The often cited reason is that portfolio rebalancing usually occurs near year-ends, which overlap January. Also that the same annual rebalancing activity is the reason why the rest of the year will reflect the changes done in January.

So again we won’t be talking about January. But we might talk about the same ideas, and apply them a little differently. Let’s go more micro. Instead of predicting yearly returns, what if we look at daily and monthly returns? What if for the same reason January might bias the year, perhaps something analogous to January might cause the same bias over shorter periods.

Could, for instance, the first trading day’s results bias the returns for that week? Could the first trading day of the month be a predictor of that month’s returns?

Why get into the exercise? Well, exactly for the same common reason January might be significant: portfolio rebalancing. Except that we won’t debate that its done yearly–which we already presume. What we can check, is if the same bias can occur over weekly and monthly time frames. If so, then we might have an indicator that ties in with shorter periods–making it practical enough to be useful.

Let’s check weekly. I took all first days of the week (mostly Mondays, but sometimes Tuesdays due to holidays) and grouped them into first days which closed up for the day and closed down. Then I took the week-on-week returns of the weeks of those days, as well as counted the number of weeks which closed up or down.

Surprise surprise. Out of 574 weeks tested since 1997, those weeks with first days closing down lost a cumulative 1,808 points! Meanwhile those weeks with first days closing up have earned cumulatively 1,479 points instead! The win rates for both sets are roughly even, but the average expectancy is very much dictated by the first day performance.

Correlating the points gained or lost by the first days with the points gained or lost by the week:

The trendline is flat, with a correlation of 0.02–practically zero. However the cumulative returns posted by those weeks depending on their first day is quite interesting.

How about monthly returns in relation to first days of the month?

Now out of 132 months tested, those months where first days closed down lost cumulatively 1,645 points and have a lower winning percentage compared to months where first days closed up, which have cumulatively gained 1,479 points.

Correlation on monthly returns vs. first day of the month:

At monthly returns, the upward sloping line here is quite discernable with a correlation factor of 0.24–showing a bias for the monthly returns to be aligned with the first trading day’s performance.

Insight And Application

This kind of analysis falls partly within the borders of seasonality-type studies, since we are looking at the first trading day of the month, which is a time-based condition. But this goes beyond seasonality since we can group the results clearly–and the results show that returns during first trading days have quite an influence on weekly and especally monthly returns.

Now simplicity is arguably one plus for this condition. What can probably lend more credibility to this pattern is that very few people (at least I’ve met) have actually put any significance on this time-based indicator.

This also serves as a reminder to traders to exercise extra caution on any trading strategies which require a lot of information and rule-sets. Simple conditions such as first trading day featured here, however unobserved, have a superior effect, simply because of the fact that no one is paying them any mind, so there are far fewer people fading these trends.

Knowledge of the first day influence can help traders with price-based trading entry and exit. Those who are currently long can use a bullish first day as supporting indicator to add to their positions. The opposite case applies if the first day closed down. Those looking to cut losses or exit on the next rally can be guided by first day indications to either go ahead and cut losses now, or wait a bit for the weeks or months for a favorable change dictated by the first trading month.

Discretionary traders aided by qualitative indicators like these can have an edge over purely mechanical trading, and surely has an edge over discretionary traders who trade purely on the traditional technical and fundamental strategies. Over time this extra edge can matter to a portfolio’s performance.

Weekly biases can be used for limited-time daily campaigns while monthly biases can assist more long-term traders/investors with their decision making.