Calendar Month Seasonality (Chart Porn Videos)

24Jun10

In my perpetual quest to improve our Monthly Seasonality Map I’ve been relooking at the subject of calendar month seasonality (i.e. strong/weak months of the year).

A bit of chart porn to ponder…(click to play)

In this first video I’ve divided the S&P 500 since 1930 into overlapping 20-year periods (1930 – 1949, 1931 – 1950, etc.)

The point of this first video is to understand visually how the market has progressed through the year over long horizons and how that may have changed over time.

In each frame the investor starts the year at 0%. In month 1 the investor achieves the average January return in that period, in month 2 the average February return, etc.

What we see is that (contrary to what we may feel year to year) the market has (on average) been mostly unaffected by the calendar month and (tended to) march onwards and upwards in a sustained manner.

But there are clearly some dips and flat spots in the video (particularly around August to October) that have been fairly persistent.

Chart Porn #2: Average Monthly Relative Return

To help better isolate these dips and flat spots, video #2 below looks at the same 20-year slices, but this time I’m showing the average return for each month minus the average return for all months in that period.

In other words, a positive/negative value doesn’t mean the return was positive or negative; rather, that the month outperformed or underperformed the average.

Why not just show actual returns? Because I don’t want to unjustly reward a weak month for pulling off a decent return in a bullish period (or vice-versa). I want to see returns relative to the “norm”.

Video #2 confirms that most months have been inconsistently bullish or bearish, but there are some notable exceptions:

1. September has been the worst month, underperforming in 59 of 61 periods by an average of -1.3%. February was a close second, underperforming in all 61 periods by an average of -0.8%.

2. December has been the best month, outperforming in all 61 periods by an average of +1.2%.

Those are reasonably compelling results.

As a very active trader I still think that there are too many intra-month opportunities to consider such broad observations like this, but if we can take this all a step further and isolate even narrower timeframes that are usually the real source of the bullish/bearishness (very late December comes to mind) that might be enough to justify biasing our trade.

This is something I’d like to look at in more detail for the Monthly Seasonality Map and as always, more to follow.

Happy Trading,
ms

Geek note: all S&P 500 returns are dividend-adjusted so they will NOT match the S&P 500 price-only index. Dividends are 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   |  2 Comments