As Goes The Week---So Goes The Year?

As is the case every January, the financial media try to draw conclusions for the remainder of the year based upon the first few trading days. These studies are so popular that two phrases have been coined to describe them – "as goes the week, so goes the year" and, "as goes the month, so goes the year." Do these adages have a basis in fact?

As Goes the Week . . .

The table below shows the historical record of the S&P 500. We defined the first week of the year as the first 5 trading days in January.

The top panel of the table shows all weeks as a predictor of the following 51 weeks. The bottom panel shows the same indicator using only the first week of every year to predict the following 51 weeks.

Since 1928, if the S&P 500 is up the first week of the year, stocks are higher 51 weeks later 69.81% of the time. While that might sound impressive, the result is similar when using any week of the year.  As the top part of the table shows, if the S&P is up on any given week, it is also up 67.37% of the time 51 weeks later.

The results are a bit more interesting when looking at down weeks in stocks. When the S&P 500 is down the first week of January, the next 51 weeks are up just 44.44% of the time.  However, on the average down week, stocks are actually up 51 weeks later 63.66% of the time.

In essence, if stocks trade up in the first week in January, the following 51 weeks should not be considered any different than any other up-week in stocks and the ensuing 51 weeks.  However, a down week in stocks during the first trading week in January is a bit different than, say, a down week in the middle of June.  Since the  S&P was up this week this indicator is offering no insight for the rest of the year.

As Goes the Month . . .

The other popular phrase often used by the media is "as goes the month, so goes the year." We tested this adage in much the same way as above.  We compared January to all months much like we compared the first week of the year to all weeks.

As the top panel of the table shows, January is no more predictive than many other months. If the month of January is up, the market is up 11 months later 80.77% of the time. While this may sound impressive, note that several months have a similar track record. Since all up months project positive returns 11 months later about 75% of the time, this suggests that the month of January is only slightly better than the average up month at predicting the direction of the market.

The bottom panel shows the results for the 11 months after a given month is down. If the month of January is down, the market is down 11 months later 45.16% of the time. While a down January ranks number one of all months in correctly predicting the direction of the market, it is no better than a coin toss in predicting the direction of the market 11 months later.

Conclusion

In this business all too often many rational people are willing to blindly accept as fact a quote or soundbite without verifying its accuracy.  Because the first 5 trading days of the year were up in 2010, they should not be viewed any differently than any other positive week for stocks when being used to predict market direction 51 weeks later.  It is only when stocks are down in the first week of the year that the market acts differently in the ensuing 51 weeks.

OT: E-mailed as well -

BR you absolutely MUST DO A POST on the St. Louis Fed “paper”. I know what we can all do with that “paper”.

http://research.stlouisfed.org/wp/2010/2010-001.pdf

“The question I ask in this paper is whether legislative attempts to promote the public disclosure of information in financial markets necessarily constitute a worthwhile social objective.”

“In particular, reputable agencies risk losing their credibility; and the punishment for this can be severe. The nondisclosure practices of reputable agencies should therefore be examined with an open mind.”

“Their result is important because it emphasizes that informationally efficient asset prices (say, as postulated by the efficient markets hypothesis) are neither necessary or sufficient to guarantee allocative efficiency.”

it’s all just CNBC nonsense- words to say that mean nothing-

the public repeats it as if its true as they mindlessly while there hours away-

that Gibbs was asked by a report whether the State of the Union would preempt “Lost”- says all you need to know about public participation in this “supposed” democracy

The top panel of the table shows all weeks as a predictor of the following 51 weeks.

No. It shows a correlation the first week to the following 51 weeks. Some folks would do well to learn the difference.

Found this post very interesting and thought to use this line of thinking for these cliches.. thanks! Oh and Lost is great, I understand the concern

To jpm:

He didn’t use “predictor” improperly…

Nor, incidentally, are the two terms correlation and predicot necessarily mutually exclusive .

Give the author a little latitude…he is describing his own table after all. I think he has a pretty good idea as to what the purpose of that table was!

I thought it was a good post. He put the trite “truism” to the test without all the BS.

Sorry…”predictor” in place of predicot.

There.

If you want to fault someone for using “predictor” improperly, go ahead and blame me.

“While a down January ranks number one of all months in correctly predicting the direction of the market, it is no better than a coin toss in predicting the direction of the market 11 months later.” Nope. The expected value of a down 11 months later is 29%. If you did a coin flip the results for the up result would be wildly off. The two cases are not independent.

A better way to calculate this is to compare the expected values if January results were independent of the 11 month on results and do a chi square. The probability that I get for these results assuming independence is about 1%- in other words we reject the premise of independence.

An even better way would be to do a correlation using the actual percentages up and down.

What about up the first week, AND up for the month?

I care more about the average gain/loss than the up/down correlation. It looks to me like on the first week of the year test, an up week correlates to an 8.25% gain while a down week relates to a .26% gain. This is quite different from the “all weeks” numbers where an up week gives 6.92% gain vs the down week’s 6.02% gain, which while still probably statistically significant, still isn’t a large difference. But the first week as an indicator gives an 8% difference which is pretty worthwhile.

Or am I missing something in the table??

You must be logged in to post a comment.

The euro has slowly slipped to flat on the day after reaching a one month high vs the US$ earlier this morning in response to the widening out in Greece 5 yr CDS which is now at a 1 yr high. Greece 10 yr bond yields rose 22 bps today to 5.86%, just 10 bps from the highest since March '09. Earlier this morning Moody's had some cautious comments on both Greece and Portugal. Also, the move comes on the heels of visits by Euro zone officials and the IMF to Greece to inspect and assist in the country's response...