The January Barometer holds that the direction of the market in January predicts the direction of the market for the whole year. The saying is:
“as January goes, so goes the year”
The January barometer was first mentioned by Yale Hirsch of the Stock Trader’s Almanac in 1972.
NB. The January Barometer is sometimes confused with the January Effect (more on this confusion here).
So, does the January Barometer work?
According to the John Dorfman in an article in January 2016
the barometer has been correct in 38 of the past 54 years for a success rate of 70 percent.
70% accuracy is not bad. Although Dorfman does point out that this under-performs a naive model (which forecasts the market will be up every year) which has an accuracy of 74% over the same period.
How does the January Barometer fair in the UK market?
Let’s jump straight in with a long-term view. The following chart plots the January and same-year full-year returns for the FTSE All-Share Index since 1800.
To support the January Barometer there needs to be a majority of points in the top-right and bottom-left sectors (i.e. both positive returns for January and the full year, or both negative for January and the full year). And, yes, broadly that does seem to be the case. The line of best fit has a positive slope which agrees with our visual inspection. The correlation (R2) is very low, but to be fair the January Barometer does not claim there is a close correlation in (January and full-year) returns, just that the sign (i.e. the direction) is the same.
We’ll now look at a shorter period to see how the relationship changes over time.
The following chart is as above except the time period is from 1900 to the present day.
Since 1900 the relationship has remained largely the same, i.e. the January Barometer broadly holds.
Finally, we look at the (relatively) short-term; the chart below shows the relationship since 1980.
And, again, the January Barometer still seems to have worked since 1980.
The following table summarises the success rate of the January Barometer over the three time periods. For example, since 1800, the Barometer has been correct for 61% of years.
The table also includes figures for the success rate for years in which the returns in January were respectively positive and negative. For example, since 1800 the Barometer was correct for 65% of years in which the market rose in January.
As can be seen, for all three periods the Barometer had success rates of over 50% for all years – and so can be considered to have worked. Since 1980 the Barometer has been notably successful.
But there can be seen a difference in accuracy between years in which January returns were positive or negative. The Barometer works better in years with positive returns.
The January Barometer since 1980
The following chart might help to visualise the behaviour of the Barometer in recent years. The chart plots the performance of the Barometer for each year from 1980; a 1 is plotted for years in which the Barometer worked, and a -1 in those years when it didn’t.
Of late the performance has been patchy, but there was a remarkable run of 12 years from 1982 in which the Barometer worked every year. It should be noted perhaps that the market was strong in this period: annual returns were positive for every year except one year. And, as we saw above, the Barometer does seem to work better in years when the market rises.
So, is the January Barometer economically significant – i.e. can you trade it profitably?
The following chart plots the FTSE All-Share returns for January and the full-year for the period 1980-2015.
A concern might be that a significant portion of the full-year returns might be accounted for by the month of January itself. In other words, by the time one acts on a signal from the Barometer, the market has already moved significantly. However, the chart shows that although a disproportionate size of the full year return have been accounted for in January, the remaining returns are significant.
Over this 1980-2015 period, if a strategy had invested in the market from February to year-end just in the years that the Barometer gave a signal for a positive market, the average annual return in those years would have been 9.5%. This is marginally better than the average market return for all years over the same period of 9.0%.
The January Barometer has had a better than 50% success rate for the three periods studied here: 1800, 1900, and 1980 to the present day.
Although one should be wary about a assigning a special relationship between January and the full-year. For example, since 1800 the month of February has been almost as good as January in predicting full year returns, and since 1900 February has had a higher succes rate in positive return years.