Does the January Effect really exist in the stock market?


With January soon upon us, traders and investors are starting to talk about the January Effect.
The January Effect is a seasonal pattern that some traders and investors use when predicting the stock market’s performance over the entire year.

What is the January Effect? (also called the January barometer)

The January effect states that the stock market’s performance in the first 5 days of every trading year can predict how the stock market will perform over that entire year.

  1. If the stock market goes up during the first 5 days of the trading year, it will go up from start-end of that year.
  2. If the stock market goes down during the first 5 days of the trading year, then the stock market’s result that year is random (i.e. the January Effect is nonexistent and has no statistical bias).

People make up all sorts of reasons for this phenomenon, most of which I think are silly (e.g. tax loss selling in December followed by increased buying in January).
I think the January Effect has no predictive value for the future. In a world with infinite data points, there will always be patterns that exist out of sheer chance and not because of any meaningful reason.
Think about it logically. Why should the first 5 days determine the course of the entire year? It shouldn’t! There is no solid fundamental reason for this pattern. Few investors/traders are going to buy stocks on e.g. October 5 just because they bought stocks on e.g. January 4.
There’s another flaw with the January Effect. “The first 5 days of January went up” even if the S&P closed higher by a mere 1 point! I doubt 1 point can determine the stock market’s fate for that entire year. This is akin to reading tea leaves or looking at stars to guess the future.

 January Effect data

How powerful is the January Effect? Believers in the January Effect often cite:

When the S&P went up during the first 5 trading days of the year, there’s an 86% chance that the S&P will be up for that year. 

This claim is misleading on 2 accounts.

  1. In any random year, the S&P has a 72% chance of closing higher vs the previous year’s close. (The S&P closed higher in 49 out of 68 years from 1950-2017).
  2. This claim cherry-picked the data. “86%” only includes data from 1980-2006, when the January Effect had the highest success rate.

The more data a study has, the more useful it becomes. I tested the January Effect on the S&P from 1950-2017 (the S&P’s entire lifespan). Here’s the data in Excel
From 1950-2017

  1. The S&P was net positive during 63% of “first 5 days of each year” (43 out of 68 years).
  2. The January Effect worked in 35 out of these 43 years. (I.e. In the 43 years that the S&P was net positive during the first 5 days, the S&P was up 35 times by year-end). This is a 81% success rate, which is not significantly higher than random (the S&P has a 72% chance of closing higher year-over-year).

As you can see, the January Effect is not very useful.

A revised version of the January Effect

A revised version of the January Effect is:

If the stock market goes up in January, it will go up from start-end of that year.

Is this true? Let’s look at the data from 1950-2017. Here’s the data in Excel

  1. The S&P closed higher during 41 out of 68 Januarys (60%)
  2. Out of these 41 Januarys, the S&P was net positive during 36 of these years (87%). This is higher than random (the S&P has a 72% chance of closing higher year-over-year).

This revised version of the January Effect is more potent than the original. However, it is far from a perfect signal. It’s not as if the revised January Effect worked in 90%+ of years.

January 2018

I will be watching the revised version of the January Effect at the end of January 2018. However, I will not put too much weight on this in my discretionary outlook. I don’t really trust patterns that don’t have a fundamental reason.

2 comments add yours

  1. I always look forward to reading your market analysis. Statistical analysis is the way to go.
    Concerning the revised version of the January effect, here is a possible reason for its effectiveness: sample size. Suppose the market goes up 18% in a twelve month period. It is likely that, for a month chosen at random during that twelve month period, the market went up that month.
    Off the top off my head, I can come up with variations of this indicator. Say, for example, if the market went up at least 2 percent in a two month period, then that is a bullish signal.
    It may be an interesting study to do: For each year that the S&P500 went up for the year, identity the number of up months and the number of down months for that year.

    • Yes that is a very valid point Ron. If the market went up 5% in January alone, then that is a buffer zone. SPX will be up that year unless it falls more than 5% during the other 11 months.

Leave a Comment