How markets work9 min read

Is the January Effect Real? 100 Years of Small vs Big Stocks, Tested

It was the strongest calendar pattern in the stock market. Then it was published.

By Pavel Penev, MScFounder, TradeWize · 10+ years trading the markets

The short answer

It was real, and it's mostly gone. The S&P 500 isn't special in January: it rose in 62 of 99 of them, about as often as any month. The real January effect is small stocks beating big ones. The smallest tenth of US stocks beat the biggest tenth in 78 of 100 Januaries since 1927, by +6.1 points on average. Before 1976, when it was published, that happened 43 times in 49. Since 2010 it's 8 of 17, and the gain now tends to arrive in late December instead.

Every December, someone tells you to buy before the January effect.

Most of the time they mean stocks go up in January. That isn't what the research found. In 1976 two economists, Michael Rozeff and William Kinney, showed that January returns were unusually high. In 1983 Donald Keim showed where most of it came from: small companies.

So we tested both versions. For the index, we used 99 Januaries of the S&P 500. For small against big, we used 100 Januaries of Ken French's size portfolios, which cover every US-listed stock back to 1926. Then we checked whether luck could produce a month that good, and what happened after the secret got out.

78 of 100
Januaries when the smallest tenth of US stocks beat the biggest tenth, 1927-2026.
43 of 49
Before 1976, when the effect was first published.
8 of 17
Since 2010. Worse than a coin flip.

Finding 1: the S&P 500 isn't special in January

The S&P 500 rose in 62 of 99 Januaries from 1928 to 2026. That's 63% of the time. Across every month in that span, 60% rose.

The average January gained +1.21%. That's good, but it's only the fourth best month. July averaged +1.67%. By how often it rose, January comes third. December rose most often, in 71 of 98 years.

The S&P 500 by month
MonthYears upAverageMedian
January62 of 99+1.21%+1.55%
February51 of 99−0.11%+0.20%
March60 of 99+0.42%+1.00%
April65 of 99+1.36%+0.94%
May60 of 99+0.13%+1.07%
June56 of 99+0.79%+0.30%
July60 of 99+1.67%+1.62%
August58 of 99+0.70%+0.88%
September44 of 99−1.11%−0.45%
October58 of 99+0.53%+0.93%
November61 of 98+1.01%+1.63%
December71 of 98+1.27%+1.48%

S&P 500 closes, price only, 1928 to October 2026. Each month runs from the last close of the month before to its own last close.

So if someone tells you the market goes up in January, they're right, about as often as it goes up in any other month.

Finding 2: small stocks were special

Now split the market by size. Ken French sorts every US stock by its market value each June. The smallest tenth are tiny companies, worth about $161 million each on average in 2026. The biggest tenth are giants, worth about $293 billion each.

In January, the small tenth beat the big tenth in 78 of 100 years. The average gap was +6.1 points, and the median +4.7 points. That's a huge gap for one month.

How often small stocks beat big ones, by month
HOW OFTEN SMALL STOCKS BEAT BIG ONESSmallest tenth of US stocks against the biggest tenth, by calendar month, 1926-2026.0%25%50%75%100%78%Jan54%Feb48%Mar49%Apr46%May51%Jun44%Jul44%Aug50%Sep33%Oct38%Nov44%DecJanuary: 78 of 100 years, by +6.1 points on average. Amber dashes: half the time.Ken French size portfolios, value-weighted, dividends in. A month counts if the small tenth did better.

How often the smallest tenth of US stocks beat the biggest tenth in each calendar month, 1926-2026. January is the green bar.

No other month comes close. The next best is February, at 54 of 100. In most months small stocks lost to big ones slightly more often than they won.

We also checked whether luck could do this. We dealt every month's gap from 1927 to 2025 at random into 12 fake months, 20,000 times. The best fake month never matched January. The highest it ever reached was +4.3 points, against January's +6.1 points. Over the whole century, this was no accident.

Finding 3: it faded after it was published

Split the record at 1976, when Rozeff and Kinney published. From 1927 to 1975, small beat big in 43 of 49 Januaries, by +9.1 points on average. From 1976 to 1999 it was 19 of 24, by +3.8 points. From 2000 to 2026 it was 16 of 27, by +2.5 points.

The January gap, decade by decade
THE JANUARY GAP, DECADE BY DECADEAverage January return of the smallest tenth minus the biggest tenth, in percentage points.0+5+10+151976: published+4.71920s2/3+13.41930s7/10+11.91940s9/10+5.01950s10/10+6.71960s10/10+9.51970s9/10+2.81980s8/10+3.21990s7/10+5.32000s8/10+0.32010s5/10+1.92020s3/7Under each bar: Januaries small won. The 1920s start in 1927; the 2020s run to 2026.

The smallest tenth's January return minus the biggest tenth's, averaged by decade. The dashed line marks 1976, when the effect was first published.

Since 2010 it's 8 of 17, with a median of −0.3 points. The average is still +0.9 points, but one year carries it: January 2021, the month of the meme-stock squeeze, at +17.8 points. Leave it out and the average is −0.1 points.

The luck test changes too. Run on 2000 to 2025 alone, the best of 12 fake months matches January 11% of the time. That's no longer a pattern you can tell apart from noise.

Small minus big in January, era by era
YearsSmall wonAverage gapMedian gap
1927-197543 of 49+9.1 points+7.2 points
1976-199919 of 24+3.8 points+3.1 points
2000-202616 of 27+2.5 points+2.4 points
2010-20268 of 17+0.9 points−0.3 points
All, 1927-202678 of 100+6.1 points+4.7 points

Ken French size portfolios: the smallest tenth of US stocks minus the biggest tenth, value-weighted, dividends in.

This is what usually happens to a pattern once it's in the textbooks. Traders buy earlier to get ahead of it, and the gain spreads out or disappears.

Finding 4: the fund version never had it

Most people who try to trade the January effect buy a small-cap fund, usually one that tracks the Russell 2000. Those aren't the tiny stocks in the research. The Russell 2000 holds mid-sized and smaller companies, and the very smallest stocks aren't in it at all.

The Russell 2000 beat the S&P 500 in only 16 of 39 Januaries from 1988 to 2026. The median gap was −0.12 points. In every other month it won 214 of 429 times, about half. From 1988 to 1999, it won only 3 of 12 Januaries. Since 2000 it's 13 of 27.

So the version you can actually buy in one ticket didn't have a January effect in the years we can measure it.

Learn it by doing

Reading about it is one thing — it clicks when you do it. Learn it hands-on with free, interactive lessons on TradeWize.

Try the free lesson →

Finding 5: what's left of it moved into December

The usual explanation is tax-loss selling. Investors sell losing stocks in December to claim the loss on that year's taxes. Small, beaten-down stocks get hit hardest. In January the selling stops and they bounce back.

If traders know the bounce is coming, they buy before it starts. So we used daily data and split the turn of the year into four windows.

The effect moved into December
THE EFFECT MOVED INTO DECEMBERSmallest tenth minus biggest tenth around New Year, average percentage points.1927-19752000-2026−20+2+4+6−1.40.01-15 Dec−1.0+1.516-31 Dec+4.8+1.2First 5 days of Jan+4.0+1.2Rest of JanLate December: small won 16 of 49 years before 1976, 22 of 27 since 2000.Ken French daily size portfolios, value-weighted. Each year is the December before its January.

The smallest tenth of US stocks minus the biggest tenth in four windows around New Year, averaged over 1927-1975 and 2000-2026.

Before 1976, small stocks lost to big ones in the second half of December most years. They won in only 16 of 49 years, with an average gap of −1.02 points. Then the first five trading days of January did the work. Small won 48 of 49 times, by +4.81 points on average.

Since 2000 it's turned around. Small stocks won the second half of December in 22 of 27 years, by +1.47 points on average. The first five days of January dropped to 16 of 27, and the rest of the month to 17 of 27. Since 2010, late December still went to small stocks 14 times in 17.

Gains of a point or so over two weeks are small. They're also on the smallest, least traded stocks, where the spread between buying and selling can eat most of it.

Finding 6: the January barometer

There's a second January claim: "as January goes, so goes the year." We tested it on the S&P 500, 1928-2025. For each year we compared January with the rest of the year, from January's close to December's.

After an up January, the rest of the year rose 48 times in 61, with a median of +11.8%. After a down January, it rose 22 times in 37, with a median of +2.5%. So there's a lean.

But it's a weak tool. The barometer called the rest of the year right 63 times in 98. Guessing "up" every single year was right 70 times. And even after a down January, the market still rose most years.

It also gets picked from a crowd. A gap this big turns up by chance in 3.7% of shuffles. January does make the best barometer of the 12 months. The rest of the year rose 19 points more often after an up January than after a down one. But with 12 months to choose from, one of them was always going to look good.

What people say

Buy stocks before January. They go up.

  • The S&P 500 always does well in January.
  • Small-cap funds jump in January.
  • A down January means a down year.

What the data says

Tiny stocks used to win January. Now it's late December, if anything.

  • The S&P 500 rose in 62 of 99 Januaries, about average.
  • The Russell 2000 beat the S&P in 16 of 39.
  • After a down January, the rest of the year rose 22 of 37 times.

What it was worth in dollars

Say you held the biggest tenth of stocks all year, but switched into the smallest tenth for every January. From 1927 to 1975, $100 would have become about $252,075. Holding the big tenth the whole time made about $4,485.

$100, with and without the January switch
YearsSwitch to small stocks each JanuaryHold big stocks all year
1927-1975$252,075$4,485
2000-2025$1,298$712
2010-2025$995$900
1988-2025, Russell 2000 in January vs S&P 500$2,652$2,771

The first three rows use Ken French size portfolios with dividends. The last row uses index prices only. No fees, spreads or taxes.

That first row is the most famous result in calendar investing, and almost nobody could have earned it. The tiniest stocks of the 1930s traded rarely, with wide spreads. Since 2010 the switch made $995 against $900, before any costs. With the Russell 2000 in place of the tiny stocks, it lost: $2,652 against $2,771.

How to read it

  • When someone says "the January effect," ask which one. The index version never really existed. The small-stock version did.
  • Don't buy a small-cap fund for January. The Russell 2000 beat the S&P 500 in 16 of 39 Januaries.
  • If you want to trade what's left, it's in the second half of December, in the smallest stocks, and it's a point or two before costs.
  • Don't sell because January was down. The rest of the year still rose 22 of 37 times.
  • It can still show up. In January 2026 the small tenth beat the big tenth by +3.5 points. One good year isn't a comeback, though.

What is the January effect?

It's the pattern of small stocks beating big ones in January. The smallest tenth of US stocks beat the biggest tenth in 78 of 100 Januaries from 1927 to 2026, by +6.1 points on average.

Is the January effect still real?

Mostly not. Small stocks beat big ones in 43 of 49 Januaries before 1976, and 8 of 17 since 2010. What's left tends to show up in the second half of December instead.

Do stocks go up in January?

Usually, but no more than in other months. The S&P 500 rose in 62 of 99 Januaries since 1928, with an average gain of +1.21%. December rose more often.

What causes the January effect?

The usual explanations are tax-loss selling in December, which pushes down small losing stocks, and fund managers selling embarrassing holdings before year-end reports. Both pressures lift in January. The data shows the pattern, not which cause is right.

Does the January barometer work?

Weakly. After an up January the S&P 500 rose over the rest of the year 48 of 61 times, and after a down one 22 of 37 times. But guessing "up" every year beat the barometer, 70 correct calls to 63.

Method, and what this cannot tell you

We used two sets of data. For small against big stocks, we took Ken French's size portfolios, built from the CRSP database of every US-listed stock: the smallest tenth by market value and the biggest tenth, with dividends included, monthly from 1926 to September 2026 and daily over the same span. For the index, we took S&P 500 and Russell 2000 closes from Yahoo Finance, price only, through October 2026. Each month runs from the last close of the month before to its own last close. To test luck, we dealt every month's small-minus-big gap at random into 12 fake months, 20,000 times, and counted how often the best fake month beat the real January.

  • The smallest tenth of US stocks is really tiny: micro-caps worth about $160 million on average in 2026. Many trade thinly, with wide spreads, so the paper returns overstate what a buyer could have kept.
  • The French portfolios include dividends and the index figures don't. That doesn't change the January comparison, since both sides of each gap are measured the same way.
  • A hundred Januaries is a decent sample, but the eras are short. Since 2010 there are only 17 of them, so one or two big years can swing the result.
  • We split the history around when the effect was published (1976 and 1983). That's a choice made after looking, and a different split would move the numbers.
  • None of this says why the effect happened. Tax-loss selling in December and fund managers tidying their books are the usual explanations, and the data here can't tell them apart.

Learn to test a pattern before you trade it

Our technical analysis track teaches you to read charts and check a pattern against the data, with charts you practise on.

Written by

Pavel Penev, MSc

MSc Investment & Finance, Queen Mary University of London · 10+ years trading the markets

Pavel founded TradeWize after years of trading and an MSc in Investment & Finance from Queen Mary University of London. He writes these guides to teach the decisions, not just the theory.

More about TradeWize →

Terms in this article

Keep reading

How markets work

Is the Santa Claus Rally Real? Every One Since 1950, Tested

The S&P 500 rose in 58 of 76 Santa Claus rallies since 1950, the best week of the year. But it's fading, and it missed in 2023, 2024 and 2025.

How markets work

Do Stocks Go Up the Week of Thanksgiving? Every One Since 1950, Tested

The S&P 500 rose in 53 of 76 Thanksgiving weeks since 1950. But a lucky week looks like that. The real edge is two days, and it's shrinking.

Technical analysis

Does the Golden Cross Work? 100 S&P 500 Crosses Since 1928, Tested

Every S&P 500 golden and death cross since 1928. A year later, golden crosses rose 74% of the time and any day 70%. Trading them tied holding.

How markets work

Do Stocks Go Up on Fed Days? Every Fed Decision Since 1994, Tested

From 1994 to 2011 the S&P 500 rose 60% of Fed days and averaged +0.34%. Since 2016 it's risen on only 44%. We tested all 262 decisions.