Do Last Year’s Best Performing Stocks Repeat? Eighteen Years of Evidence
We took the top 25 stocks of every year since 2007 and looked at what they did next. On average 2.6 of them stayed in the top 25 — and 2.6 fell into the bottom 25. Last year’s winner was as likely to become next year’s worst performer as to repeat.
The short version
- Across 18 consecutive year-pairs, an average of 2.6 of each year's top 25 stocks were still in the top 25 the following year.
- An average of 2.6 of them landed in the bottom 25 instead. A previous year's winner was about as likely to finish last as to repeat.
- In 3 of those 18 years, not one of the top 25 repeated.
- But they did not systematically lag either: the previous year's winners beat the universe average in 11 of 18 following years. Unpredictable is not the same as bad.
Every January the same lists appear: last year's best performing stocks, ranked. The implicit promise is that the list is useful going forward. We have 18 years of data to test that with — the top 25 of every year since our record begins, and what each of them did in the twelve months that followed.
What we measured
For each pair of consecutive calendar years, we ranked every company with a complete return in both years, took the top 25of the first year, and looked up where each of them finished in the second. Restricting to companies present in both years matters: a business that listed in between has no previous-year rank to have fallen from, and including it would quietly change what “the top 25” meant from one pair to the next.
Returns are December close to December close on the adjusted price, so dividends are counted as reinvested. Only complete years are used, which is why the current year never appears.
Winners rarely stay winners
| Year → next | Still in top 25 | Fell to bottom 25 | Winners' next year | All stocks |
|---|---|---|---|---|
| 2007 → 2008 | 0 of 25 | 2 of 25 | -67.2% | -56.2% |
| 2008 → 2009 | 0 of 25 | 5 of 25 | +80.0% | +144.8% |
| 2009 → 2010 | 1 of 25 | 4 of 25 | +20.3% | +31.1% |
| 2010 → 2011 | 1 of 25 | 3 of 25 | -23.0% | -24.4% |
| 2011 → 2012 | 3 of 25 | 0 of 25 | +60.5% | +48.9% |
| 2012 → 2013 | 4 of 25 | 3 of 25 | +22.6% | +5.2% |
| 2013 → 2014 | 3 of 25 | 3 of 25 | +91.4% | +82.0% |
| 2014 → 2015 | 4 of 25 | 0 of 25 | +37.7% | +17.1% |
| 2015 → 2016 | 0 of 25 | 5 of 25 | +4.3% | +11.6% |
| 2016 → 2017 | 5 of 25 | 0 of 25 | +107.6% | +69.3% |
| 2017 → 2018 | 1 of 25 | 5 of 25 | -21.9% | -11.6% |
| 2018 → 2019 | 2 of 25 | 2 of 25 | +8.1% | -0.9% |
| 2019 → 2020 | 4 of 25 | 0 of 25 | +65.2% | +33.1% |
| 2020 → 2021 | 3 of 25 | 0 of 25 | +194.3% | +72.7% |
| 2021 → 2022 | 2 of 25 | 4 of 25 | +15.0% | +16.5% |
| 2022 → 2023 | 6 of 25 | 3 of 25 | +111.9% | +65.1% |
| 2023 → 2024 | 7 of 25 | 0 of 25 | +90.2% | +34.1% |
| 2024 → 2025 | 1 of 25 | 7 of 25 | -5.6% | +4.5% |
The average across all 18 pairs is 2.6 of 25. Roughly one in ten. In 3of those years the figure was zero — not a single one of the previous year's twenty-five best performers was among the next year's twenty-five best.
They were just as likely to finish last
The striking part is the symmetry. An average of 2.6 of the top 25 stayed at the top, and an average of 2.6 of them fell into the bottom 25 of the same universe.
On this record, a stock that had just finished in the top 25 was about as likely to finish in the bottom 25 the following year as to stay where it was.
That is close to what you would expect if last year's rank carried no information about this year's at all — which is the useful finding, because it is precisely the information a “best performers of last year” list implies it is giving you.
But they did not lag on average
One honest correction to the obvious conclusion. Failing to repeat is not the same as underperforming, and the data separates the two. The previous year's top 25 beat the universe average in 11 of 18 following years — a little over 61%.
So the right conclusion is narrower than “avoid last year's winners”. What collapses is the ranking, not the returns: those stocks kept performing roughly in line with everything else, while their position in the league table became close to random. A list of last year's leaders is a poor predictor of next year's leaders and not obviously a bad set of companies. Those are different statements and only the first is supported here.
What this does not prove
- This is a description, not a mechanism. We measured that ranks do not persist. We have not shown why, and the plausible explanations — mean reversion, one-off re-ratings, cyclical sectors moving together — are not separated by this data.
- 18 pairs is not a large sample. It is the whole record we have, which is not the same thing as being statistically comfortable.
- Top 25 of a few hundred is a narrow slice. The finding is about the extreme tail, and says nothing about a company that finished, say, fortieth.
How to use this
Mostly as a reason to treat any “best performing stocks of last year” list — including the ones we publish — as a record of what happened rather than a shortlist. That is why those pages exist year by year: reading two of them back to back makes this finding obvious without any statistics at all.
If a company on such a list interests you, the useful next step is the longer record rather than the recent one. Rolling returns show every holding period a stock has been through, and the drawdown pages show what the falls looked like along the way. Our methodology covers how each figure is calculated.
HELP CENTER
Frequently asked questions
Everything you need to know about this page and the data presented.
Do last year’s best performing stocks repeat the following year?+
Rarely. Across eighteen consecutive year-pairs in our Nifty 500 data, an average of about 2.6 of each year’s top 25 stocks were still in the top 25 twelve months later — roughly one in ten. In three of those years, not one of the top 25 repeated. Past calendar-year rank carried very little information about the next year’s rank.
Are last year’s winners more likely to crash?+
They were about as likely to finish at the very bottom as at the very top: an average of 2.6 of the top 25 stayed in the top 25, and an average of 2.6 fell into the bottom 25 of the same universe. That symmetry is the clearest way to see that the previous year’s ranking carried little forward-looking information — but it is not evidence that winners are especially prone to collapse.
So should I avoid last year’s top performers?+
This research does not support that, and we do not make recommendations. Failing to repeat as a leader is not the same as performing badly: the previous year’s top 25 beat the universe average in 11 of the 18 following years. What collapses is their position in the ranking, not their returns. The narrow, supportable conclusion is that a list of last year’s winners is a poor predictor of next year’s winners.
Why measure calendar years rather than rolling periods?+
Because that is how the lists people actually read are constructed — "best stocks of 2024" is a calendar-year claim, so testing it requires calendar years. It is an arbitrary window in every other respect: nothing about a business changes on 31 December, and a run that straddles two years can be invisible in both. Our rolling-returns pages measure every possible start date instead, which is the better tool for almost any other question.
Is this affected by survivorship bias?+
Yes. Only companies still in today’s Nifty 500 are measured, so a business that had a spectacular year and then collapsed entirely cannot appear in the following year’s figures. That means the real record of past winners was somewhat worse than shown here, since the most complete failures are missing by construction.