Quantum BiasBuying Stocks at a 52-Week High: Is It Really Dangerous?
The fear of buying the top, tested on the index and on the stocks people hesitate to buy the most.
In this article
A new high wasn't a warning sign. In the 335 weeks since 2000 when the S&P 500 closed at a 52-week high, the next 13 weeks were positive 71% of the time, against 70% for any week, with a slightly lower average return (+1.9% vs +2.5%). The best average returns came after deep drops, but with way more risk. Research on individual stocks goes even further: stocks close to their 52-week high have tended to keep beating other stocks.
| Question | Data | Answer |
|---|---|---|
| How did the S&P 500 do after closing at a 52-week high? | SPY weekly total return, Nov 2000 – Sep 2026, 1,353 weeks | Next 13 weeks up 71.4% of the time, against 70.2% for all weeks |
Why the question matters
"I'll wait for a pullback" is one of the most common things you hear in trading. A market at a record feels expensive, and buying it feels like paying for someone else's profit, so a lot of traders lean bearish at new highs almost on reflex.
That reflex actually mixes up two different questions:
- Is the same market more likely to fall after a new high? That's a time-series question: it compares a market with itself.
- Are stocks near their highs worse buys than other stocks? That's a cross-sectional question: it compares stocks with each other.
Academic research has mostly looked at the second question. We tested the first.
What the research says
In The 52-Week High and Momentum Investing, published in the Journal of Finance in October 2004 (vol. 59, no. 5), Thomas J. George and Chuan-Yang Hwang studied U.S. stocks from 1963 to 2001. They found that how close a stock trades to its 52-week high predicts its future returns better than its past returns do. Stocks near their highs tended to beat stocks far from them, and those gains didn't reverse later.
Their explanation is behavioral. Traders anchor on the 52-week high, so when good news pushes a stock toward that level, they hesitate to bid it higher and the price adjusts slowly. Once the news is fully priced in, the stock keeps climbing.
So the level that makes traders nervous is exactly where the research found continued strength, at least when you compare stocks against each other.
Our test: the same market, before and after a new high
We went through every week from 2000 to 2026 and flagged the ones where the market closed at or above its highest close of the previous 52 weeks. Then we measured the next 4, 13 and 26 weeks, dividends included, and compared them with all weeks. For contrast, we also flagged the weeks when the market was 20% or more below its 52-week high.
S&P 500 (SPY):
| After a week that closed… | Weeks | Next 4 weeks | Next 13 weeks | Next 26 weeks | Next 13 weeks up |
|---|---|---|---|---|---|
| At a 52-week high | 335 | +0.46% | +1.93% | +5.02% | 71.4% |
| 20%+ below its 52-week high | 111 | +1.96% | +5.60% | +11.13% | 72.1% |
| Any week | 1,353 | +0.77% | +2.47% | +5.10% | 70.2% |
Exact values from table 2 in this article. Bar lengths share a zero baseline; no new calculation or simulated price path. Read the table and its stated sample and limits in this article.
New highs turned out to be ordinary, not dangerous. The share of positive quarters after a new high (71.4%) was basically the same as after any week (70.2%), and over 26 weeks the average was almost identical (+5.0% vs +5.1%). Over 4 and 13 weeks, it was a little lower.
Deep drops were followed by higher average returns: after weeks 20% or more below the high, the next 26 weeks averaged +11.1%. But every one of those weeks fell in 2001–2003, 2008–2009, 2020 or 2022, when the market was swinging hard. Buying them meant sitting through more losses before the rebound, as our bear market article shows.
The seven mega-caps
We ran the same test on Apple, Microsoft, Nvidia, Amazon, Alphabet, Meta and Tesla, pooling all their weeks together:
| After a week that closed… | Stock-weeks | Next 13 weeks | Next 26 weeks | Next 13 weeks up |
|---|---|---|---|---|
| At a 52-week high | 1,337 | +7.0% | +16.3% | 62.3% |
| 20%+ below its 52-week high | 1,952 | +11.6% | +23.8% | 65.5% |
| Any week | 8,011 | +8.3% | +17.3% | 65.0% |
The averages are huge because of survivorship bias: these are today's seven biggest companies, picked with hindsight, and any stock that became a giant went up a lot. That inflates every row, especially the "far below the high" row, because we already know these companies came back from every drop. Plenty of companies that fell 20% never did.
What the table can show is the comparison between rows, and it tells the same story as the index: a new high wasn't followed by unusually bad returns. Individual results varied. Tesla did better after new highs, while Alphabet and Meta, and to a lesser degree Microsoft, Amazon and Nvidia, did worse than their own averages.
Why traders fear new highs anyway
- Anchoring. The previous high becomes a reference point, and prices above it feel "too high". That's exactly the mechanism George and Hwang describe.
- Hindsight. Every crash started from a high, so highs feel dangerous in memory, even though most highs were followed by more highs. Our hindsight bias article explains why the exceptions stick with you.
- Mean-reversion gut feeling. After a strong run, "it has to come back" feels logical. In weekly index data, momentum near highs and bounces after crashes both existed side by side.
What this means if you trade ES, NQ or stocks
- Don't short a market just because it's at a high. On this data, that was a coin flip at best, against a market that went up most of the time.
- Don't treat deep lows as a sure thing either. The higher average after big drops came with a lot more volatility, and for stocks, with a sample flattered by hindsight.
- Use the high as a reference level, not a forecast. A prior 52-week high is a natural spot for a reaction. Whether the market breaks through it is what your bias has to judge, with evidence.
How to test it on your own market
- Export weekly closes adjusted for splits and dividends.
- Flag the weeks where the close is at or above the highest close of the prior 52 weeks.
- Measure the next 13 and 26 weeks and compare them with all weeks.
- For single stocks, include companies that failed or got delisted, or your results will be biased upward.
Limits
- Overlapping windows. New highs come in clusters, so back-to-back weeks share most of their forward returns.
- Survivor stocks. The seven mega-caps were picked with hindsight.
- A different question from the research. George and Hwang compared stocks with each other. We compared each market with itself.
- Weekly closes. Intraday new highs happen more often and may behave differently.
How we did it
- Weekly adjusted prices for SPY and the seven mega-caps from Alpha Vantage (
TIME_SERIES_WEEKLY_ADJUSTED), through October 9, 2026. - 52-week high: the highest adjusted weekly close of the previous 52 weeks; the first year of each series is only used to set the first high.
Sources
- George, T. J., & Hwang, C.-Y. (2004). The 52-week high and momentum investing. Journal of Finance, 59(5), 2145–2176. IDEAS/RePEc
- Alpha Vantage. TIME_SERIES_WEEKLY_ADJUSTED. Data retrieved October 2026.
FAQ
Is it bad to buy stocks at an all-time high? On S&P 500 weekly data since 2000, the 13 weeks after a 52-week high were positive 71% of the time, about the same as after any week.
Do stocks near their 52-week high outperform? George and Hwang (2004) found that U.S. stocks close to their 52-week highs beat stocks further from them between 1963 and 2001.
What happens after the S&P 500 falls 20% below its high? The next 26 weeks averaged +11.1% in our sample, but those periods were very volatile and included more drops.
Why do traders anchor on the 52-week high? It's a very visible reference point. Research suggests traders are reluctant to bid prices above it, which slows the market's reaction to good news.
Does this apply to Tesla and Nvidia? After new highs, Tesla's next quarter was better than its average and Nvidia's slightly worse. Both samples are flattered by survivorship.
Research and education only. Not financial advice. Past results do not predict future returns. A backtest is not real performance.
How this article was made: drafted with AI assistance from public data and the sources listed above; facts and figures checked by Gabriel Décarie and Karel Panneton.
The week just closed, summarized, and the economic calendar for the week ahead.