Opulence Alpha Research · Working paper

Do stocks that just jumped keep winning?

Opulence Alpha Research · Published Sep 25, 2026 · Data through Aug 21, 2026

Abstract

Not on average: they tend to lag the next week. Since 1995, the fifth of US stocks that beat their industry most over three sessions beat the median stock the following week 48 times in 100; the fifth that lagged most, 51 times. A coin flip gives 50. The relation pointed the same way in 32 of 32 years: steady across hundreds of stocks, a nudge for any one.

Keywords: short-term reversal, do stocks that go up keep going up, mean reversion stocks, weekly reversal effect, short-term reversal anomaly, residual reversal·JEL classification: G11, G12, G14, C12, C58·Concepts: Multiple testing, Point-in-time data, Backtest overfitting

Study design
Universe
The same for every study: 1,767 US common stocks in the 11 GICS sectors, 422 of them since delisted; S&P 500 members since 1996 plus large and mid-sized companies outside the index
Period
Every Wednesday from Jan 4, 1995 to Aug 19, 2026: 1,634 Wednesdays across 7,959 trading sessions
Sample
n = 1,763 stocks of the 1,767-stock universe; 1,889,758 stock-weeks, a median of 1,148 stocks per Wednesday
Outcome
Return over the next 1, 5, 21 and 63 trading sessions against the same day's median stock
Inference
t on non-overlapping dates; proven only when |t| ≥ 3 and the same sign in at least 60% of years

1Introduction

A stock that has just jumped is often expected to keep going: strength is read as news the market has only begun to price. Over short horizons the academic record points the other way. Jegadeesh and Lehmann documented that monthly and weekly winners tend to give back part of their gains, and Blitz, Huij, Lansdorp and Verbeek found the reversal stronger and steadier once a stock's return is measured net of what common factors explain. This study measures the relation again on one universe, every Wednesday since 1995, using a stock's three-session return against the median of its own industry, and follows it from the next session to three months.

2Data and method

2.1Sample design

Population and frame. The population is US common stocks listed on the NYSE and Nasdaq; funds, ETFs, trusts, preferred shares, warrants and units are excluded. The sampling frame is a fixed universe of 1,767 companies, drawn once when the platform was built and not re-sampled since, in two strata. Stratum 1 is a census of the S&P 500: every company in the index at any time since 1996 whose price history could be recovered, 1,027 companies of which 413 have since delisted; it holds 76% of the index's members in 1996 and at least 96% in every year from 2010. Stratum 2 is 740 large and mid-sized companies outside the index, selected in proportion to the market's sector weights from the stocks that passed a minimum share price of $15 and a minimum average daily trading value of $25 million; 9 of them have since delisted.

Sample. The unit of observation is a stock-week: one stock on one Wednesday. A stock enters a Wednesday's cross-section when it has a valid close that day, a value of the signal and a measured outcome; bars flagged as bad data and returns that cross a change of issuer are left out. This study's sample is n = 1,763 stocks of the 1,767: 1,889,758 stock-weeks on 1,634 Wednesdays from Jan 4, 1995 to Aug 19, 2026, a median of 1,148 stocks per Wednesday (range 820 to 1,371). 331,954 stock-weeks (17.6%) come from the 422 companies that have since delisted. Every stock and every Wednesday carries equal weight.

Representativeness. Table 1 gives the sample by GICS sector beside the S&P Composite 1500: 7.9% of companies would have to change sector for the two to match exactly. By latest market value, 55% of the active companies are large (at least $10bn), 38% mid ($2–10bn) and 6% small. Because Stratum 2 was chosen from companies listed at construction, its history carries survivorship bias. Section 4 repeats the lead result among the stocks that were S&P 500 members on each date, the part of the sample largely free of that bias.

Table 1. The sample by sector

Companies in this study's sample, of them those since delisted, the size of the active companies, the sample's share of stock-weeks and the sector's share of the S&P Composite 1500.

SectorSampleDelistedLargeMidSmallStock-weeksS&P 1500
Information Technology28267122761813.9%12.7%
Financials2566011474814.6%17.2%
Industrials25553126671115.9%17.5%
Health Care2374792811612.4%10.9%
Consumer Discretionary2093969871212.3%12.9%
Energy10530393155.9%4.7%
Consumer Staples10138401755.9%4.9%
Materials9434352425.6%5.1%
Real Estate8410403225.7%6.9%
Communication Services8031281353.9%3.3%
Utilities6013351204%4%
All sectors1,76342274051484100%100%

Sample: n = 1,763 of the 1,767 companies. Size by latest market value for the 1,338 active companies with one: large ≥ $10bn, mid $2–10bn, small < $2bn. S&P 1500 shares count the constituents of the S&P 500, MidCap 400 and SmallCap 600 (1,506 companies, lists read Sep 25, 2026). Sectors are each company's current GICS sector.

2.2Signals

Table 2. Signals studied
SignalDefinitionSinceWednesdays
Three-session return vs industryA stock's return over the last three sessions minus the median return of its GICS industry groupJan 4, 19951,634
Three-session return vs sectorThe same three-session return, measured against the median of its GICS sectorJan 4, 19951,634
A month beyond its factorsThe last 21 sessions' return left over once the usual market factors are taken outJan 4, 19951,634
Fitted trend slopeThe slope of a curve fitted to recent closing pricesJan 4, 19951,634
A month vs its sectorThe last 21 sessions' return against its sector'sJan 4, 19951,634

2.3Measurement

Outcome. The return from the close on the Wednesday to the close 1, 5, 21 and 63 trading sessions later, on closes adjusted for splits and dividends, compared with the same day's median stock; half of all stocks beat the median by construction, so chance is 50 in 100 on every date. No delisting return is added.

Each Wednesday the stocks are ranked on the signal. The study reports how often each fifth of that ranking beat the median stock, and the rank correlation (IC) between the signal and the return that followed.

The t-statistic uses non-overlapping dates only. A result is called proven when |t| is at least 3, it held in at least 60% of years and it is large enough to matter; with 1,470 tests across the studies, a looser bar would pass dozens by luck.

Full data and methods

3Results

Three sessions ahead of the industry, then the next week

Every Wednesday since 1995, we rank US stocks by how far they beat or trailed their own industry over the last three sessions, split them into fifths and follow each fifth for a week. The fifth that had beaten its industry most beat the median stock 48 times in 100; the fifth that had lagged most, 51 times. On average the first trailed the average stock by 0.16 percentage points over that week and the second led it by 0.31, before trading costs.

Share of each fifth that beat the median stock over the next week

Fifths by three-session return against the industry: lowest = fell furthest behind it, highest = beat it most. The dashed line is chance.

505644chance 5051lowest50low50middle49high48highest
Out of every 100 stocks, the next week

Each square is one stock in the highest or the lowest fifth; a filled square beat the median stock. A coin flip fills half.

48 of 100 of the biggest three-session winners beat the median stock
51 of 100 of the biggest three-session laggards beat the median stock

Strongest over a week, faint by three months

Follow the same fifths for longer. The tilt is there from the next session, peaks over the next week and is still measurable over the next month: the biggest three-session winners beat the median stock 49 times in 100 over that month, the biggest laggards 51 times, and the relation pointed the same way in 29 of 32 years. Over three months it is weaker still, the same way in 25 of 32 years, and falls short of the bar in the method below: a faint tilt, not reliable on its own.

For the curious: over the next week the rank correlation (IC) was -0.027, with a t-statistic of -12.4 on non-overlapping dates, measured across 1,633 Wednesdays.

How strong the relation is, by horizon

Rank correlation between the three-session return against the industry and the return that followed. Below the line means reversal.

-0.022next session-0.027next week-0.019next month-0.013next three months
next session: Proven · next week: Proven · next month: Proven · next three months: Faint
The next session

Proven: a measured tendency across hundreds of stocks

505644chance 5051lowest50low50middle49high48highest
The next month

Proven: a measured tendency across hundreds of stocks

505644chance 5051lowest50low50middle50high49highest
The next three months

Faint: not reliable on its own

505644chance 5051lowest50low50middle50high49highest

Four more ways to measure it, one answer

Against the sector instead of the industry, the three-session relation tells the same story: over the next week it pointed the same way in 31 of 32 years. Three slower measures agree: a month's return that the usual market factors do not explain, the slope of a trend line fitted to recent prices, and a month's return against the sector. In each, the fifth with the strongest recent run beat the median stock less often over the next week than the fifth with the weakest.

Strongest recent run (highest fifth)

Share that beat the median stock over the next week. Filled square = proven relation.

4648chance 5052543 sessions vs industry48a month beyond its factors49fitted trend slope49a month vs its sector49
Weakest recent run (lowest fifth)

Share that beat the median stock over the next week. Filled square = proven relation.

4648chance 5052543 sessions vs industry51a month beyond its factors50fitted trend slope51a month vs its sector51

A tilt, not a forecast

Across hundreds of stocks the tilt is steady; for any one stock it is a nudge. Even in the fifth that had beaten its industry most, 48 in 100 still beat the median stock the next week. The averages on this page are before trading costs. Short-term reversal is a long-studied relation, and the papers under References are the standard reading; here it is measured again on our own recomputed data, with outcomes through Aug 21, 2026.

It shows up in chart patterns too

Over the next week, two chart patterns went against their own call: 48 in 100 rounding bottoms beat the median stock, and only 48 in 100 rounding tops trailed it. Both are consistent with the short-term reversal measured here.

Chart patterns against a coin flip

4Robustness

Table 3 re-measures the lead relation. Over the next week it was -0.027 (t -12.4) across all stocks and -0.021 (t -7.7) among S&P 500 members on the date, with the same sign in 11 of 11 sectors and 3 of 3 decades, though smaller in each decade than in the one before. Over the next month it was -0.019 (t -5.5) overall, with the same sign in 11 of 11 sectors and 3 of 3 decades, but it fell short of the bar among S&P 500 members (-0.015, t -2.6) and is too small to measure in the latest decade.

Table 3. Robustness of the lead relations

Rank IC and its t-statistic on non-overlapping dates.

SampleICtWednesdays
Three-session return vs industry, next week
All stocks, whole period-0.0273-12.41,633
By decade: 1995–2004-0.0445-14.3516
By decade: 2005–2014-0.0221-5.9518
By decade: 2015–-0.0170-4.1599
S&P 500 members on the date-0.0205-7.71,581
Sectors with the overall sign11 / 11
Three-session return vs industry, next month
All stocks, whole period-0.0190-5.51,630
By decade: 1995–2004-0.0326-6.2516
By decade: 2005–2014-0.0156-2.9518
By decade: 2015–-0.0103-0.3596
S&P 500 members on the date-0.0149-2.61,578
Sectors with the overall sign11 / 11

5Limitations

  • Part of a next-session reversal can be the bid-ask bounce: a close at the ask tends to be followed by one nearer the bid. The weekly and monthly readings are less exposed to it, and none of them deducts the cost of trading on the relation.
  • The relation has shrunk from decade to decade, so its full-period average overstates how large it has been in recent years.
  • Before costs. Averages exclude trading costs, taxes and market impact.
  • Same-close timing. Returns start at the close the signal is computed from; a real trade would start later.
  • Survivors among smaller companies. The non-index names were chosen from companies listed when the universe was built; no delisting returns are added.

6Conclusion

Stocks that have just beaten their industry did not keep winning on average. Over the next week they tended to lag the median stock, and recent laggards tended to catch up: the steadiest relation in these studies, pointing the same way in 32 of 32 years. It is also small and shrinking, a tilt across hundreds of stocks and a nudge for any one of them, measured before the costs that trading on it week after week would bring.

References

  1. Jegadeesh, N. (1990). Evidence of Predictable Behavior of Security Returns. Journal of Finance 45(3).
  2. Lehmann, B. N. (1990). Fads, Martingales, and Market Efficiency. Quarterly Journal of Economics 105(1).
  3. Blitz, D., Huij, J., Lansdorp, S. and Verbeek, M. (2013). Short-Term Residual Reversal. Journal of Financial Markets 16(3).
  4. Benjamini, Y. and Hochberg, Y. (1995). Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society, Series B 57(1).
  5. Grinold, R. C. and Kahn, R. N. (2000). Active Portfolio Management, 2nd ed. McGraw-Hill.
  6. Harvey, C. R., Liu, Y. and Zhu, H. (2016). … and the Cross-Section of Expected Returns. Review of Financial Studies 29(1).
  7. Shumway, T. (1997). The Delisting Bias in CRSP Data. Journal of Finance 52(1).

Appendix A. Questions readers ask

Do stocks that went up last week keep going up?
Measured against the median stock, not on average: over the next week they tend to lag. Since 1995, the fifth of US stocks that beat their industry most over three sessions beat the median stock over the following week 48 times in 100, and the fifth that lagged most 51 times in 100, where chance is 50. It is a small tilt across hundreds of stocks and a nudge for any one.
What is short-term reversal?
The tendency of stocks that have just beaten their peers to lag them for a while, and of recent laggards to catch up. On US stocks since 1995 it is the most reliable relation we measured: over the next week it pointed the same way in 32 of 32 years.
How long does short-term reversal last?
It is strongest over the next week and still measurable over the next month, when the biggest three-session winners beat the median stock 49 times in 100 against 51 for the biggest laggards, the same way in 29 of 32 years. Over three months it is weaker, the same way in 25 of 32 years: a faint tilt, not reliable on its own.
Does a month-long run-up reverse too?
A little, over the next week. The fifth of stocks that beat their sector most over 21 sessions beat the median stock the following week 49 times in 100; the fifth that lagged most, 51 times. Over the next month that tilt is faint, not reliable on its own.
How big is the short-term reversal effect?
Small. Over the next week the fifth of stocks that beat their industry most over three sessions trailed the average stock by 0.16 percentage points, and the fifth that lagged most led it by 0.31, before trading costs. Steady across hundreds of stocks, a nudge for any one.
Do chart patterns show short-term reversal?
Two fit it. Over the next week, 48 in 100 rounding bottoms beat the median stock and only 48 in 100 rounding tops trailed it: both went against their own call, as short-term reversal would suggest.

Data availability and citation

Data availability

Every figure in this paper is quoted from one frozen snapshot, published as JSON with the sample description, the robustness results and the test counts. The same snapshot feeds the research console, so the two cannot disagree.

How to cite

Opulence Alpha Research (2026). Do stocks that just jumped keep winning? Opulence Alpha Studies, Sep 25, 2026. https://opulencealpha.ai/studies/short-term-reversal

Research, not advice. A measured tendency across hundreds of stocks is a nudge for any one of them, never a forecast of its price.