Stock buyback announcements are often treated as an instant bullish signal. A company is putting its own money behind its shares, the share count may fall, and earnings per share may rise. But an authorization is permission to buy—not proof that the company will spend the money, buy at a sensible price, or outperform the market.
We tested the modern version of that belief: what happens to a stock after a company announces a buyback?
The short-term pop is real—but uneven
Across 431 companies with usable five- and 20-session price windows, the equal-weighted average return exceeded SPY at both horizons. The mean excess return was 1.51% after five sessions and 1.85% after 20 sessions.
Those averages are not the experience of most individual stocks. The median excess return was 0.76% after five sessions and 1.02% after 20 sessions. Just 54.5% of companies beat SPY over five sessions, falling to 53.1% over 20 sessions.
That distinction matters. The mean describes a diversified basket that catches every outlier. The median is closer to the result from choosing one announcement at random.
| Horizon | Companies | Median stock return | Median SPY return | Median excess return | Mean excess return | Beat SPY |
|---|---|---|---|---|---|---|
| 5 sessions | 431 | 1.24% | 0.40% | 0.76% | 1.51% | 54.5% |
| 20 sessions | 431 | 1.46% | 0.74% | 1.02% | 1.85% | 53.1% |
| 63 sessions | 430 | 5.26% | 7.10% | -0.92% | 2.00% | 47.9% |
The edge disappears by three months
The 63-session result is the most revealing. The average stock still showed a 2.00% excess return, but the median stock lagged SPY by 0.92%. Only 206 of 430 companies outperformed the index.
Why can the average remain positive while most stocks do not win? The return distribution has a long right tail. One company in the sample gained more than 300% relative to SPY over the 63-session window. That single result added roughly 0.8 percentage points to the portfolio average.
Trimming the largest and smallest 5% of outcomes reduces the 63-session mean excess return from 2.00% to 0.66%. The median remains negative.
The uncertainty says “signal,” not “strategy”
We bootstrapped the excess-return distribution 10,000 times.
The 95% confidence interval around the median excess return crossed zero at every horizon:
| Horizon | Median excess return | Bootstrap 95% interval | Sign-test p-value |
|---|---|---|---|
| 5 sessions | 0.76% | -0.01% to 1.68% | 0.067 |
| 20 sessions | 1.02% | -0.45% to 2.57% | 0.210 |
| 63 sessions | -0.92% | -2.70% to 1.06% | 0.412 |
The mean result was more favorable. Its 95% bootstrap interval was 0.29% to 2.78% after five sessions and 0.34% to 3.35% after 20 sessions. By 63 sessions, that interval widened to -0.65% to 4.82%.
The honest reading is not that nothing happens. The short-term average reaction is positive. It is that the reaction is too uneven to make “buy every buyback announcement” a dependable rule for selecting one stock.
How we built the sample
We started with 3,375 verifier-approved buyback extraction records in the Equibles filing database as of September 3, 2026. Those records come from company-filed SEC documents and preserve the source statement behind each amount and date.
To keep the study from learning about an old announcement through a much later filing, an event qualified only when:
- The filing explicitly supplied an announcement date.
- The announcement occurred from July 1, 2025 through June 1, 2026.
- The supporting filing arrived on or after that date and no more than 14 calendar days later.
- The ticker resolved to its exact listed security.
- Only the first qualifying event for each company was retained.
That produced 442 distinct companies. Of those, 431 had a usable split-safe price window for the five- and 20-session tests. One additional company did not yet have all 63 subsequent traded closes, leaving 430 at the longest horizon.
For each event, we measured the raw closing-price return from the last traded close before the announcement date to the fifth, 20th, and 63rd traded close on or after the announcement. We calculated SPY over matching dates and subtracted the index return from the stock return.
Every company received the same weight. Only positive-volume traded closes were used. Price histories were clipped at the latest captured stock-split boundary, so a return never combined incompatible share-price bases. Dividends were excluded from both the stock and SPY returns.
Why this result differs from the buyback folklore
Classic research often found positive long-run abnormal returns after open-market repurchase announcements. A long-run study of U.S. repurchases summarizes evidence of multi-year outperformance in older samples.
The effect has not been constant. Later research on announcements from 1994 through 2014 found that post-announcement performance became much weaker after the early sample years.
There is also an important gap between announcing and doing. Research on operating performance after repurchase announcements found that improvements were concentrated among companies that actually repurchased shares in the same fiscal quarter.
Our study is narrower and more current. It covers short and medium horizons around recent U.S.-listed announcements. It does not test the multi-year drift reported in older academic samples.
An authorization is not a purchase
A board authorization sets a ceiling. It does not require management to buy any shares. A company can pause the program, let it expire, or spend only a fraction of the amount.
That is why the announcement alone is incomplete evidence. Investors should ask:
- How large is the authorization relative to market capitalization?
- Is management actually repurchasing shares in subsequent quarters?
- Is the diluted share count falling, or are repurchases merely offsetting stock compensation?
- Is the company funding purchases with durable free cash flow or additional debt?
- What valuation did the company pay for its own shares?
Equibles’ largest stock buybacks ranking focuses on cash actually spent in the latest reported fiscal year. That is a different—and often more useful—question than the headline authorization.
Limitations
This is an observational event study, not proof that a buyback caused a return.
- Announcement dates are day-level; filing acceptance and press-release timestamps were not used.
- The fifth traded close can include slightly different amounts of post-announcement trading for before-open and after-close releases.
- SPY controls for the broad market, not industry, size, value, momentum, or earnings-announcement exposure.
- The filing pipeline covers announcements supported by recent company filings; it is not a complete history of every U.S. buyback.
- Using the first qualifying event per company prevents repeat issuers from dominating the sample but discards later information.
- Delisted or unresolved securities without usable portal price histories are absent from the return sample.
- Raw price returns exclude dividends, taxes, spreads, and trading costs.
- A 63-session window cannot answer whether the older multi-year buyback anomaly still exists.
Bottom line
Stock buyback announcements produced a small positive average reaction over the next five to 20 trading sessions. The result was broad enough to be visible, but not broad enough to be dependable: barely more than half of the stocks beat SPY.
By 63 sessions, the typical company lagged the index and fewer than half outperformed. The arithmetic average stayed positive because a small group of extreme winners pulled it upward.
A buyback announcement is useful information. It is not, by itself, a durable investment thesis. The stronger question is whether the company follows through—and whether it buys shares at a price that creates value for the owners who remain.