Extreme winners, not special memes? No robust evidence of a meme-specific long-run reversal after the 2021 retail squeeze

Purpose To test whether the long-run reversal of the 2021 meme stocks reflects a meme-specific effect, or simply the reversal that follows any extreme short-run appreciation. The cohort is defined by realized event-window returns, not by any observed measure of social-media activity or retail order flow, so the study characterizes long-run return behavior and makes no causal claim about retail coordination. Design/methodology/approach In a long-horizon event study, we track 44 CRSP common stocks with event-window cumulative abnormal returns above 100% during the January 2021 squeeze through December 2025 (48- and 58-month horizons), via matched-firm buy-and-hold abnormal returns (BHAR) and calendar-time factor models. The primary test is a within-2021 regression of abnormal return on event-window CAR and a meme indicator, with a same-window benchmark, long-short portfolio, squeeze/attention proxies and a cross-decade placebo as robustness. Findings The cohort reverted sharply (median 48-month abnormal return −27.4%; 81.8% underperformed their matched controls). Its multi-year decline is consistent with that of the extreme winners used as comparison groups: the within-2021 meme-threshold coefficient is insignificant (−51 points, t = −1.76), the long-short five-factor-plus-momentum alpha is insignificant (−0.31% per month, t = −0.37) and neither short interest nor turnover predicts reversal. The raw same-window gap is marginal (Mann–Whitney p = 0.07) and does not survive risk adjustment. We report a bounded failure to reject, not equivalence. Originality/value Among the first multi-year cross-sectional studies of the return-defined 2021 meme cohort, combining matched long-run returns, same-window benchmarks, measured squeeze and attention proxies, and explicit construct-validity and BHAR-tail caveats.