Do smaller companies' shares gain a fifth of a point more against the benchmark fund before market holidays?
Study 73 · Inconclusive · recorded 28 October 2026 · answer: Not settled · full paper with figures
The question
We averaged the session return of the smaller listed companies the project follows, priced at five or more, less the benchmark fund's, and compared each session before a market holiday with the other sessions of its weekday. Published studies report such a premium for United States shares; we ask whether it reaches a fifth of a point a session here.
Rules fixed before the result
Fixed before any outcome: the answer reads the 95% range of the eve difference against zero and 0.20 a session. A range wholly below zero meant a reversal; wholly above zero, a premium, at the full size only if the point reached 0.20; within a fifth of a point either side of zero, no premium at that size; any other range, unsettled. The earlier years had to show the test could detect 0.20 with false alarms within 5% a side.
Data and method
Each session, the shares priced at five or more at the previous close, 400 or more of them, were averaged with equal weights, less the benchmark fund's return. Before any 2018 to 2022 return was formed, 2013 to 2017 set the expected precision: the smallest detectable difference, 0.16, was within 0.20, and on 2,000 sets of ordinary sessions drawn to match the eves' weekdays and month turns, false alarms stayed under 5% a side.
The answer
Not settled. Over 46 holiday eves in 2018 to 2022, the shares' return less the benchmark fund's exceeded that of the same weekday's other sessions by 0.09 percentage points a session, with a 95% range of -0.08 to 0.25: it neither excludes zero nor rules out a fifth of a point. The benchmark fund's own eve difference was 0.15 (-0.16 to 0.45) and the shares' own, the fund not netted, 0.23 (-0.10 to 0.57), neither settled. Eves not on a Friday, against their own weekday, came to 0.27 (0.02 to 0.52) and Friday eves to -0.07; the overnight (0.02) and daytime (0.07) parts and the session after the holiday (-0.05) each had a range spanning zero.
Main estimate
0.088 percentage points (95% range -0.076 to 0.252), from 46 observations
What it changes
By the rule set in advance, the question goes back to the list of open questions with the spread and count measured here, and no holiday control is proposed for the return windows later studies measure against the benchmark fund. No family of questions is closed by this answer. The plan had expected a premium at the full size, with low confidence, and gave this unsettled answer a chance of 0.19.
What remains uncertain
Earlier studies used these years, so this is a check within our development data, not fresh data, and 46 holiday eves leave most side readings unsettled. The comparison is with the benchmark fund on 2018 to 2022 alone, so it says nothing about whether a published decline or reversal held. The check against false alarms matched the holiday eves' weekdays and month turns, not the rest of the calendar, so a seasonal pattern could still pass for a holiday premium.
How it was made
Run took 10 minutes; independent code review: approve; the recorded calculation reruns to the same result.
Fingerprints
plan c3e1eeb16ef2542a389f2712aa7db4f9dab7f1307834d01dcf2f74dfa64a8792 result b0e446dc5829cfe32255ad33a4b88f173b848046c57b9e6f63bd0d3b07b09b58
Exploratory work on development data; not a performance claim.