Matched on trading volume, do one-day falls of a tenth or more without a company filing still recover more?
Study 67 · Changed a rule or plan · recorded 24 October 2026 · answer: Yes · full paper with figures
The question
An earlier study found falls with a company filing within a day kept falling from the next opening price while falls without one held; another found the falls with a filing had traded more by value over the prior fifty sessions than those without. This test asks whether that gap survives within the same broad group of fall-day trading volume.
Data and method
Each fall's return from the next opening price to the third close, less the market's, was compared between falls without and with a company filing within a day, on the same day, price band and volume group (under two, two to five, or five or more times the share's usual daily volume). The filing's content was not read. The plain average's power check failed, so the clipped average decided; its false-alarm rate was 5.85% (4.90% to 6.97%), inside its band.
The answer
Within the same day, price band and volume group, falls without a filing recovered 0.80 points more over three sessions than falls with one by the clipped average, which decided (range 0.25 to 1.35), and 0.90 by the plain average (0.19 to 1.62): above zero, with a one-point gap still possible. The answer speaks of the clipped average, which trims each price band's most extreme returns; the plain average gives the same answer. On the same falls without the volume groups the gap was 0.75 (0.23 to 1.29), and the groups carried none of it measurably (-0.05, -0.15 to 0.04). Among falls with a filing, heavy-volume falls differed from quiet ones by 0.05 points (-0.85 to 0.95), but too few quiet ones were counted to settle it. Among falls without one, heavy-volume falls recovered 0.76 points less (-1.48 to -0.04).
Main estimate
0.008 fraction (95% range 0.002 to 0.013), from 16617 observations
What it changes
By the rule set in advance, the filing separates the falls within the same broad volume group, so fall-day volume carries at most part of the gap; matching reduced that explanation without ruling it out, since a finer volume effect inside the groups is not excluded. The gradient among falls with a filing could not be read, so the programme next groups the filings by their form; nothing else changes. The plan expected this gap, with low confidence.
What remains uncertain
These falls were studied before, so this is a check within our development data, not a final test. Trading volume stands in for attention and is partly news itself; the three broad groups come from an earlier look at the first years, and the filing's time of day is not held. The false-alarm checks used ordinary days, not days after a fall, and with made-up unequal tails the clipped average's false-alarm rate exceeded its target.
How it was made
Run took 114 minutes; independent code review: approve; the recorded calculation reruns to the same result.
Fingerprints
plan 7ce1af6a4aee432cea066f6a9757c564668f1d9e1017485d4ec63ea0ea0c72e6 result d5af2cb20c8c25b79c557c15faaa849cb18c596347dff7989a60efc3dc8abfed
Exploratory work on development data; not a performance claim.