After a large one-day fall that a company filing dates, do the overnight gaps carry a further decline of at least 0.5%?
Study 75 · Concluded · recorded 29 October 2026 · answer: Not at the size tested · full paper with figures
The question
After a large one-day fall that a company filing dates, the smaller listed shares we follow kept falling for three sessions. We ask whether the two overnight gaps, close to next open, carry a further decline of at least 0.5% net of quiet comparison shares, a size fixed in advance at a third of an earlier study's measured decline.
Rules fixed before the result
Fixed before any outcome: the answer reads the 95% range of the two overnight gaps' summed return, net of comparison shares and of the same gaps 30 sessions earlier, a continued decline counted positive. A range starting at or below zero and ending below 0.5% meant no overnight part that size; starting above zero but below 0.5%, a visible part; starting at 0.5% or more, that size; otherwise, open. The study first needed 6,000 kept falls and the planned precision.
Data and method
Each fall's two overnight gaps between the next three sessions were summed, less the same sum 30 sessions earlier, and netted against quiet shares measured alike on the same days, in the same volatility tenth and price band (cheaper or dearer shares, split at one fixed price). The deciding prices were the source's, adjusted for splits and payouts; our own adjustment of its unadjusted prices agreed. The 95% ranges allow for falls on neighbouring dates and repeat companies.
The answer
Not at the size tested: the range rules out an overnight part of 0.5% or more. Net of comparison shares and of the same gaps 30 sessions before each fall, the two overnight gaps over 7,238 falls rose 0.23 percentage points (95% range 0.04 to 0.41), while the three trading-hours parts fell 1.42 points (1.06 to 1.77). In the source's daily opening and closing prices, the decline was found in trading hours, not overnight. Without the falls whose second or third opening price may not reflect trading (equal to the previous close, on no volume, or far from it), the overnight range held zero on the remaining 4,692 falls (a rise of 0.24 to a fall of 0.11 points).
Main estimate
-0.002 fraction (95% range -0.004 to -0.000), from 7238 observations
What it changes
By the rule set in advance, the result is consistent with the account under test, that holders' selling in trading hours carries the further decline, on where the decline happens in the source's daily opening and closing prices. Because it leaned on falls whose opening prices may not reflect trading, the next choice of study weighed those falls first: without them, the overnight part still stayed below 0.5%. Whether trading volume carries the decline was not settled.
What remains uncertain
These falls were studied before, so this is a check within our development data, not a test on fresh data. Opening prices come from one source: a biased opening price would move the decline between the gaps and trading hours, and falls with no filing showed a similar overnight rise. The result locates the decline in daily prices; it does not show why shares fell.
How it was made
Run took 474 minutes; independent code review: approve; the recorded calculation reruns to the same result.
Fingerprints
plan f29f4092d1cbd5a2528879824979fe5db2ee109b316f25701fae93badf090943 result bec8e6b7354fa2211000df421245239f81d36e36c1a8b980f0824b62358bdff3
Exploratory work on development data; not a performance claim.