Do the most speculative shares gain overnight and give it back during the trading day?
Study 72 · Changed a rule or plan · recorded 27 October 2026 · answer: Yes · full paper with figures
The question
Each week we rank shares priced at five or more on volatility, low price, largest daily rise and thin trading, comparing the most with the least speculative fifth over the next week. We split that gap into its overnight part (close to open) and its daytime part (open to close), asking whether the first is positive and the second negative.
Rules fixed before the result
Fixed before any outcome: the answer is read on the 95% range of the weekly gap against zero and a size of 0.35 percentage points a week, and the word is chosen by the overnight and daytime parts' ranges. Yes needed the gap's range above zero and reaching 0.35, with the overnight part's range above zero and the daytime part's below. It required 200 weeks, and a check on resampled earlier years held each side's false-alarm rate to 5%.
Data and method
Shares priced at five or more at each week's last close were ranked into fifths on four traits; their overnight and daytime returns, payout- and split-adjusted, were summed over the next week. Before 2018 to 2022 was read, 2013 to 2017 showed the gap detectable at 0.32 percentage points a week with market moves held flat, 0.38 without, above the 0.35 size, so, as fixed in advance, the flat-market comparison decided. The 95% ranges allow for neighbouring weeks' correlation.
The answer
Yes. Over 260 weeks of 2018 to 2022, with the market's own weekly moves held flat, the most speculative fifth's overnight return less its daytime return, measured against the least speculative fifth's, averaged 0.77 percentage points a week, with a 95% range of 0.40 to 1.14, wholly above the 0.35 set in advance. The overnight part alone was 0.32 (0.09 to 0.54) and the daytime part -0.46 (-0.71 to -0.21), so the gain before the open was given back, and more, during the day. Without holding the market flat the gap was 0.95 (0.59 to 1.30). On the week's first day alone the overnight part was 0.16 (0.09 to 0.23) and the daytime part -0.18 (-0.32 to -0.04).
Main estimate
0.772 percentage points (95% range 0.405 to 1.140), from 260 observations
What it changes
By the rule set in advance, the project proposes, for its records, that every later study of a gap between speculative and other shares reports the overnight and daytime parts side by side, and the first day's daytime loss sends to the owner the idea of buying such shares after the open, never at it. The plan expected this answer at low confidence, chance 0.37, but a flat market gap of 0.25 to 0.45 points a week, smaller than found.
What remains uncertain
Earlier studies used these years, so this is a check on development data, not a fresh test. No quotes are held, and an opening trade may print near the asking price, inflating the overnight gain where trading is costly; it held at 0.20 (0.04 to 0.37) in the lowest-cost third, cost estimated from daily closes, shares whose estimate was zero left out. The 2020 weeks carried the largest gap; without any one year it stayed at 0.58 or more.
How it was made
Run took 41 minutes; independent code review: approve; the recorded calculation reruns to the same result.
Fingerprints
plan 4e7ffe8df8ec6c3808a09ec73cdfae854d84b3ed26f59c9f97356ebacfe1533d result dffac4a3d5a5184693b8387df3c4c09c0df0d76d9d081af7041e3e1e765ac523
Exploratory work on development data; not a performance claim.