Which day of the week is most volatile for stocks?

Two pieces of weekday folklore about stocks get repeated in trading rooms more than almost anything else. Monday is volatile, because a weekend of news gets priced in at once. Friday is quiet, because nobody wants to carry a position into the weekend. Both are plausible. Neither holds up.

1. The range by weekday

We took 180 sessions and, on each, the median intraday range across every US symbol with a prior close of at least $1 and $1,000,000 of turnover. A median of 6,186 symbols per session. Then we grouped the sessions by the day they fell on. Thursday is widest. Tuesday is narrowest. Monday comes second, well below Thursday and only just above Wednesday.

2. Most of the difference is not there

The quartiles in that table overlap almost completely, and there are only 34 to 38 sessions behind each row. Reading a winner off a table like that is how noise becomes a trading rule. So we tested it directly. Shuffle the weekday labels across the 180 sessions, recompute the gap between the highest and lowest weekday, and repeat 2,000 times. If a gap as large as the real one turns up often under random labels, the weekday is not doing anything. Dollar volume has no weekday effect whatsoever. A gap of $62.4B between the busiest and quietest weekday is what random labels produce three quarters of the time. The range result is more interesting, and it needs to be read carefully.

3. Thursday, with a caveat

A p of 0.022 means random weekday labels produce a range gap this large about one time in forty-five. On its own that would count as a finding. It is not on its own. We ran three tests, and the more tests you run the more likely one of them clears an arbitrary line by chance. 0.022 does not clear a correction for having run three tests at once. So the honest reading is that Thursday's wider range is suggestive and consistent across the window, not established. There is also a plausible mechanism, which we did not test. US weekly jobless claims are released every Thursday morning before the open, and scheduled macro data is exactly the kind of thing that widens a whole market's range on a known day. If that is the cause, it would show up as a Thursday effect concentrated in the morning. That is a testable follow-up, and until it is tested it is a hypothesis.

4. What the folklore gets wrong

The Friday result deserves a second look, because it is the opposite of the story. Friday's dollar volume is the highest median of the five days, though the permutation test says that ordering is indistinguishable from chance. What the data rules out is Friday being systematically quiet.

5. What this means if you trade

The day of the week is a weak variable at best. If you are choosing which days to trade on the basis of Monday or Friday folklore, the data does not support you. If you have noticed Thursdays feeling wider, that is consistent with what we measured, but not strong enough to plan around without a longer sample. The variables that actually move day to day, and move a lot, are the ones inside the session. We measured when the high of a big move tends to print in what time of day do big movers peak, and the time-of-day structure there is far stronger than anything the weekday produces. Limitations Nine months is short for a calendar effect, and 34 to 38 sessions per weekday leaves the medians sensitive to a handful of unusual days. A single volatile week can shift one weekday's median, and the permutation test does not account for sessions clustering in time: if a stretch of volatile Thursdays all fell in one turbulent month, the test would read that as a weekday effect. Holidays shift the calendar. A week with a Monday holiday has no Monday session, and the Tuesday that follows absorbs a longer gap; we did not adjust for that. The range is measured on regular-session highs and lows, so pre-market moves after weekend news are outside it. A Monday effect that lives entirely before 09:30 would not appear here. The universe includes funds, warrants and preferred issues alongside common stock, filtered only by price and turnover.