Rare Day, Oct 16, 2026: VIX rise on a small S&P 500 drop, only 8 other sessions since 1990
Session 2026-10-16, from Cboe daily index closes. Card published 2026-10-17 15:58:09 UTC.
Next card: the Oct 19 session, after 6:30 a.m. PT (9:30 a.m. ET) on Tuesday, Oct 20, 2026.
Each session after the close, 21 S&P 500 and VIX stats are counted against every earlier session in Cboe's daily histories (S&P 500 from 1975, VIX from 1990, the newer indexes from their first year). A stat earns a card when its reading was this rare or rarer on 1% or fewer of earlier sessions. Every earlier match is listed with its date, beside what the S&P 500 did after it and after any session.
Not a forecast. These are counts of past sessions and what the S&P 500 did after them; past sessions say nothing certain about the next one, and nothing here is advice.
VIX rise on a small S&P 500 dropSMALL SAMPLE
The VIX rose 3.53 points while the S&P 500 fell 1.09%. Only 8 other sessions since 1990 had a VIX rise of 3.53 points or more on an S&P 500 drop of 1.09% or less.
Calibration: on 0.93% of earlier sessions, this stat's reading was this rare or rarer.
8 matching sessions, 8 after de-clustering (one per 20 trading days, earliest kept). Small sample: fewer than 10 sessions remain.
What the S&P 500 did after
Horizon
After the 8 counted matches
Any session since 1990
Compared
1 week later
higher in 7 of 8
median +1.91%, worst -1.66%, best +5.94%
higher in 58% of 9,259 sessions since 1990
median +0.33%
about the same as any session
1 month later
higher in 8 of 8
median +2.99%, worst +0.64%, best +16.96%
higher in 64% of 9,243 sessions since 1990
median +1.24%
about the same as any session
S&P 500 change 1 month after each counted match, by match date. Dashed line: the median for any session since 1990.
Every match
Session
Counted
1 week later
1 month later
Context
1991-01-14
yes
+5.94%
+16.96%
1998-09-03
yes
+2.73%
+0.64%
2001-10-12
yes
-1.66%
+2.44%
Stocks fell about 2% midday after an anthrax case in an NBC employee. Dawn (Reuters)
2016-06-13
yes
+0.20%
+3.53%
Dow fell 133 points as UK polls leaned Leave before Brexit vote. Investing.com
Fed minutes showed a 2021 bond purchase taper was possible; Dow fell 383 points. UPI
2025-10-16
yes
+1.65%
+1.58%
Zions and Western Alliance disclosed fraud-tied loan losses; bank shares slid. Bloomberg Law
How a card is counted
For each stat, N counts the earlier sessions that matched or beat this one on every part of the stat, and M counts the earlier sessions with every input present. The sentence keeps that raw count. Whether a card shows is decided separately, by calibration: the share of earlier sessions on which this stat's own reading was this rare or rarer. That share is not the count; a two-part stat such as a VIX rise on a small drop can have few matches and still a larger calibration share. Two views of the same VIX jump do not both get a card; at most one card per family shows, three in all.
What followed, and why it is shown this way
Every match is used, never a picked example. Matches within 20 trading days of a kept match count once, so one stretch like October 2008 is not counted five times; the page gives the count before and after. Beside the matches sits the same number for any session in the stat's history, and the comparison reads "about the same as any session" unless an exact binomial test separates them. Fewer than 10 counted matches is marked small sample. A context label appears beside a date only where a named source supports it.
Source
S&P 500, VIX, VIX9D, VIX3M, SKEW and VVIX daily histories from Cboe, archived once per night. When a night's files have not arrived, this page keeps the last published card with its own session date. The same card is in plain text at /rare-day.md and as JSON at /api/public/rare-day.