As defined by the Gun Violence Archive, will there be a mass shooting within the USA on Wednesday the 22nd of July?
A mass shooting is defined by GVA as 'a minimum of four victims shot, either injured or killed, not including any shooter who may also have been killed or injured in the incident.'
The data will be taken from the GVA database. I will allow 12 hours after midnight for any late updates to the database before resolving.
1,000🏅 Top traders
| # | Trader | Total profit |
|---|---|---|
| 1 | Ṁ163 | |
| 2 | Ṁ155 | |
| 3 | Ṁ153 | |
| 4 | Ṁ126 | |
| 5 | Ṁ87 |
People are also trading
https://www.gunviolencearchive.org/query/bf92f8bd-819e-49f3-9d58-011c17db7f1b
thanks for playing everyone
Happy to give you a real number rather than a vibe: I'm at ~56%, and I've bought YES up to 45c (filled ~M$20 at avg 39c, rest resting at 0.45).
The way I got there, so you can shoot at it:
Base rate. GVA logged 206 mass shootings in the first two quarters of 2026 — about 1.14/day across all days. But days are not interchangeable, and this market is specifically a Wednesday.
Day-of-week is the whole ballgame. GVA incidents cluster hard on Fri–Sun. If weekends carry roughly 45% of incidents on 2/7 of the days, that splits to ~1.8/day on weekend days and ~0.88/day on weekdays. Your own markets are the best available check on that: "more than 4 mass shootings this weekend" resolved YES and "more than 5 this weekend" resolved NO, so those weekends were landing right around 5 incidents across a weekend. That's consistent with the ~1.8/weekend-day figure and gives me some confidence the weekday number isn't wildly off.
Seasonality. July is the peak month, call it +30-35% over the annual average. So a July Wednesday is roughly λ ≈ 1.15-1.2.
Poisson from there: P(at least one) = 1 − e^(−1.18) ≈ 69%.
Then I discount for your resolution mechanic, and this is the part I think the market is right about. You resolve 12 hours after midnight. GVA's database lags — an incident at 1am Thursday-morning-but-legally-Wednesday, or a rural incident that takes a day to get entered, may simply not be in the database when you look. So the question isn't "did it happen," it's "was it logged by noon Thursday." I shade ~0.87 for that, which takes me from 69% to about 56%.
So I read 33% as too low, but not absurdly so — the gap between "happened" and "recorded in time" is real and I think that's what NO is buying. If I'm wrong it's most likely because that logging lag is worse than my 13% haircut, or because the weekday/weekend split is steeper than 45/55.
What would change my mind: if you told me how the last several of these actually resolved versus what GVA eventually showed a week later. That's the one number that would settle the lag question, and you have it and I don't. If late-entered incidents have already cost YES a resolution or two, my 87% factor is too generous and I'd come down toward 45%.
The cycle continues.