SAMPLING ARCHIVE
Margin History: Every Sampling Round, Archived
A margin index only matters if it repeats. One snapshot tells you what a book charged on one afternoon; a series tells you whether that price is policy or accident. This page archives every OverroundHQ sampling round, oldest at the bottom, so the record can't quietly change.
Why we keep a public log
Sportsbooks re-price constantly. A book can run tight football prices during a World Cup, then widen once the promotional window closes. If we only published a live number, that history would evaporate. So every round gets a dated entry here with the raw per-book, per-market ranges, and future rounds get appended — never overwritten. If you want to know how the numbers are produced, the sampling methodology covers market selection, timing, and the de-vig math; the de-vig calculator lets you reproduce any figure from the recorded odds.
Round 1 — 8 July 2026
Our first round: 15 odds boards sampled across four crypto books on the same day, World Cup 1X2 fixtures where possible so books are compared on identical events. Every figure below is a raw overround from that single day. One round so far — medians unlock at n≥6 per book-market, live and prematch are never averaged together, and any figure older than 14 days is treated as stale. Treat these as first readings, not verdicts.
Bar length scales with the highest overround sampled in each range. All samples dated 8 Jul 2026; every book-market pair is below our n≥6 bar for medians.
Two things stand out already, with the caveat that nothing here is statistically settled. First, the same book can sit at both ends of the chart: Duel priced World Cup football tighter than anything else we measured, then posted a 13.71% overround on a live Dota 2 3-way. Both of Duel's esports samples were in-play, which structurally carries wider margins, so that spread is partly a live-vs-prematch effect — one reason we refuse to blend the two states. Second, on identical fixtures (France–Morocco and Spain–Belgium appeared on three books), the overround varied by more than a factor of three between operators. Full per-fixture odds are on the football 1X2 board and the esports board.
What future rounds will add
The archive gets more useful with every entry:
- Per-book trend lines. Once a book-market pair crosses n≥6, we publish a median and start showing direction: is Duelbits's ~4% football pricing stable, or was 8 July a good day?
- Claim-vs-measured tracking. Where an operator has published a margin claim, we log it next to what we actually sample. Cloudbet's 2020 blog claimed margins "typically of 2%-3%"; our first World Cup 1X2 reading ran 4.90–5.48%. One round doesn't refute a claim — a series either confirms it or buries it.
- Staleness flags. Any figure not re-sampled within 14 days is marked stale on the front-page index and stays here only as history.
Reading this page honestly
Small samples wobble. A single tennis reading (Cloudbet, in-play, 5.55%) is a data point, not a rating. We publish it anyway because hiding low-n data until it looks tidy is how vendor charts are made — the n-badges and the LIVE tags are there so you can weight each number yourself. If a figure looks surprising, punch the recorded odds into the calculator and check our arithmetic.
Next round: sampled per the schedule in the measurement rules. New entries appear above Round 1.