Sportsbook Odds Report: September 2026
DraftKings was at or tied for the best available price more often than any other venue with unaffected pricing in September 2026, leading on 57.0% of the 2,568 sides it quoted, which were 13.0% of every price recorded that month. Pinnacle was second at 41.4% of 1,002 sides, then Caesars at 31.4% of 1,730 sides.
Sample: 19,776 recorded prices across 20 sportsbooks and prediction markets, 2026-09-01 to 2026-09-01 (1 day with recorded scans, out of 30 in September).
This month has recorded scans on 1 of its 30 days. Every figure below covers only the days listed, and is not scaled up to a full month.
Which sportsbook had the best odds most often in September 2026?
Best-price share is the share of the sides a venue quoted where its price was at or tied for the highest payout of every venue quoting that same side. Each venue is measured against its own coverage, so a book that prices fewer games is not penalised for the games it skipped. Sample size is listed for every row because the denominators differ.
Coverage is the share of every price recorded this month that came from that venue. It is the check on the share beside it: a high best-price share on low coverage means the venue won most of the few sides it chose to quote, not that it beat the market. With 20 venues recording prices, an even split would put each near 5.0%.
- DraftKings: 57.0% of 2,568 sides quoted, over 1 day (13.0% coverage; worst price on 48.2%)
- Pinnacle: 41.4% of 1,002 sides quoted, over 1 day (5.1% coverage; worst price on 4.8%)
- Caesars: 31.4% of 1,730 sides quoted, over 1 day (8.7% coverage; worst price on 48.7%)
- LowVig: 27.7% of 754 sides quoted, over 1 day (3.8% coverage; worst price on 8.0%)
- Novig (prediction market): 25.7% of 584 sides quoted, over 1 day (3.0% coverage; worst price on 12.5%)
- Bally Bet: 23.9% of 1,246 sides quoted, over 1 day (6.3% coverage; worst price on 20.1%)
- Bovada: 23.9% of 1,376 sides quoted, over 1 day (7.0% coverage; worst price on 22.7%)
- BetParx: 22.9% of 1,246 sides quoted, over 1 day (6.3% coverage; worst price on 20.5%)
- Kalshi (prediction market): 22.4% of 688 sides quoted, over 1 day (3.5% coverage; worst price on 27.2%)
- Hard Rock Bet: 20.2% of 820 sides quoted, over 1 day (4.1% coverage; worst price on 15.6%)
- BetOnline: 19.4% of 830 sides quoted, over 1 day (4.2% coverage; worst price on 12.4%)
- BetRivers: 18.1% of 1,244 sides quoted, over 1 day (6.3% coverage; worst price on 27.0%)
- FanDuel: 17.1% of 1,050 sides quoted, over 1 day (5.3% coverage; worst price on 21.4%)
- theScore Bet: 16.7% of 962 sides quoted, over 1 day (4.9% coverage; worst price on 27.7%)
- BetMGM: 13.3% of 968 sides quoted, over 1 day (4.9% coverage; worst price on 26.9%)
- Polymarket US (prediction market): 13.3% of 482 sides quoted, over 1 day (2.4% coverage; worst price on 19.1%)
- Fanatics: 11.4% of 832 sides quoted, over 1 day (4.2% coverage; worst price on 9.6%)
- MyBookie: 10.4% of 740 sides quoted, over 1 day (3.7% coverage; worst price on 25.0%)
- ProphetX (prediction market): 9.5% of 422 sides quoted, over 1 day (2.1% coverage; worst price on 12.6%)
- Fliff: 0.4% of 232 sides quoted, over 1 day (1.2% coverage; worst price on 47.8%)
Do prediction markets beat sportsbooks on price?
September 2026, measured the same way on both sides. Each group's denominator is the sides its own members actually quoted, so this is not distorted by the exchanges covering fewer games.
| Venue type | Venues | Sides quoted | At or tied for best |
|---|---|---|---|
| Prediction markets (Novig, Kalshi, Polymarket US, ProphetX) | 4 | 2,176 | 18.8% |
| Traditional sportsbooks | 16 | 17,600 | 26.9% |
Exchange prices in this comparison are fee-adjusted: Kalshi and Polymarket US include each venue's taker fee and ProphetX its commission on winning bets, so an exchange number is what a taker actually pays. We publish this comparison and not a "share of all best prices" figure, because best-price credit is shared among tied venues, which would hand the larger group an advantage that is an artefact of counting rather than of pricing.
Which book led each sport?
No sport in September 2026 reached the sample this site requires for a per-sport ranking.
Sports with insufficient sample
These sports were scanned in September 2026 but did not reach 2,000 recorded sides across at least 3 days, so no leader is named for them. The counts are real; they are simply too small to rank on.
- NFL: insufficient sample (n = 4,018 sides over 1 day)
- NCAAF: insufficient sample (n = 3,888 sides over 1 day)
- Tennis: insufficient sample (n = 1,954 sides over 1 day)
- MLB: insufficient sample (n = 1,516 sides over 1 day)
- MLS: insufficient sample (n = 1,108 sides over 1 day)
- EPL: insufficient sample (n = 1,012 sides over 1 day)
- NHL: insufficient sample (n = 910 sides over 1 day)
- Serie A: insufficient sample (n = 862 sides over 1 day)
- La Liga: insufficient sample (n = 840 sides over 1 day)
- Bundesliga: insufficient sample (n = 780 sides over 1 day)
- NBA: insufficient sample (n = 754 sides over 1 day)
- Ligue 1: insufficient sample (n = 682 sides over 1 day)
- Champions League: insufficient sample (n = 600 sides over 1 day)
- UFC: insufficient sample (n = 586 sides over 1 day)
- WNBA: insufficient sample (n = 266 sides over 1 day)
What was the widest price gap of the month?
The largest single moneyline gap recorded in September 2026 was 8.8 percentage points of implied probability, on Philadelphia Eagles at San Francisco 49ers (NFL). Caesars paid +150 on Philadelphia Eagles, where DraftKings paid +105.
Drawn from 848 gap observations recorded in September 2026. This is a single recorded scan, not an average.
How is this measured, and what does it not measure?
On every scan we record each venue's price on both sides of every game, then count how often each venue was at or tied for the highest payout on a side it quoted. Ties credit every venue holding the top price, so a book that consistently matches the market's best is not scored as though it lost. The full arithmetic, including how prediction-market fees are applied before any comparison, is on the methodology page.
This is best-or-tied share of each venue's own quoted sides. It is not the matched-lines metric. Because the denominator is only the sides a venue chose to quote, this measure rewards selective quoting: a venue that prices a narrow slice of the market and skips the rest can post a high share without ever beating the field on a like-for-like bet. That is what the coverage column beside each figure is there to expose.
The stricter measure, and this site's single source of truth for who actually has the best prices, is matched lines: it keeps only the games where every venue in the comparison quoted the same market at the same line, and ranks who won the price on those. It discards far more data and is much harder to lead. On that measure the exchanges do not lead, and no book dominates. The live matched-lines view is on Book Comparison, and the reasoning is set out in the correction of July 26, 2026. Where the two measures disagree, matched lines is the one to believe.
Data: 2026-09-01 to 2026-09-01. Generated September 01, 2026 at 01:07 AM ET. September 2026 is still in progress; this page updates daily until the month closes, then it is final.
The Odds Gap publishes where the same bet paid more. It does not sell picks and makes no income claims. Always verify a price at the book before placing any bet.