How accurate are Polymarket’s NBA odds? Favourites won 103 of the first 145 games of 2026-27, and every one priced 80% or more won
We took Polymarket’s last price before tip-off for 145 of the 146 games in the first three weeks of the 2026-27 NBA regular season and compared it with the result. Favourites won 71% of the time against an average price of 67%. Teams priced at 80% or more won all 22 of their games, while near coin flips were exactly that. Nearly $300 million traded on the match-winner markets alone.
In this article
Polymarket’s pre-game NBA prices have been a little too cautious about strong favourites so far this season. We took the last price before tip-off for 145 of the 146 regular-season games played from 20 October to 8 November 2026 and compared it with the result. The favourite won 103 games, 71%, against an average price of 67%.
The gap came from the top of the range. Teams priced at 80% or more won all 22 of their games, and teams priced 70% to 79% won 82% against an average price of 74%. Close games, priced 50% to 59%, went to the favourite 58% of the time, almost exactly the price.
The short version
Source: Polymarket price history and ESPN results; PredictDesks analysis
- Favourites won 103 of 145. Their average price was 66.9%; they won 71.0%.
- Heavy favourites were too cheap. All 22 teams priced at 80% or more won, against an average price of 85%.
- Close games were coin flips. Favourites priced 50-59% won 29 of 50.
- Home court was worth little. Home teams won 52% of games, against an average price of 55%. Away favourites won 41 of 54, 76%.
- The markets are big. The match-winner markets on these games traded $298 million, a median of $1.6 million a game; all markets on the games together traded $502 million.
How we measured it
For every regular-season game from opening night on 20 October to the games of 8 November (US time), we found Polymarket’s game market and took the last traded price of the match winner at least five minutes before the scheduled tip-off, from Polymarket’s price history at ten-minute resolution. The favourite is the team priced above 50%. We matched each game with the final score on ESPN’s scoreboard.
ESPN lists 146 completed regular-season games in that window. We found Polymarket prices for 145; the one we could not match was the Spurs’ win at Sacramento on 6 November. Preseason games are left out.
We also scored the prices with the Brier score, the average squared gap between the price and the result, where 0 is perfect and always saying 50% scores 0.25 (Brier score explained). Polymarket’s pre-game prices scored 0.185.
Three weeks is a short time. With 145 games, a win rate can be several points away from the true figure by chance alone, so treat the patterns below as early evidence, not a law.
Results by price
Source: Polymarket price history and ESPN results; PredictDesks analysis
Show the dataHide the data
| Value | |
|---|---|
| 50-59% | 58.0% |
| 60-69% | 62.5% |
| 70-79% | 81.8% |
| 80%+ | 100.0% |
The bottom half of the scale lined up well. Favourites priced in the 50s won 58% of the time against an average price of 56%. Those in the 60s won 62.5% against 65%, slightly below the price.
The top half did better than its prices. Favourites priced in the 70s won 27 of 33, 82% against 74%. Favourites priced at 80% or more won 22 of 22, against an average price of 85%. At those prices, three or four losses would have been normal; there were none.
The most frequent heavy favourites were the Spurs, priced at 80% or more in six games, and the Celtics, in five. The highest price before any game was 92.5% for the Spurs at home to the Kings on 27 October; they won.
The upsets
The biggest upsets by price were all favourites in the low to mid 70s:
- Kings at Mavericks, 2 November: Dallas was 74.5%; Sacramento won 115-100.
- Grizzlies at Trail Blazers, 29 October: Portland was 72.5%; Memphis won 124-103.
- Rockets at Suns, 7 November: Houston was 71.5% on the road; Phoenix won 102-99.
- Timberwolves at Rockets, 4 November, Warriors at Nuggets, 26 October and Mavericks at Heat, 26 October: each favourite was 70.5% and lost.
Of the 42 games the favourite lost, the average final margin was 10.4 points; when favourites won, they won by 14.6 on average.
Home and away
Source: Polymarket price history and ESPN results; PredictDesks analysis
Show the dataHide the data
| Value | |
|---|---|
| Away favourites: won | 75.9% |
| Away favourites: priced | 65.7% |
| Home favourites: won | 68.1% |
| Home favourites: priced | 67.6% |
Home teams won 75 of 145 games, 52%, while the market priced them at 55% on average. Home favourites won 68% of the time, almost exactly their average price of 68%. Away favourites did much better than their price: 41 wins in 54 games, 76%, against an average price of 66%.
The sample is small, and early-season schedules are uneven, so this may not last. But it is the opposite of the usual worry that markets underrate home court.
Week by week
Source: Polymarket price history and ESPN results; PredictDesks analysis
Show the dataHide the data
| Value | |
|---|---|
| 20-26 Oct | 70.2% |
| 27 Oct-2 Nov | 66.7% |
| 3-8 Nov | 75.5% |
Favourites beat their prices in the first and third weeks and matched them in the second. Prices also became more confident as the season went on: the average favourite was priced at 64.7% in the first week and 69.2% in the third, as the markets learned which teams were good.
Late moves
We also took each game’s price 24 hours before tip-off. In 10 games the favourite changed in the last day. The team that became the favourite won 5 of those 10. Late flips are where the market has the most new information, and in this small sample it was no better than a coin flip at using it.
Team by team
Source: Polymarket price history and ESPN results; PredictDesks analysis
Show the dataHide the data
| Value | |
|---|---|
| 76ers (9-2) | 2.5 |
| Lakers (9-2) | 2.4 |
| Warriors (7-3) | 2.3 |
| Timberwolves (8-3) | 2.0 |
| Mavericks (6-4) | 2.0 |
Adding up each team’s pre-game prices gives the number of games the market expected it to win. Comparing that with real wins shows which teams have surprised the market most.
- 76ers: expected to win about 6.5 of their 11 games; they won 9. They were favourites in seven and won six, and underdogs in four and won three.
- Lakers: about 6.6 expected, 9 won, including both games as an underdog.
- Warriors: 4.7 expected, 7 won, three of them as underdogs.
- Timberwolves: 6.0 expected, 8 won.
- Mavericks: 4.0 expected, 6 won, four of them as underdogs.
At the other end, the Jazz won one game when the prices expected about three, and the Bucks won one of ten against 2.9 expected. The Spurs matched the market almost exactly, 7.0 expected and 7 won in the games we checked (they also won the game at Sacramento that we could not match). The Celtics were favourites in nine of their ten games and won eight of them.
Three weeks in, most of these gaps are within normal luck: a difference of two wins over ten games is not unusual. But they show where the market’s early-season view of a team has been furthest from the results.
Underdogs
The flip side of strong favourites winning is that big underdogs have had a thin season. In games where the favourite was priced at 70% or more, the underdog won 6 of 55. In the closest games, priced 50-59%, the underdog won 21 of 50, 42%, close to its average price of 44%.
That fits what is often called the favourite-longshot bias, where long shots attract more money than their chances justify and end up slightly overpriced. Over three weeks it is a hint, not proof, but it points the same way as the strong favourites’ record above (how accurate prediction markets are overall).
How this compares with the NFL
We ran the same test on Kalshi’s NFL prices for the first eight weeks of the 2026 season (our NFL study). There, favourites won 68 of 107 games, close to what their prices implied, and the eight favourites priced above 80% went 4-4. In the NBA so far, the 22 favourites above 80% went 22-0.
Heavy favourites are also more common in basketball: 22 of 145 NBA games, 15%, had a favourite above 80%, against 8 of 107 NFL games, 7%. The NBA markets have priced the middle of the range about right and the top too cautiously.
The busiest games
Source: Polymarket public market data
Show the dataHide the data
| Value | |
|---|---|
| Cavaliers at 76ers, 22 Oct | $11.2m |
| Trail Blazers at Lakers, 27 Oct | $10.1m |
| 76ers at Knicks, 20 Oct | $9.2m |
| 76ers at Hornets, 31 Oct | $9.2m |
| Nets at Celtics, 27 Oct | $8.2m |
Opening night’s 76ers at Knicks, where New York was 61.5% and Philadelphia won, drew $9.2 million. The busiest game so far, Cavaliers at 76ers on 22 October, traded $11.2 million; the 76ers were a 54.5% favourite and won. Trail Blazers at Lakers drew $10.1 million and was another upset at a near coin-flip price.
What this means if you trade NBA games
- Near coin flips are fairly priced. There is no obvious edge in games priced 50-59%.
- Strong favourites have been too cheap so far. Teams at 70% and above won 49 of 55, 89%, against an average price of 79%. Whether that lasts is the question for the rest of the season.
- Do not overpay for home court. Home teams won less often than their prices implied.
- Fees matter at the top. On a favourite at 85¢ the upside is 15¢ a share, so a taker fee and a spread eat a larger share of it (Polymarket fees).
For how NBA markets work on Kalshi and Polymarket, see how to bet on the NBA. For the same test on NFL games, see our NFL study, and for the season-long markets, NBA MVP odds and NBA championship odds.
What the prices do not tell you
A price before tip-off is a snapshot. It already includes the injury report and the rest decisions known at the time, and it can move sharply in the last hour. We used the last trade at least five minutes before the scheduled start, so a late scratch announced after that point is not in our prices, and a game that started late is measured against its scheduled time.
We also looked only at the match-winner market. Polymarket lists spreads, totals and player props on most games, and those markets may be priced better or worse than the winner market. And the sample covers the first three weeks of the season, when teams are still settling rotations and the market has little current-season data to go on. The same test after the All-Star break could look quite different.
Tonight’s slate
On 9 November, Polymarket’s prices for the evening’s games put one team, Detroit, in the 80%-plus range where favourites have won every game so far, and four more in the 70s, where favourites have won 82%:
- Pelicans at Warriors: Golden State 78.5%.
- Wizards at Pistons: Detroit 81.5%.
- Spurs at Suns: San Antonio 77.5% on the road.
- Cavaliers at Hornets: Cleveland 76.5%.
- Nuggets at Rockets: Houston 73.5%.
- Grizzlies at Raptors and Pacers at Mavericks: the home team 58.5% in both.
- Lakers at Hawks: Atlanta 50.5%, a pure coin flip.
Live prices change through the day as injury reports come out, so check the market before you trade. This study will tell you how often prices like these have held up, not what will happen tonight.
Data
Every game, with the favourite, its price before tip-off and 24 hours earlier, the score and the volume, is free to download: nba-2026-27-polymarket-game-prices.csv.
Games to 8 November 2026 (US time). Prices from Polymarket’s public price history; results from ESPN. Nothing here is financial advice.
