Home › Guides › Why favorites lose

Can a 90% favorite lose?

Updated September 25, 2026 · numbers refresh automatically from this site's data

Yes. A 90% price means the outcome should fail about one time in ten. In our sample of 1,918 resolved Polymarket markets that closed between September 25, 2025 and September 21, 2026, outcomes priced 90% or more a week before the close happened 95% of the time: 3 of 62 lost. Long shots priced 20% or less came through 82 times out of 1,413 (5.8%, against an average price of 3.2%).

How often each price level came true

If prices are well calibrated, outcomes priced near 90% should happen about 90% of the time, and those priced near 10% about 10% of the time. Here is the full sample, priced seven days before each market closed:

Price a week before closeMarketsAverage priceHappened
0–10%1,2431.7%3.9%
10–20%16514.2%19.4%
20–30%11624.5%25.9%
30–40%7834.8%42.3%
40–50%7045.1%45.7%
50–60%7154.3%60.6%
60–70%4965.0%73.5%
70–80%4174.7%63.4%
80–90%2385.1%87.0%
90–100%6296.1%95.2%

Method, caveats and the data file: How accurate is Polymarket? Small bands move a lot by chance.

Recent upsets

Outcomes that were priced at 20% or less a week before the close and still happened:

MarketOutcome that wonPrice a week outClosed
What price will Bitcoin hit in August?↑ 80,0000.5%August 24, 2026
What price will Bitcoin hit in August?↑ 77,5001.1%August 21, 2026
What price will Bitcoin hit in August?↑ 75,0002.6%August 21, 2026
What price will Bitcoin hit in August?↑ 72,5007.5%August 20, 2026
What price will Bitcoin hit in August?↑ 70,00015.0%August 19, 2026
NBA: LeBron James Next TeamPhiladelphia 76ers8.9%July 27, 2026
2026 Women's Wimbledon WinnerLinda Nosková4.5%July 11, 2026
Argentina vs. Cabo VerdeDraw (Argentina vs. Cabo Verde)11.5%July 4, 2026

Full list: biggest Polymarket upsets.

Why favorites lose

Short, mechanical markets behave differently: in 1,008 daily Bitcoin and Ethereum price strikes, the side priced above 50% a day ahead won 95% of the time, and all 527 strikes priced 90% or more happened (crypto accuracy study).

What to do with this

Size positions so that a favorite losing does not hurt: the Kelly calculator gives a stake for your edge and the risk of ruin calculator shows how often a stake size halves a bankroll. The calibration and favorite–longshot bias entries explain the terms.

Frequently asked questions

Can a 90% favorite lose on Polymarket?

Yes. A 90% price means the outcome should fail about one time in ten. In our sample of 1,918 resolved Polymarket markets that closed between September 25, 2025 and September 21, 2026, outcomes priced 90% or more a week before the close happened 95% of the time: 3 of 62 lost. Long shots priced 20% or less came through 82 times out of 1,413 (5.8%, against an average price of 3.2%).

How often do Polymarket favorites win?

Outcomes priced at 50% or more a week before the close happened 75% of the time in our sample of 1,918 resolved markets.

Do long shots ever win?

Yes: 82 of 1,413 outcomes priced 20% or less a week out happened in our sample.

Polymarket View is independent and not affiliated with Polymarket. Educational information, not financial advice. All guides · How market types work · Glossary · Polymarket odds today