Are Polymarket Sports Prices Right? 34,065 Markets Checked

The short answer

Mostly, yes. In Polymarket "who wins" markets, the price before the game predicted the result about as often as it promised. The gaps we found were about the size of the fee or smaller, or they changed direction when we checked them again on new data.

We checked 34,065 closed match-winner markets in 17 sports and esports from 29 June to 26 September 2026. We tested nine ideas for beating the price. None of them gave an advantage over the price that was still there after the fee.

This covers "who wins" markets before the start, in one 90-day summer period. We did not check totals, handicaps or betting during the game. "Accurate to within the fee" is not the same as "nobody can win", and nothing here is betting advice.

What we checked

We took every closed Polymarket "who wins" market with $2,000 or more of trading between 29 June and 26 September 2026: 34,065 markets in 17 sports and esports. Draw markets in soccer were left out. The largest groups:

How we measured. For each market we took the price at a set time before the start and compared it with how the game ended. A side that won counts as 100, a side that lost counts as 0, and we subtract the price in cents. A favourite bought at 70¢ that wins gives +30; one that loses gives −70. Averaged over many games, the result is close to zero if the price was right. A positive average means that side won more often than its price said; a negative one, less often.

Why we used real trades. Polymarket keeps the price history of a market for about 30–45 days, so for July and most of August there was no price history left. Instead, we took the price from the real trades in each market: the middle of the last three trades before the time we needed. These are prices people actually paid, not prices on a screen. Prices from real trades were there for 29,191 of 30,141 markets (96.8%).

Ideas first, data second. We wrote down seven ideas, and what would count as a pass, before we looked at the results. We split the period in two: 29 June to 12 August to look for a pattern, and 13 August to 26 September to check it. To pass, an idea had to point the same way in both halves, be clearly above zero in the second half, and be larger than the fee at that price. Two more ideas came out of this first pass; we checked those on April to June 2026, a period we had not used, with real buys and the fee taken out.

What we found

A simple test first: does the favourite's price move in a predictable direction between soon after the market opens and the start of the game? If it does, buying early would beat the closing price. The table shows the average move, in points of the price.

All close to zero. In cricket the move was −6.5 and −5.3, but in the second half the range of error still reached above zero, so we can't tell it apart from chance. MLB had too few markets for this test.

Then favourites priced 60–80¢ two hours before the start. The numbers are "result minus price", in points:

Esports and soccer changed direction between the halves. Tennis favourites did a bit worse than their price in the first half, but in the second half the gap was too small to tell from chance. MLB passed on paper, so we checked it again on April to June with real buys and the fee taken out: −5.45. The direction did not hold.

For scale, the sports fee on one share is 0.05 × price × (1 − price). At 70¢ that is 0.05 × 0.70 × 0.30 = $0.0105, or 1.05 points of the price. Most of the gaps above are that size or smaller, and the larger ones did not keep their direction from one period to the next.

Nine ideas, no advantage

1. Young markets are less accurate. The price soon after opening moves to the start in a direction you can predict. It didn't.
2. Popular teams are priced too high. The teams and players with the most trading lose more often than their price says. The result changed direction between halves.
3. Late moves. When the price moves 5 points or more in the last hour, it has moved too far or not far enough. No pattern.
4. Quiet markets are less accurate. Markets with the least trading are further from the result. Not as we stated it.
5. Favourites at 60–80¢ win more often than the price says. Not in esports, tennis or soccer; MLB is idea 9.
6. Best-of-one against best-of-three in esports. No steady difference.
7. After an upset. A favourite at 70¢ or more loses; its next match is priced wrong. No pattern.
8. Big tennis matches: the outsider wins more often than its price says. It looked real. It wasn't. See the next section.
9. MLB favourites at 60–80¢ win more often than the price says. On April to June: −5.45.

None of the nine ideas left an advantage over the price after the fee.

The trap that looked like an advantage

In the first pass, one group stood out: tennis matches in the busiest quarter of markets, with $184,000 of trading or more. There, favourites won clearly less often than their price said, in both halves. So we wrote a new rule before checking it: in these big matches, buy the outsider. Then we tested the rule on April to June, with real buys made 24 to 2 hours before the start and the fee taken out.

The result: +3.11 points (± 1.22) after the fee, over 1,547 matches, with the outsider at about 34¢ on average. On paper, a pass.

The problem was how we picked the "big" matches. We used the market's total trading at the end, and that total includes trading during the match. When an outsider starts winning, trading jumps. So picking matches by their final trading quietly picks the matches where the outsider was already winning. The rule knew something that nobody could know before the first serve.

Two checks confirmed it:

• By tournament. The gain was not in the star matches: Masters −1.13, Grand Slams +1.78. It was in small tournaments with surprisingly large trading: +4.41.
• By trading before the start, which you can know when you buy. In the busiest quarter by this measure, favourites were not priced too high: −0.16 and +2.91 in the two halves, against −4.08 and −4.51 when we used final trading.

The lesson for anyone testing their own system: pick bets only by what is known at the moment you buy. Final volume, final match length and anything else that is fixed after the start will make a rule look better than it is.

What it means for your bets

If the price before the game is right on average, then two things weigh heavily on your result: the price you bought at and the fee you paid. Buying a few points above the last price before the start, or paying the fee twice by buying and then selling instantly, comes straight out of the result.

The free check above shows the fees from your last 30 days, to the cent. The full report for $19 shows, for bets placed before the start, whether you bought cheaper or dearer than the last price before the game. How we work: method.

Questions

Are Polymarket odds accurate?

On "who wins" sports and esports markets before the start, mostly yes. In 34,065 closed markets from 29 June to 26 September 2026, the price before the game predicted the result about as often as it promised, and the gaps were about the size of the fee or did not repeat. We did not check totals, handicaps or prices during the game.

Can you beat Polymarket sports markets?

We tested nine ideas, written down before we looked at the results, and none gave an advantage over the price after the fee. That does not mean nobody can win: it means the common patterns we checked, by sport, price range, timing, popularity and trading size, were not there. Some people do finish ahead; more on that in do most Polymarket traders lose money.

Are favourites overpriced on Polymarket?

Not in a steady way. For favourites at 60–80¢, the result minus price was −0.14 and +2.86 in esports, −2.74 and −0.83 in tennis, and −1.46 and +0.72 in soccer, in the two halves of our period. MLB was +1.58 and +6.83, then −5.45 on April to June. The direction changes, so there is no rule here to follow.

Is Polymarket more accurate than sportsbooks?

We don't know: we did not compare Polymarket prices with sportsbook odds. What we can compare is the cost. Polymarket charges a fee on each instant buy and sale, and a sportsbook builds its margin into the odds. The numbers side by side are in Polymarket vs sportsbook fees.

How can I check the price I paid?

The full report for $19 goes through your finished bets and shows, for bets placed before the start, whether you bought cheaper or dearer than the last price before the game. To turn a Polymarket price into odds and back, use the odds calculator. The free check above shows your fees to the cent.