Exchanges Beat the Sportsbooks on 74% of Moneylines. After Fees, It Is 55% — and Which Exchange Decides Everything

Editor's note: TrueEdge Wire and TrueEdge Odds are run by the same team. Every number in this piece comes from TrueEdge Odds' own price collection, and TrueEdge sells tools that make exactly this comparison. Read it with that in mind; the method and its limits are set out below so you can check it.

The cheapest moneyline in American sports is usually not at a sportsbook. It is on a prediction-market exchange. Whether you actually get that price is decided by something the screen does not show you: the fee.

We took every game-winner market that TrueEdge Odds collected for games that started between 22 and 29 September 2026, across college football, the NFL, the NHL, MLB, the WNBA and MMA. For each of the 509 games we kept each venue's last price before kickoff, looking back no more than 12 hours, and compared the best exchange price with the best sportsbook price for every side where both existed. That gave 579 head-to-head sides.

Before fees, it is not close

On 73.6% of those 579 sides, an exchange — Kalshi, Polymarket US, Novig or ProphetX — showed a better price than the best of the regulated and offshore sportsbooks we track. That is the number the exchanges' marketing leans on, and it is real. An exchange lists a price that another trader has offered, with no house margin built into it, so the listed two-way market is tight: the median Polymarket US game market added up to just 0.5% over fair, and the median Kalshi market to 1.0%.

Set that against the sportsbooks. The median two-way hold at the big retail books sat in a narrow band from 4.34% (theScore) to 4.87% (Bally Bet), with FanDuel at 4.49% and DraftKings and BetMGM at 4.55%. The market-making sharp books were cheaper — Circa at 2.25%, Pinnacle at 2.36% — but nowhere near 0.5%.

After fees, the picture splits in two

An exchange's listed price is not what a taker pays. Polymarket US charges 0.0695 × contracts × price × (1 − price) on every fill that removes liquidity; Kalshi charges a fee of the same shape on a contract's expected earnings, modelled at 0.07 and halved on MLB game markets, where Kalshi's own API sets a 0.5 multiplier. The fee peaks on the even-money contracts that make up most game lines. At 50 cents, 0.0695 of 0.5 × 0.5 is 1.7 cents a contract — about 3.5% of what you stake.

Put the fee back in and the exchange's best-price share falls from 73.6% to 55.4%. It is still better than a coin flip, but a quarter of the exchanges' apparent advantage was fee.

And the average hides a split that matters more than the average. Here is how often each exchange, after its own fee, matched or beat the best sportsbook price on sides where it had a quote:

• Novig: 60.0% (138 of 230 sides). TrueEdge models Novig with no per-trade fee, so its tight listed market survives intact: a median game hold of 0.5% before and after.
• ProphetX: 41.5% (190 of 458). ProphetX's feed prices arrive already net of its commission on winnings, so its 1.61% median hold is the all-in number.
• Polymarket US: 19.1% (86 of 450). A 0.5% listed hold becomes 3.84% once the taker fee is counted.
• Kalshi: 17.6% (79 of 449). A 1.0% listed hold becomes 3.47%.

That is the finding. The two biggest exchanges, the ones with the lowest listed margins, are after fees more expensive than Pinnacle, Circa, Bookmaker and — at 3.84% for Polymarket US — BetOnline, whose median hold was 3.63%. They still beat every retail sportsbook we track, which is why they win so often on a single side. But the "no vig" framing is wrong for anyone who takes the displayed price.

Where the gap is widest, and how big the edge really is

By sport, the exchange-best share after fees was highest in the NFL (22 of 32 sides, 68.8%), the WNBA (59.4%), college football (145 of 246, 58.9%) and MLB (55.7%), and lowest in the NHL (42.5%) and MMA (4 of 15, 26.7%). The NFL, WNBA and MMA samples are small; treat those three as directional.

When an exchange did have the best price after fees, the advantage was usually modest. The median improvement over the best sportsbook was 0.65 percentage points of implied probability; one side in ten beat it by 1.73 points or more. On a −110 favourite that median is the difference between −110 and roughly −107. Worth taking every time, and not worth a round-trip fee on a thin price.

The bigger effect shows up when you shop everything at once. On the 292 games where we had both sides from the sportsbooks and at least one exchange price, the best line available across sportsbooks alone had a median hold of 1.51%. Adding the exchanges, after fees, cut it to 0.52%.

The case for the exchanges, stated properly

The strongest objection to all of this is that we have measured the wrong trader. We priced every exchange fill as a taker: someone who accepts the price on the screen and pays the fee to cross it. A trader who instead rests an order and waits to be filled pays no taker fee, and on Polymarket US collects a 0.0125 maker rebate. For that trader the listed 0.5% market is close to the real one, and the exchanges are the cheapest venue in the country by a distance. The exchanges also do something a sportsbook will not: they do not limit a customer for winning.

That objection is correct, and it is why this piece says "for anyone who takes the displayed price" rather than "for everyone". But a resting order is a different product. It fills only if the market comes to you, which on a game that is about to start often means it does not fill at all, or fills only after the price has moved against you. Most bettors comparing a Kalshi price with a FanDuel price are deciding where to take a number now. For that decision, the after-fee column is the one that is real.

What this data can and cannot tell you

The prices are each venue's last quote before kickoff, not a synchronised snapshot. The median sportsbook quote was 59 minutes old at kickoff and the median exchange quote 40 minutes, because a quote is only recorded when it changes. That is fine for comparing typical pricing and wrong for claiming tradeable arbitrage: on paper, 65 of the 292 games had a combination of prices that summed to less than 100% once exchanges were included, against 36 using sportsbooks alone, but prices recorded up to an hour apart were never available at the same moment, so we are not counting those as arbs.

We also did not account for depth. An exchange's best price may have only a few hundred dollars behind it, and a sportsbook may accept far more at its number. The fee model is TrueEdge's own: Polymarket US at its published 0.0695, Kalshi at 0.07 with the MLB multiplier, ProphetX taken at its commission-adjusted feed price, Novig at no per-trade fee. If a venue's schedule differs from our model on a given market, the net figures move with it.

What would change the conclusion is a fee cut. If Kalshi or Polymarket US lowered the taker coefficient on sports, their after-fee share would climb toward Novig's. Until then, the rule for line shoppers is simple and slightly counterintuitive: the exchange with the tightest-looking market is often not the cheapest place to bet. Compare after fees, every time — the free arbitrage and hedge calculator takes the Kalshi and Polymarket fee off each leg before it shows a result.