Kalshi and Polymarket arrive at the 2026 NFL season with the first genuinely competitive fee fight prediction markets have ever had, and the timing could not be sharper. Kalshi's published taker fee formula runs a 7% coefficient. Polymarket's sports schedule runs a 5% coefficient. Both platforms are federally regulated by the Commodity Futures Trading Commission and are technically legal in all 50 states and Washington, D.C. That combination — a national footprint plus a visible price war — is something the traditional sportsbook industry has not previously had to answer.
The volume is already there. Before a single meaningful game of the 2026 season had been played, Kalshi had recorded over $59 million in trading volume on the Super Bowl LXI winner market alone. That is futures money, parked months ahead of resolution, on a market where the Los Angeles Rams topped the board at 16%. The NFL MVP market shows a similar depth of engagement, with Josh Allen leading at 11% and Joe Burrow, Justin Herbert and Lamar Jackson each sitting at 10%.
Two Different Ways to Charge for a Bet
To understand why the 5% versus 7% headline matters less than it appears, you have to start with the fact that sportsbooks and prediction markets extract revenue in structurally different ways.
A traditional sportsbook is a counterparty. It sets a price, takes the other side of your wager, and builds its margin directly into the odds it publishes. The classic -110 on both sides of a point spread is the textbook example: you risk $110 to win $100 on either outcome, and the two implied probabilities sum to more than 100%. That excess — the overround, or hold — is the book's expected revenue. It is never itemised. You do not see a line on your bet slip that says "fee." The cost is embedded in the number you accepted the moment you clicked.
A prediction market or exchange does not take a position. It matches one participant against another and charges an explicit fee on the transaction. The price itself floats on supply and demand: if the market thinks a team is a 16% shot, that is where the contract trades, and nobody is shading it to build in a margin. The platform's revenue arrives as a separate, disclosed charge on the taker side of the trade.
That distinction is why the fee war between Kalshi and Polymarket is not directly comparable to the vig conversation happening at regulated and offshore sportsbooks. One number is an explicit charge on a trade. The other is an implied cost baked into a price. They are not the same unit of measurement, and treating them as interchangeable is the single most common mistake bettors make when they first move between the two models.
Implied Hold Versus Explicit Taker Fee
Consider a worked example — clearly hypothetical, purely to illustrate the arithmetic, not a quote from any live market.
Suppose a sportsbook offers -110 on both sides of a coin-flip proposition. The implied probability of each side is roughly 52.4%. Together they total about 104.8%, meaning the book has built in an overround of around 4.8%, which on a two-way market translates to a theoretical hold of roughly 4.5% of handle. That cost applies to the full notional amount you staked, and it applies whether you win or lose, because it is embedded in the price you accepted.
Now suppose an exchange lists the same proposition at a true 50/50 and charges a percentage fee on the trade. Whether that fee is more or less expensive than the sportsbook's 4.8% overround depends entirely on what base the fee is applied to, whether it is charged on both sides or only the taker, and whether it is charged at entry, at resolution, or both. A fee coefficient is an input to a formula, not a flat percentage of your stake — and formulas that scale with contract price behave very differently at 50 cents than they do at 5 cents.
This is the crux of it. A 5% coefficient and a 7% coefficient are not two points apart in any meaningful sense until you know what they are multiplying. Longshot contracts priced in the single digits, like most of the field in a Super Bowl outright market, sit at a completely different point on the fee curve than a near coin-flip Week 1 spread. Anyone comparing platforms should be running the actual formula against the actual contract price they intend to trade, not comparing headline coefficients.
Why a Lower Fee Does Not Mean a Better Price
Even a correctly calculated fee comparison tells you only part of what you are paying. The rest is the spread.
On an exchange, the price you can actually transact at is not the midpoint. It is the best available offer on the side you want. If the bid-ask spread on a contract is wide — two, three, four cents — that spread is a real cost, and it is often larger than the fee difference between competing platforms. A market with a 5% fee coefficient and a four-cent spread can easily deliver a worse effective price than a market with a 7% coefficient and a one-cent spread. Fees are visible and easy to compare. Spreads are variable, contract-specific, and much harder to shop, which is precisely why they get overlooked.
The same logic applies to depth. It is one thing for the best offer to be one cent wide; it is another for there to be meaningful size resting at that level. A bettor moving small amounts can take the top of the book and never notice. A bettor moving size will eat through the first level, then the second, then the third, and the average fill price will drift meaningfully away from the quoted top-of-book number. That slippage is not a fee, it is not disclosed anywhere, and for anyone betting real volume it will dwarf the difference between a 5% and a 7% coefficient.
This is why the $59 million already traded on the Super Bowl LXI winner market is a more important number than either fee figure. Liquidity depth is the thing that determines whether a market is genuinely usable at scale. A platform can advertise the cheapest fees in the industry and still be the worse venue if the book is thin when you need it. Conversely, deep liquidity tends to compress spreads on its own, which quietly improves effective pricing for everyone trading that market.
The Regulatory Asymmetry
The structural argument for prediction markets has never really been about price. It is about geography.
A CFTC-regulated national footprint means one account works the same way in every state and in Washington, D.C. There is no state-by-state licensing patchwork, no separate app per jurisdiction, no geofence that stops working when you cross a state line on a road trip. State-licensed sportsbooks operate under a fundamentally different model: each state regulator sets its own rules, its own tax rate, its own permitted market types, and its own approval process. That is why the same operator can offer different markets, different promotions and different limits depending on where a customer happens to be standing.
Bettors who have historically dealt with that fragmentation by using offshore sportsbooks instead are now looking at a third category — one that is federally regulated rather than state-licensed or unlicensed. That is a genuinely new option, and it is the reason the incumbent industry is paying attention to a fee dispute between two platforms that did not meaningfully exist in this form a few years ago.
That said, the legal position is not settled. Prediction markets remain the subject of legal challenges from several US states, and the outcome of those challenges is not something anyone can currently predict with confidence. Readers should check the rules in their own jurisdiction rather than assuming that federal regulation ends the conversation.
The Season That Frames the Argument
The 2026 schedule gives both platforms an unusually strong showcase. The season opens on a Wednesday, for only the second time in 75 years, with a Super Bowl LX rematch between the New England Patriots and the Seattle Seahawks. A midweek opener with a built-in narrative is exactly the kind of event that concentrates liquidity, and concentrated liquidity is where exchange pricing looks its best.
Week 1 also delivers the NFL's first-ever regular season game in Australia, with the San Francisco 49ers facing the Los Angeles Rams on Thursday Night Football. It is one of the tightest matchups on the Week 1 board — and tight matchups are where the fee-versus-vig comparison actually bites, because the closer a market sits to even money, the more the cost structure determines whether a position is viable at all.
Traditional books are not standing still through any of this. Operators like SportsBetting.ag and TigerGaming compete on reduced-juice lines, market breadth and promotional value rather than on an itemised fee, and for a large share of recreational bettors that package still wins. A prediction market cannot offer a parlay boost or a deposit match. It can only offer a price.
What This Means for Bettors
Run the formula, not the headline. A fee coefficient is an input, not a rate. Calculate what the actual charge would be on the specific contract and price you intend to trade before concluding one platform is cheaper.
Price the spread, not just the fee. The gap between bid and ask is a real cost that varies by market and by moment. On thin contracts it will routinely exceed the difference between competing fee schedules.
Size determines which factor matters. If you are betting small, fees dominate your cost. If you are betting size, depth dominates — because slippage through a thin book is unbounded in a way that a disclosed fee is not.
Do not assume the models are interchangeable. Sportsbook hold is embedded and invisible; exchange fees are explicit and disclosed. Comparing a 4.5% theoretical hold to a fee coefficient without normalising the units produces a meaningless answer.
Promotional value still counts. Explicit fees are easier to see, but a sportsbook bonus or reduced-juice offer can change the effective cost comparison in ways a headline fee percentage never captures. Value the whole package.
Cost structure does not replace discipline. Whichever venue prices best, the variable that decides long-run outcomes is staking, and bankroll management does more work than any two-point fee differential ever will.
Check your jurisdiction. Federal regulation and state-level litigation are running in parallel. Verify your own position rather than relying on a general claim of nationwide legality.
The Real Contest
The fee war is the visible part of a larger structural question: whether a market-priced, nationally available venue can take meaningful share from a counterparty-priced, state-licensed one. Kalshi's $59 million in Super Bowl LXI volume before a meaningful snap suggests the demand exists. Whether that translates into competitive pricing at the level where most bettors actually operate — Week 1 spreads, live markets, mid-season totals — is a question the 2026 season will answer far more convincingly than any comparison of published fee schedules.