Sports prediction-market volume on Kalshi and Polymarket has exploded in 2026, but the cannibalisation of regulated US sportsbook handle is real-but-bounded: Bank of America puts prediction markets at roughly 8% of the online sports-wagering market, and the apparent threat is inflated by a volume-versus-handle accounting gap that few headlines explain.
Prediction markets account for about 8% of the regulated US online sports-wagering market, according to Bank of America (BofA) estimates cited in April 2026 — not the wholesale disruption that record volume figures imply. US commercial sportsbooks took roughly $167 billion in handle last year (American Gaming Association, 2025), while Kalshi’s weekly volume recently hit about $2.8 billion, nearly three times its 2024 election peak (Pew Research Center, May 27, 2026). This Deep Dive separates volume from handle, tests the substitution thesis, and maps where it would break.
Key Facts:
• US commercial sportsbook handle reached roughly $167 billion in 2025 — American Gaming Association
• Kalshi weekly volume hit about $2.8 billion in 2026, ~3× its 2024 election peak; sports has been 75–80% of volume — Pew Research Center, May 27, 2026
• Sports was Polymarket’s largest category in Q1 2026 at $10.1 billion in volume — research compiled by The Defiant, 2026
• Polymarket’s CFTC-regulated US unit generated over $700 million in March 2026, up 167% month-on-month — The Defiant, 2026
• BofA pegs prediction markets at ~8% of the US online sports-betting market; Kalshi’s penetration is ~0.6% versus high-single-digit for the books — Bank of America, January–April 2026
• Citizens estimates prediction-market expansion across 50 states has coincided with a ~5% decline in legal sportsbook handle — Jordan Bender, Citizens
• The American Gaming Association claims states and tribes have forgone more than $1 billion in gaming-tax revenue — AGA, 2026
Methodology and its limits
This analysis draws on four source classes: regulator and trade-body handle data (American Gaming Association), prediction-market volume aggregates (Pew Research Center, TRM Labs, The Defiant), sell-side equity research (Bank of America, Jefferies, Citizens), and CFTC registration disclosures for the regulated US venues. The window is mid-2025 through early June 2026, the period in which CFTC-regulated sports event contracts scaled. Two caveats are decisive. First, handle and volume are not the same unit: on Kalshi a 10-cent position contributes a dollar of notional volume, whereas a 10-cent bet at DraftKings is 10 cents of handle — so raw volume systematically overstates prediction markets relative to sportsbooks. Second, no clean out-of-sample causal test exists yet; the substitution estimates are association, not proof, and are revised in near-real time as fresh data lands.
The data: volume up, share contained
The headline numbers are genuinely large. Aggregate prediction-market volume reached roughly $21 billion in a single month in 2026 (TRM Labs, 2026), and sports has been the dominant driver — 75% to 80% of Kalshi’s activity even after the Super Bowl, and Polymarket’s biggest single category at $10.1 billion in Q1. Yet share of the actual wagering economy tells a quieter story. BofA’s estimate that prediction markets sit near 8% of the online sports-betting market, with Kalshi itself around 0.6% penetration, is an order of magnitude below what the volume charts suggest.
The gap is the volume-versus-handle distinction. Because prediction-market volume counts the full notional of each contract rather than the bettor’s stake, a direct comparison of “Kalshi volume” to “sportsbook handle” is apples-to-oranges. Normalised for that, the regulated books still command the overwhelming majority of genuine consumer wagering dollars. The clearest signal of real overlap comes from the books’ own numbers: BofA’s Julie Hoover noted that online sports-betting handle growth went negative “for the first time ever” as Kalshi volume kept climbing — a coincidence in timing that fuels the substitution debate without settling it.
| Venue / metric | Recent volume or handle | Sports share | US-regulated status | Est. market penetration |
|---|---|---|---|---|
| US commercial sportsbooks | ~$167bn handle (2025) | 100% | State-licensed | High-single-digit (baseline) |
| Kalshi | ~$2.8bn weekly volume (2026) | 75–80% | CFTC-regulated | ~0.6% penetration |
| Polymarket (group) | $10.1bn sports volume (Q1 2026) | Largest category | Offshore + CFTC US unit | ~8% of online sports market (sector) |
| Polymarket US (CFTC) | $700m+ (March 2026) | Sports-led | CFTC-regulated | 6.6% of platform activity |
Sources: American Gaming Association (2025 handle); Pew Research Center (May 27, 2026, Kalshi volume); The Defiant research compilation (2026, Polymarket); Bank of America (Jan–Apr 2026, penetration). Time window: mid-2025 to June 2026. Volume and handle are different units and are not directly comparable.
“Every single piece of news that’s come out from the prediction market landscape has had an impact.”
— Shaun Kelley, Managing Director and Senior Research Analyst, Bank of America (NEXT.io)
The mechanism: why overlap is partial, not total
Prediction markets and sportsbooks are structurally different products that happen to converge on sports outcomes. A sportsbook sets a price, takes the other side, and books a margin (the vig). A prediction market is a peer-to-peer exchange: Kalshi and Polymarket match buyers and sellers of binary event contracts and earn fees, not spread. That distinction shapes who switches. Sharp, price-sensitive bettors are drawn to exchange-style pricing and the absence of limits or stake restrictions that books impose on winners. Casual parlay bettors — the books’ most profitable cohort — have far less reason to migrate to a contract exchange that does not offer same-game parlays or boosted odds.
The crypto rail matters here too. Polymarket settles in USDC on a public blockchain, which gives it global reach and frictionless onboarding for crypto-native users but also exposes it to the regulatory perimeter that the CFTC and state gaming regulators are still drawing. That on-chain plumbing is why prediction-market growth shows up in the same dashboards as crypto adoption, and why thefairstake has tracked how prediction-market volume decoupled from Bitcoin’s slide earlier this cycle. The steelman for full cannibalisation is simple: if the CFTC umbrella lets prediction markets offer sports nationwide — including in states with no legal sportsbook — the marginal new bettor may never open a DraftKings account at all. That is a distribution argument, and it is the strongest bear case for the books.
What the comparison misses
The substitution narrative leans on a single suggestive coincidence — handle growth stalling as Kalshi volume rose — and extrapolates it. Citizens’ estimate of a roughly 5% handle decline tied to prediction-market expansion is the most concrete substitution figure on the table, but even that sits inside a wide band of “up to 10%,” and analysts are blunt about the uncertainty. The profitability question is separate from the volume question, and it is where the bull case thins: high notional volume on a fee-only exchange does not translate into sportsbook-style margins. A market can be enormous in turnover and still be a poor business, which is the crux of the contrarian read.
There is a useful prior-cycle analogue. When daily fantasy sports (DFS) scaled in 2015–2016, incumbents and analysts warned it would hollow out traditional wagering; instead, DraftKings and FanDuel converted their DFS user bases into the two largest US sportsbooks once the Supreme Court cleared state-by-state legalisation in 2018. A new wagering format expanded the overall funnel rather than simply transferring it. That precedent argues for coexistence and conversion over zero-sum substitution — but it is not a guarantee. DFS and sportsbooks were ultimately governed by the same state licences and operators; prediction markets answer to a different regulator (the CFTC) and, in Polymarket’s case, run on crypto rails outside the state framework entirely. The structural break from the DFS template is the regulatory arbitrage, which is why this cycle could rhyme rather than repeat.
“While reported prediction trading volumes will likely continue to grow due to the breadth available (entertainment, financials, elections, sports, etc), and we see a global potential TAM of >$100B, this greatly overstates the potential profitability of these markets.”
— David Katz, Analyst, Jefferies (Covers)
Where this thesis breaks
The “real-but-bounded, ~8% and complementary” reading rests on assumptions that could fail. Three conditions would force a rethink toward genuine, large-scale substitution:
- Sustained handle decline at the books that tracks prediction-market growth state-by-state. If multiple mature states report year-on-year handle contraction of more than 10% while Kalshi and Polymarket US volume rises in those same states, the coincidence becomes a pattern and the substitution case strengthens materially.
- CFTC entrenchment plus state-sportsbook stasis. If federal courts affirm that CFTC-regulated event contracts pre-empt state gaming law, prediction markets gain nationwide sports distribution the books lack — particularly in non-legal states — and the addressable-market gap closes from the wrong side for sportsbooks.
- Parlay-style product parity. If exchanges launch correlated multi-leg contracts that replicate same-game parlays, they would attack the books’ highest-margin product directly, and the “casual bettors won’t switch” assumption collapses.
Conversely, the thesis is reinforced if the AGA’s tax-and-licensing campaign forces prediction markets into state frameworks, neutralising their structural distribution edge.
What to watch next
Three data streams will resolve the debate over the next two quarters. First, state-regulator handle reports for the 2026 NFL season opening in September — the first full football slate with CFTC sports contracts live nationwide — will show whether handle growth re-accelerates or keeps stalling. Second, the AGA-versus-Kalshi tax dispute and any CFTC rulemaking on sports event contracts will set the regulatory perimeter; a re-classification either way is binary for the model. Third, the sell-side prints from BofA, Jefferies and Citizens each earnings cycle will recalibrate the penetration estimate as cleaner state-level data arrives. Watch the gap between notional volume and operator revenue — that spread, more than any headline volume record, tells you whether prediction markets are a business or a billboard.
TL;DR
Sports prediction markets are booming in turnover but still small in share. Bank of America pegs them at roughly 8% of the US online sports-betting market, with Kalshi near 0.6% penetration, even as Kalshi clears about $2.8 billion in weekly volume (Pew Research Center, May 2026). The cannibalisation case rests largely on handle growth stalling as volume rose — suggestive, not proven, and inflated by counting notional volume against bettor handle. The thesis breaks if mature states show 10%+ handle declines tracking prediction-market growth, or if CFTC pre-emption hands exchanges nationwide sports distribution.
FAQ
Are prediction markets the same as sports betting?
Not legally or structurally. Sportsbooks set odds, take the opposite side of the bet, and earn the margin; prediction markets such as Kalshi and Polymarket are exchanges that match buyers and sellers of binary event contracts and earn fees. The CFTC regulates the US event-contract venues, while state gaming regulators license sportsbooks — a jurisdictional split at the heart of the current dispute.
How much US sports betting do prediction markets actually capture?
Bank of America estimates prediction markets sit at roughly 8% of the US online sports-wagering market, with Kalshi’s own penetration near 0.6% versus high-single-digit shares for established books. Headline volume looks far larger because prediction-market volume counts full contract notional rather than the bettor’s stake.
Why is volume a misleading comparison?
Because a 10-cent position on Kalshi adds a dollar of notional volume, while a 10-cent bet at a sportsbook is 10 cents of handle. Comparing prediction-market “volume” to sportsbook “handle” therefore overstates the exchanges. Normalised, regulated books still command the large majority of real consumer wagering dollars.
Is there proof that prediction markets are cannibalising sportsbooks?
No clean causal proof yet. Citizens estimates a roughly 5% handle decline tied to prediction-market expansion, and BofA noted handle growth turned negative for the first time as Kalshi volume climbed. But analysts stress the evidence is associative and revised in real time, with cannibalisation estimates ranging up to about 10%.
What role does crypto play?
Polymarket settles trades in USDC on a public blockchain, giving it global reach and easy crypto-native onboarding. That on-chain settlement is why prediction-market growth appears alongside crypto-adoption metrics and why its regulatory status remains contested in the US.
This article is informational analysis only and is not financial, investment, or wagering advice. Cryptocurrencies are highly volatile and can lose substantial value rapidly. Prediction markets and betting markets carry total-loss risk. Past performance and historical correlations do not guarantee future results. Do your own research. If your article touches gambling or betting products: see Responsible Gambling notice in our footer or visit GamCare (UK), 1-800-GAMBLER (US).
Gambling carries financial risk and can be addictive. If you or someone you know needs help, visit GamCare (UK), call 1-800-GAMBLER (US), or see our Responsible Gambling page. 18+ / 21+ depending on jurisdiction.
Further reading: our analysis of Kalshi’s regulatory squeeze at $12.35bn a month, the sportsbook super-PAC fighting state betting curbs, and the NBA Finals odds picture.
