← Research
Operator Research Prediction Markets 16 min read • March 2026

FanDuel Predicts: Breaking Down the PM vs. Sportsbook Margin Math

FanDuel Predicts charges a 2% fee but cedes roughly half to CME Group — leaving an effective net margin 6–8× lower than its own sportsbook. Here is the unit economics breakdown operators need before pricing their B2B stack decisions.

By the Metrics
$39.5B
PM volume needed to match FanDuel’s top-5 state net profit
~1%
FanDuel’s effective net margin on PM volume after CME split
7–15×
more volume PM needs per user vs. a traditional sports bet
Problem
FanDuel Predicts charges a 2% fee but cedes ~50% to CME Group, leaving a net margin 6–8× lower than its own sportsbook—yet operators are pouring $200–300M into PM expansion.
Approach
We model the fee structures, revenue splits, and volume thresholds of FanDuel Predicts, Kalshi, and traditional sportsbooks to quantify exactly where the margin math breaks—and what scale is required to close the gap.
📈
Outcome
Operators and their B2B partners must understand the structural economics before pricing AI-driven betslip and CRM tools into a PM-first product stack.
in 𝕏

When FanDuel Predicts launched on December 22, 2025, the coverage focused on the novelty: a major US sportsbook operator entering the CFTC-regulated prediction market space, targeting states where traditional sports betting is illegal. What received less scrutiny was the structural unit economics baked into the CME Group joint venture — economics that make FanDuel’s PM product fundamentally unprofitable at any reasonable near-term scale.

This is not a critique of the strategy. Flutter CEO Peter Jackson has been clear that PM is an addressable market play, not a margin play. But for operators evaluating their own product mix, and for B2B vendors pricing their tools into a PM-first stack, the math deserves rigorous examination. The gap between sportsbook hold and PM net margin is not a rounding error — it is a structural 6–8× difference that reshapes every downstream investment decision.

What FanDuel Predicts Actually Is — and What It Isn’t

FanDuel Predicts is a CFTC-regulated prediction market product built as a joint venture with CME Group, one of the world’s largest derivatives exchange operators. The product launched in five states — Alabama, Alaska, South Carolina, North Dakota, and South Dakota — none of which have legal sportsbooks. That is not a coincidence.

Flutter’s stated rationale is explicit: prediction markets open the door to California, Texas, and Florida, states that collectively represent more than 20% of the US population and have so far blocked traditional sports betting legalization. In a market where FanDuel already commands approximately 40% national share (34% GGR share), organic growth in existing sportsbook states is increasingly constrained. PM is the growth vector, not the margin engine.

The costs confirm this framing. Flutter reported $40–50M in incremental Q4 2025 EBITDA impact from the PM launch alone, then guided $200–300M in additional 2026 EBITDA investment — the majority expected to land in H2 2026 as national rollout accelerates. (Source: Flutter Q4 & Full Year 2025 Results)

That is a $240–350M loss-leading investment funded by sportsbook profits. The strategic logic is defensible. The margin math is not pretty.

Key structural advantage of PM vs. sportsbook: Prediction markets operate under CFTC regulation, not state gaming regulators. This means no state licensing fees, no gaming taxes, and availability in large states (CA, TX, FL) where sportsbooks cannot legally operate. The regulatory arbitrage is real — but it does not fix the revenue-split problem.

The 2% Fee That Isn’t Really 2%

FanDuel Predicts charges a flat 2% fee on potential payout — not on the stake. This distinction matters enormously for understanding the effective take rate across different market types. (Source: DeFiRate: FanDuel Predicts Fee Structure)

Consider a simple example: a trader takes a position on a near-certain outcome, wagering $98 to win $2 (implied probability ~98%). The 2% fee on the $100 potential payout is $2 — representing 100% of the expected profit. At the other extreme, a 50/50 market where $50 is wagered to win $50 produces a $1 fee on $50 of expected profit, or a 2% effective rate on profit.

In practical terms, most actively traded sports markets sit somewhere in the 60/40 to 70/30 range, where the fee structure produces a meaningful but variable drag. On a $100 payout with $25 profit, the $2.50 fee represents 10% of expected profit — equivalent to a 10% effective vig on that specific trade.

Compare this to Kalshi’s fee formula: 0.07 × contracts × price × (1−price). Kalshi concentrates fees on 50/50 markets and reduces them at probability extremes. On a $1,000 position in a 50/50 market, the taker fee is approximately $35. On a high-confidence 90/10 market, the fee compresses significantly — a structurally different risk and incentive profile than FanDuel’s flat percentage-of-payout approach.

Market Type Implied Probability FanDuel 2% Fee (on $100 payout) Effective % of Expected Profit
Near-certain outcome 95% $2.00 ~40%
Moderate favourite 75% $2.00 ~8%
Even money 50% $2.00 ~4%
Moderate underdog 30% $2.00 ~2.9%

This asymmetry penalizes traders seeking value on near-certain outcomes while appearing cheap on even-money markets. For sports bettors used to a uniform sportsbook vig applied via odds shading, the FanDuel fee structure creates a meaningfully different user experience — and a different operator revenue profile depending on which market types users prefer.

Why the Revenue Split Kills FanDuel’s Margin Math

The gross fee mechanics are only half the problem. The CME Group joint venture structure means FanDuel cedes approximately 50% of gross revenue before deducting a single dollar of its own operating costs. (Source: CME Group / FanDuel Predicts launch announcement)

This creates an inverted cost model. In a traditional sportsbook, FanDuel earns the full 7.8% hold on handle, then deducts state taxes, licensing fees, promotional spend, customer acquisition costs, and operating expenses to arrive at net profit. The gross margin is FanDuel’s to work with before costs hit.

In the PM structure: gross revenue is split ~50/50 with CME first. FanDuel then bears 100% of non-exchange costs — marketing, customer acquisition, product development, compliance, and customer support — against only its ~50% share of the gross fee. The effective net margin on PM trading volume is therefore approximately 1% or less.

$39.5B in annual prediction market volume required—at optimistic 40% net margins and 1.5% fee rates—just to replace the net profit FanDuel generates from its top five sportsbook states. (Source: SigmaSquirrel analysis)

The break-even analysis published by SigmaSquirrel makes the scale problem concrete. FanDuel’s top five sportsbook states — New York, Illinois, New Jersey, Ohio, and Pennsylvania — produced $31.97B in combined handle, $2.05B in GGR, and $241.5M in estimated net profit. To replace that net profit through PM alone, at optimistic assumptions of 40% net margins and a 1.5% effective fee rate, requires $39.5B in annual prediction market volume. (Source: SigmaSquirrel analysis)

For context: Kalshi — the current PM volume leader — did $22.88B in total volume across all of 2025. FanDuel would need to generate $39.5B in PM volume annually just to hold ground on the net profit it is already generating from five sportsbook states today.

FanDuel Sportsbook Hold
7.8%
Top-5 states: $31.97B handle → $2.05B GGR → $241.5M net profit
FanDuel PM Net Margin
~1%
After CME ∼50% revenue split, before any operating costs are deducted
The Gap
6–8×
More net revenue per dollar wagered from sportsbook vs. PM at current fee-split structure

Exchange Ownership Changes Everything: Kalshi’s Structural Advantage

The contrast with Kalshi illustrates why exchange ownership is the decisive structural variable in PM economics. Kalshi owns its exchange infrastructure outright, retaining approximately 100% of its ~1.15% effective fee rate. In 2025, that translated to $263.5M in fee revenue on $22.88B in total volume, with sports representing 89% of that handle. (Source: Kalshi 2025 annual disclosures)

Where FanDuel earns roughly $0.50 of every gross dollar before costs, Kalshi keeps the full dollar. At comparable gross fee rates, this structural difference translates directly to bottom-line profitability. Kalshi’s annualized sports revenue is already approximately $1.3B — roughly 25% of DraftKings’ total sportsbook revenue — despite being a far more recent entrant to sports prediction markets.

The user growth trajectory reinforces that PM demand is structural, not FanDuel-driven. Kalshi monthly active users grew from 600,000 at the start of 2025 to 5.1 million by February 2026 — a 750% surge — the vast majority of which predates FanDuel Predicts’ December launch.

750% growth in Kalshi monthly active users—from 600K to 5.1M—in just 14 months, demonstrating that prediction market demand is structural and not dependent on FanDuel’s entry.

Robinhood adds a third fee architecture worth noting. Robinhood traded 8.5 billion event contracts in Q4 2025 alone, monetizing at a flat $0.01 per contract — a volume-driven model completely different from FanDuel’s percentage-of-payout approach or Kalshi’s price-sensitive formula. At 8.5B contracts, $0.01/contract generates $85M in a single quarter. The model scales without the margin erosion that affects percentage-based structures at high implied-probability markets.

Operator Fee Model Exchange Ownership 2025 Volume Effective Net Rate
Kalshi 0.07 × contracts × p × (1−p) 100% owned $22.88B ~1.15%
FanDuel Predicts 2% of potential payout JV with CME (~50% split) Launch Dec 2025 ~1% or less
Robinhood (via Kalshi) $0.01 flat per contract Distributor (no exchange) 8.5B contracts Q4 2025 Volume-dependent
FanDuel Sportsbook Odds shading (vig) N/A (books risk directly) $31.97B (top-5 states) 7.8% hold

The exchange-ownership vs. JV distinction is not a minor operational detail — it is the entire difference between a structurally profitable PM business (Kalshi) and a market-access play that requires billions in volume to justify its cost base (FanDuel Predicts at current fee-split terms).

What FanDuel Is Actually Giving Up: The Sportsbook Margin Benchmark

To fully appreciate the PM margin gap, the sportsbook baseline needs to be stated plainly. US sportsbook total annual revenue reached a record $16B in 2025 — up $3B versus 2024 — with a national hold rate of 10.16%, the highest annual win rate ever recorded. Total trailing-twelve-month handle reached $117.9B, producing $10.8B in GGR at a 9.14% hold. (Source: CasinoReports US Sports Betting Database)

FanDuel’s 7.8% hold rate — below the national average but above DraftKings’ ~6.4% as of Q1 2025 — is driven by high parlay penetration. Parlays carry structural hold advantages: a two-leg parlay at standard -110 lines carries roughly 10% expected hold, versus ~4.5% on a single-game moneyline. FanDuel’s product investment in parlay builder tools is a direct hold-rate optimization play.

The per-user economics tell an equally important story. DraftKings ARPMUP (Average Revenue per Monthly Unique Payer) reached $151 in Q2 2025 (up 29% YoY) and $139 in Q4 2025 (up 43% YoY). These figures represent the revenue ceiling per active sportsbook user that PM has not yet approached. At PM’s effective net rate of ~1%, generating $139 in net revenue per user requires $13,900 in trading volume per monthly active user — a level that remains far beyond current PM ARPMUP benchmarks.

The fundamental ratio: PM requires 7–15× more wagering volume per user to generate equivalent net revenue to a traditional sports bet. This is the core challenge for any operator attempting to run PM as a margin-equivalent product line alongside sportsbook. It does not mean PM is unviable — it means the per-user economics require either massive volume or a fee structure evolution before PM can fund the same CRM, personalization, and retention investment that sportsbook revenue currently supports.

What This Means for B2B Stack Decisions in a Hybrid PM/Sportsbook World

Operators running both sportsbook and PM products face a two-speed economics problem. Sportsbook users generate $7–9 of net revenue per $100 wagered. PM users currently generate roughly $1. Every dollar of CRM spend, every AI tool license, every data platform cost must be evaluated against those fundamentally different revenue-per-user baselines.

For reactivation campaigns, the math is stark. A churned sportsbook player worth $7.80 in hold per $100 wagered supports a meaningful CRM investment to win them back. A churned PM user generating ~$1 per $100 wagered requires a very different reactivation economics model — lower offer cost thresholds, lower acceptable cost-per-reactivation, and a longer breakeven timeline per user.

Betslip AI illustrates the asymmetry most clearly. AI tools that optimize parlay construction, improve odds presentation, and drive parlay penetration have their highest ROI in the sportsbook context. A 0.1% improvement in hold rate on $117.9B in annual US handle generates approximately $118M in additional GGR industry-wide. Applied to FanDuel’s $31.97B top-five-state handle alone, a 0.1% hold improvement is worth $32M annually — from a single product optimization. No comparable leverage exists in PM at current volume levels.

The B2B investment priority: Until PM ARPMUP approaches sportsbook levels, AI betslip tools and personalized CRM that drive parlay uptake and hold rate improvements generate 6–8× the revenue impact per dollar of PM volume. The B2B iGaming platform and sportsbook software market is valued at $15.4B in 2025, growing at 14.22% CAGR through 2033. Operators who optimize sportsbook yield now — while PM matures — will be best positioned when PM economics eventually close the gap. (Source: B2B iGaming market sizing, 2025)

The break-even horizon for FanDuel PM is not before 2028 per analyst consensus. That creates a multi-year window where sportsbook yield optimization remains the highest-leverage investment for operators on both sides of the B2B relationship. Segmented tooling — separate pricing models, separate ROI frameworks, separate CRM logic for PM vs. sportsbook users — is not optional. It is the correct architecture for a hybrid product environment where the two products have structurally different margin profiles.

PM’s regulatory arbitrage in CA, TX, and FL is real and strategically valuable. But PM user LTV at current fee rates does not support the same per-user CRM investment as sportsbook. Operators who treat PM users identically to sportsbook users in their retention spend will systematically over-invest in PM reactivation and under-invest in sportsbook yield optimization — the exact wrong allocation given the current margin structure.

The Right Frame: PM Is Market Expansion, Not Margin Replacement

FanDuel’s PM investment is rational as a land-grab for large currently-unregulated states. The regulatory arbitrage is real. The user base expansion into CA, TX, and FL is a genuine strategic opportunity that FanDuel’s 17M+ existing users make more viable than for any competitor starting from zero. The $200–300M 2026 investment is funded by the $241.5M net profit FanDuel already generates from five sportsbook states alone — this is not financial recklessness, it is capital redeployment from a high-margin core into an addressable market expansion.

But the framing matters. FanDuel’s JV model trades margin for speed-to-market and CME’s regulatory credibility. Kalshi’s exchange-ownership model is structurally superior for PM economics at scale. The gap between ~1% net PM margin and 7.8% sportsbook hold will likely narrow as PM volume scales and as fee-split terms evolve over JV renegotiation cycles — but that convergence is measured in years, not quarters.

For operators and their B2B vendors, the implication is clear: optimize sportsbook yield now while PM matures. AI betslip tools that drive parlay uptake and hold rate improvements generate 6–8× the revenue impact per dollar of PM volume at current fee rates. Sportsbook CRM tools that maximize reactivation and player LTV have a structurally higher ceiling than PM-equivalent tools until the margin gap narrows.

The prediction market opportunity is real, large, and growing. Kalshi’s 750% MAU growth and $22.88B in 2025 volume make that undeniable. But understanding the margin math — who owns the exchange, who splits the revenue, what volume is required to break even — is the prerequisite for any rational product stack or B2B pricing decision in a hybrid PM/sportsbook world.

Data Sources & References

Related Articles

Want to Model Hold Rate Improvements for Your Sportsbook?

BidCanvas AI Bet Slips drives parlay uptake and hold rate optimization — the highest-leverage investment while prediction market economics mature.

Request Demo See AI Bet Slips