September sportsbook handle was virtually flat year over year across 15 states, even as prediction markets shattered trading records as football returned.
Prediction markets set trading records as football returned in September but sportsbook handle in 15 major states was nearly flat year over year, suggesting traditional operators held steady even as a fast-growing rival expanded.
An analysis of September sports betting results across 15 states shows approximately $11.11 billion in handle, down an estimated 0.3% year over year, even as the return of the NFL and college football seasons brought a substantial rebound from August.
Seven states recorded year-over-year increases, seven reported declines, and Iowa was virtually unchanged. The results suggest that traditional sportsbooks saw substantial wagering activity despite the rapid expansion of prediction markets, although growth remained weak across several major markets.
The relatively flat performance followed a difficult August for sportsbooks and coincided with unprecedented prediction-market trading.
According to TickerTracker, Kalshi generated $60.28 billion in September notional volume, up 56.4% from August. Across four major prediction-market exchanges, trading reached $75.5 billion. Unlike sportsbook handle, prediction-market notional volume measures contracts at their $1 settlement value, not the amount traders paid. The figures are therefore not directly comparable.
The figures raise an important question for the U.S. gambling industry: Are prediction markets already taking significant business away from sportsbooks, or is football generating enough demand to support both?
World Cup Boost Fades Before Football Season
The 2026 FIFA World Cup provided an unusual summer boost for both traditional sportsbooks and prediction markets.
According to the American Gambling Markets Atlas, online sportsbook handle increased approximately 24% year over year in June and 30% in July, reversing several months of declining activity.
Prediction markets also broke records. Trading across Kalshi and Polymarket reached a then-record $50.59 billion in July, according to The Block, before retreating approximately 14.5% in August.
Traditional sportsbooks experienced a similar slowdown. The Atlas recorded a 3% year-over-year decline in online handle during August. The American Gaming Association (AGA) reported a smaller nationwide decline of 0.7%, with sportsbooks accepting $10.89 billion in wagers.
The AGA noted that, excluding World Cup-boosted June and July, monthly sports betting handle had declined year over year since December 2025. The organization cited prediction-market competition and the later start of college football as contributing factors in August.
Against that backdrop, September’s virtually flat handle across the 15 states could indicate stabilization, but it could also represent a continuation of the industry’s broader slowdown.
Sportsbook Revenue Diverges From Handle in September
Like handle, revenue figures were mixed. In several states, revenue declined more sharply than volume, while in others, the opposite was true.
New York generated $159.4 million in September gross gaming revenue, down 17.7% year over year despite a 5% increase in handle. Meanwhile, New Jersey’s sports wagering revenue fell 14.3% to $77 million, while wagering volume was down only 3.6%. States such as Maryland and North Carolina followed similar patterns.
However, the pattern was not universal.
Ohio’s sportsbook revenue increased 12.8% to $87.7 million despite a 2.2% decline in handle. Pennsylvania reported a 50.5% increase in taxable sports wagering revenue to $44.7 million, even as wagering volume fell 6.1%. Massachusetts and Nevada showed similar trends.
The differences highlight the importance of sportsbook hold, the percentage operators retain after paying out winning bets.
For example, New York’s hold fell from approximately 8.47% in September 2025 to 6.64% this year, explaining why revenue declined even as wagering increased. Maryland’s statewide hold similarly dropped from 11.2% to 9.0%. Lower operator margins can result from more favorable outcomes for bettors, even when wagering volume remains stable or increases.
These differences demonstrate why sportsbook revenue declines cannot automatically be attributed to competition from prediction markets. Changes in betting outcomes and operator margins can substantially affect revenue independently of wagering demand.
Prediction Markets Set Records in September
Meanwhile, prediction-market platforms saw record growth in September.
Prediction Frontier’s Q3 report shows that Kalshi, Polymarket, Polymarket US, and Novig generated $179.6 billion in notional trading volume between July and September, an 84.3% increase from the previous quarter.
September alone accounted for $75.5 billion, the largest monthly total in the research firm’s dataset.
Kalshi dominated activity. Prediction Frontier estimated the exchange’s trading volume at $60.7 billion. Meanwhile, TickerTracker estimated approximately $60.3 billion traded on Kalshi in September. The two trackers use separate datasets and report slightly different totals.
TickerTracker’s data shows that $34.98 billion, or 58% of that activity, was combos (similar to parlays at sportsbooks). Meanwhile, straight sports contracts contributed another $15.03 billion, representing 24.9%. Together, combos and straight sports accounted for nearly 83% of Kalshi’s September notional volume, although not every combination contract necessarily involved sports.
Football weekends saw the biggest growth.
Kalshi generated $4.89 billion over September 12-13, the first full NFL weekend, a then-record for two-day trading. TickerTracker recorded $5.66 billion across seven platforms. The following weekend, eight exchanges generated $7.64 billion, with Kalshi accounting for $5.83 billion.
By September 26-27, Kalshi set consecutive single-day records of approximately $3.03 billion and $3.04 billion, respectively. Prediction Frontier separately recorded September 27 as the quarter’s busiest day, with $4.19 billion across its four tracked exchanges.
The momentum continued into October. TickerTracker recorded $95.42 billion in Kalshi notional volume, up by 54.2% from its September estimate. Combos’ share grew to 67.6%, while traditional sports added another 18.1%.
Prediction-Market Volume Is Not Directly Comparable to Sportsbook Handle
The distinction between notional trading volume and the actual cost of contracts matters when assessing the scale of prediction markets relative to traditional sportsbooks.
Sportsbook handle figures measure how much bettors wagered. Meanwhile, prediction market volume figures show contracts traded at their $1 settlement value, regardless of the price paid. For example, a contract purchased for five cents can contribute $1 in notional volume, even though it requires only five cents from the buyer.
Market-making, repeated trading, and increasingly popular low-priced combos can further amplify reported activity.
Prediction Frontier estimated that traders paid $56.1 billion for the $179.6 billion in contracts traded during Q3. Similarly, Kalshi’s approximately $60.7 billion September notional volume represented $14.7 billion in dollars paid, according to the research firm.
The dollars-paid measure provides a different perspective on the scale of prediction-market activity. However, it is still not equivalent to sportsbook handle, and Kalshi’s platform-wide figures cannot be directly compared with sportsbook results from only 15 states.
Macquarie analysts separately estimated that U.S. prediction markets generated approximately $20.1 billion in taker-side volume during September, including roughly $17 billion tied to sports.
Are Prediction Markets Taking Significant Business From Sportsbooks?
The September figures complicate arguments that prediction markets are already causing a widespread contraction in regulated sports betting.
Ahead of the season, the AGA projected $29.5 billion in legal NFL wagering, virtually unchanged from its $29.4 billion estimate for the previous season. The September figures broadly align with the AGA’s expectation of limited growth. However, the AGA figures cover the entire season and only NFL wagering, rather than all sports in the 15-state September sample.
Football’s return brought the expected seasonal rebound from August, but year-over-year sportsbook handle remained virtually flat across the states examined. At the same time, prediction markets generated record activity.
September figures suggest that football is supporting demand across both sectors, but they don’t show whether prediction markets are attracting new users, cannibalizing sportsbooks, or both. A flat sportsbook handle does not necessarily mean prediction markets have no impact.
Some indication could come from third-quarter earnings, particularly from DraftKings and Flutter Entertainment, FanDuel’s parent company. However, both have largely limited their sports event contracts to states where traditional sports betting is unavailable, meaning their results may offer only a partial picture of the competitive impact.
Another important data point will be October’s sports betting figures, once available. A second month of football-season results could provide a clearer picture of sportsbook trends as prediction markets continued expanding in October. However, it would not establish whether customers are shifting between the two sectors.
For now, results point to an established sportsbook industry facing a rapidly expanding competitor, rather than a clear transfer of customers from one market to another.
The post Prediction Markets Exploded, but Sportsbooks Held Their Ground in September appeared first on Gambling Insider.