Big, bold numbers command attention. Dr. Evil understood this (even if he was a bit disoriented at first and out of step with inflation when first unfrozen in the 1990s). And the American Gaming Association (AGA) understands this.
That’s part of why AGA President and CEO Bill Miller appeared on CNBC’s Squawk Box when he did a little under two weeks ago, on the occasion of the AGA’s running counter of “State Gaming Tax Dollars Lost Since Prediction Markets Began Offering Sports Event Contracts” crossing … (cue Dr. Evil voice) … $1 billion.
State and tribal governments, Miller said on CNBC, “are losing literally a billion dollars today in … revenue that would otherwise go to fund important community projects and pay taxes to these states.”
The number turned heads and spurred headlines, as it was intended to. But it also raised skeptical eyebrows among some in the gambling community, whether they side with prediction markets in their fight to offer sports betting under the federal oversight of the Commodity Futures Trading Commission (CFTC), or they’re neutral on the issue but just don’t care for the methodology behind the AGA’s big, bold, billion-dollar number.
Back when the AGA was reporting in April that its counter of sports betting tax dollars lost to prediction markets was over $700 million, Alex Kane, the founder and CEO of exchange operator Sporttrade, responded on social media site X: “I’d love to see the math on how this number is constructed. For the sake of the AGA I really hope it’s not: Kalshi daily sports volume x avg Sportsbook hold of 7% x avg state ggr tax rate of 20%.”
And in response to that, Wake Forest Economics Professor Koleman Strumpf expressed his belief that this was indeed how the AGA’s estimate was calculated:
AGA Vice President of Research David Forman, however, shined a degree of light on the mathematical formula in his reply to Kane, writing at the time, “It assumes handle [equivalent] is 50% of PM volume and that kalshi is 55% of the US market this year.”
In contrast to that relatively cordial back-and-forth, there’s this statement that Kalshi spokesperson Elisabeth Diana gave to CNBC regarding the AGA’s billion-dollar estimate: “This is fake math from casinos, who are worried about losing their monopoly power.”
“Fake math” is awfully bold wording — perhaps just as bold as tossing the “1 billion” number out there.
But it you want to call it “disputed math,” that seems a fair assessment.
AGA: Numbers are not static
Forman told InGame last week that the “50%” and “55%” numbers he cited in April continue to be adjusted and that the formula is perpetually evolving.
“The billion-dollar tax loss is based on the public data that Kalshi puts out there, and then we make assumptions around their share of the U.S. market and other estimates around how prediction market trading volume might compare to legal sportsbooks’ handle,” Forman said. “We’ve updated it over time as we’ve seen other research come out in terms of Kalshi’s share of the U.S. market and what the handle equivalent might be.
“Right now, we estimate 65% for Kalshi’s share of the U.S. market — which is in line with some of the other estimates we see out there — and we assume that the equivalent handle is about 30% of the sports betting volume that they report, which we’ve been lowering over time. I know some people think it’s as low as 12%, but I would call that into question.”
Kane, while expressing respect for Miller and the rest of the AGA as professionals, feels they are engineering numbers that suit their objectives.
“Your job there at the AGA is to produce the highest, scariest number possible. Your job, representing the state-regulated operators, is to evoke fear, evoke action,” Kane told InGame. “So either they’re trying to just come up with the highest number possible, or they really don’t understand sportsbook handle is not exchange volume.”
There are several components to Kane’s argument, but the most significant piece is that he thinks it’s a mistake for the AGA to assume there’s heavy overlap between prediction market customers and sportsbook users.
Kane uses himself as an example. Whether he’s a “sharp” or not, he’s at least a price-sensitive sports bettor who looks for edges, and he’d already mostly stopped using state-regulated mobile sportsbooks that charge high vigs and/or limit successful bettors.
“The core predisposition of the AGA argument,” Kane said, “is that because it’s all sports betting, it’s all completely a substitute product and interchangeable — and for that reason, if we just got rid of Kalshi, look at how much more tax money we’d make. Just removing Kalshi doesn’t somehow bring invisible tax revenue back, but the AGA’s assumption is that it does, because it creates the highest number.
“But for me, why would I go somewhere that I can buy a side at 59 cents or sell it at 54 cents, when I can cut that 5% spread in half and bid 57 cents? The customers who are betting lots of parlays in the sportsbook model are not the same people who are using prediction markets.”
A theory on handle and revenue
The AGA notes that betting handle numbers have declined at regulated sportsbooks since prediction market sports contracts have risen to the fore, and the monthly reports from individual states bear that out. In New Jersey in February, betting handle was down 14.4% compared to the prior year. In March, New Jersey handle dropped 8.6%. In New York in March, handle dipped 4.6%. In Ohio in January, the decrease measured 8%.
Numbers like this don’t apply to every state in every month in 2026, but the trend is coming through. The AGA noted in late May that, nationally, “March sports betting handle fell 2.6% and, through Q1, sports betting handle is down 0.8%. This marks the first quarter year-over-year handle decline since June 2020.”
However, there’s a second trend, one the AGA acknowledges in the next sentence: “Sports betting revenue rose 8.9% in Q1, driven by favorable hold comps.”
Looking back at those state-by-stand handle numbers cited above, the 14.4% handle drop in New Jersey in February was accompanied by a less steep 10.3% revenue decline; the 8.6% revenue dip in March in Jersey came with a 22.8% increase in revenue; New York’s 4.6% handle decrease in March ran alongside a revenue spike up 34.3%; and when handle went down 8% in Ohio in January, revenue rose 22.9%.
Some of that has to do with the short-term ups and downs of hold percentages, such as March Madness results going far better for the house in 2026 than they did in 2025. But Kane has another theory: that handle may be dropping at sportsbooks in part because sharp bettors have migrated to prediction markets, and that sportsbooks have benefited from removing some of those customers from their ecosystem and are enjoying higher hold rates and increased revenue as a result.
“If you ask a sportsbook operator, ‘Would you rather live in the March 2026 world or the March 2025 world?’ they would tell you, ‘I want to live in the March 2026 world,’” Kane asserted. “They’d say, ‘I don’t have to spend nearly the same resources trying to find people multi-accounting and getting in and trying to evade all my limits. I’m paying less federal excise tax. And I don’t have to contend with all the sharp flow.’”
Forman disagrees with Kane about the degree of customer overlap, and he points to the handle decline as evidence of the AGA’s position.
“I think the idea that the people betting on sports on prediction markets are betting there because they like the plumbing behind it, rather than the plumbing behind the sportsbook, is frankly ridiculous,” Forman opined. “People are betting on sports in both spots and they’re obviously the same consumer, which we see in the handle numbers.
“Different sportsbooks operate different models when it comes to who they tailor their book to. Some prefer not to deal with sharp bettors, and some welcome sharp bettors. So the idea that the only place that certain bettors could go to get their sports bets down is on these prediction markets, and that’s leading to a decline in national handle, is a little far-fetched.”
Overall, it seems the philosophical position is of greater importance to Forman than is the precise math, as he said, “Ultimately, it doesn’t matter if the number is 700 million or 3 billion. What matters is prediction markets don’t pay any taxes, so they’re not helping to support the critical infrastructure and the local communities that the gaming industry does. And the idea that sports betting on prediction markets is somehow not sports betting is not believable.”
Again, there’s a perfectly civil disagreement taking place between the Millers and Formans of the world and the Kanes and Strumpfs.
But Forman is a bit more aggressive in his comments directed toward Kalshi.
“People can disagree with the assumptions that we use,” Forman said. “But we’re happy to talk about them on the record and be transparent — whereas a lot of the prediction market companies like to call things fake without providing any actual truth.”
