Close the Event-Contract Gap

Say in federal law that a contract paying out on the result of a game is a bet, and let the states regulate it as one.

AI evaluation · not yet reviewed by a human

This evaluation was produced and sourced by an AI model; a human review is still pending. Figures and conclusions may still change. The review log is at the foot of the page.How review works →

Federally licensed exchanges now sell contracts that pay one dollar if a named team wins and nothing if it loses. Because they are filed as derivatives under the Commodity Exchange Act, they fall outside state gambling law: no state licence, no state tax, an age limit of 18 rather than 21, and no connection to the register a person signs when they ban themselves from every sportsbook in their state. Thirteen states are in litigation over it and the federal courts have mostly sided with the exchanges. This measure would amend the Act so that contracts on sporting events are treated as wagers under the law of the state where the customer sits, leaving contracts on elections, economic data and other public outcomes where they are. This evaluation looks five years ahead.

Balance

Better for the future · 0.69 previous scale

Balance on the previous scale. The Bilanz 2.0 simulation is not yet available for this evaluation. The category comes from the share of the debate on the pro side (r).

For 50 · 69 % Against 22 · 31 %

Size class: medium Scale of this evaluation: Normalised Impact — unitless, calibrated to this topic. For comparison: one point here is worth roughly 200 million euro per year. The first version of this evaluation assumed an exchange was cheaper for the person betting, because its stated fee is a fraction of a sportsbook's margin. Transaction-level data on 313,972 Kalshi contracts show the opposite: the average return is about minus 20 percent and minus 31 percent for those accepting quoted prices. The one thing that would restore the old answer is if switchers moved to parlay bets, where sportsbook margins reach 25 to 30 percent. How we score →

Arguments for

Arguments against

9 arguments evaluated · Scoring v1.3 Δ absolute +28

Arguments — For

5 arguments · top 3 shown

People lose less at a sportsbook

25of 100

The stated fee on an exchange is a fraction of a sportsbook's margin, which is the industry's central argument for being left alone. What people actually lose is not the fee. Across 313,972 contracts the average return was about minus 20 percent, and minus 31 percent for those who accepted a quoted price rather than posting one.

Value 5 · Household budgetsImpact 10Plausibility 5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money staying with the person who staked it rather than moving to whoever was on the other side of the trade. It is priced at the middle of the scale like any other money. What makes it a gain rather than a wash is where the two ends sit: the people accepting quoted prices on a sports contract are retail customers, and the people posting them are professional market makers and the platform. That gap is counted in the Impact. What the customer loses along with the money — the ability to sell out of a position before an event ends, which a sportsbook does not offer — is real and is counted as its own argument against this measure. The value is the middle of the scale, because the stream is a transfer of money and the distance between its two ends is priced in the Impact.

About 24.1 billion euro a year is staked on sports event contracts, of which 65 percent is assumed to reach a licensed sportsbook instead, in a range from 40 to 85 percent — 15.7 billion euro. What that money loses differs sharply by venue. American sportsbooks kept 10.2 percent of everything staked in 2025, which is measured and published monthly ^[2]. On the exchange, transaction-level data covering 313,972 contracts put the average return at about minus 20 percent, and minus 31.46 percent for the side accepting quoted prices against minus 9.64 percent for the side posting them ^[6]. A dollar-weighted retail loss rate of 20 percent is used here, in a range from 10 to 31 percent, because the published figure weights contracts equally and the cheapest contracts carry the smallest stakes. The difference of about 9.8 points on 15.7 billion euro is 1.54 billion euro a year that stays with the customers, weighted at 1.3 for the incomes concerned. The Impact is the largest in this debate and it runs opposite to the direction the industry's own fee comparison suggests.

▸ Show calculation ▾ Hide calculation

	Staked on sports event contracts a year Setting, range 12 to 60 billion dollars a year [1]	1.7 billion dollars in one month on one exchange, four fifths of it sport	24.1 billion euro
×	Share reaching a licensed sportsbook instead Setting, range 40 to 85 percent: some customers stop, some go offshore	65 %	15.7 billion euro
×	Difference in what is lost per euro staked Setting on the exchange rate, range 10 to 31 percent: the measured average return across 313,972 contracts is about minus 20 percent and minus 31.46 percent for those accepting quoted prices; the sportsbook figure is 16.96 billion dollars kept from 166.94 billion staked [2][6]	20 % on the exchange against 10.2 % at a sportsbook	1.54 billion euro
×	Weight of a euro at these incomes the customers accepting quoted prices are retail; those posting them are professional market makers	1.3	2 billion euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	10

Score 10 Impact × 5 Value × 5 Plausibility ÷ 10 = 25 of 100

Both loss rates are measured rather than modelled, and the step between them is not. The sportsbook figure is a regulatory return: 16.96 billion dollars kept from 166.94 billion staked in 2025, filed monthly by every licensing state ^[2]. The exchange figure comes from transaction-level data supplied by the platform itself, covering 313,972 contract prices, with the maker and taker side identified on every trade ^[6]. Neither is an estimate. What has no design behind it is the counterfactual this argument needs: whether the same person, moved to a sportsbook, would lose 10.2 percent. Two thirds of exchange contracts are priced below 10 cents or above 90, and contracts under 10 cents lose more than 60 percent on average — which is the signature of longshot betting. A customer with that appetite does not buy the average sportsbook product; they buy parlays, where the margin is 25 to 30 percent rather than 10. If most switchers behave that way, this argument is worth nothing. That is the single largest uncertainty in the evaluation and it is why the plausibility sits at the middle rather than higher. The Plausibility is at the middle: both figures are measured exactly, and nothing establishes that the same person carries the same behaviour across the two venues.

evidence basis: Converging studies · P ceiling 7 identification: Controlled · rung ceiling 7

Counterfactual: the same customers staking the same money at a licensed sportsbook — not observed; no design links the two populations. Design: controlled — two administrative measurements compared, one a regulatory filing [2], one transaction-level platform data on 313,972 contracts [6], with no exogenous variation between them. Confounder: selection by betting appetite, since the favourite–longshot pattern in the exchange data suggests switchers would buy parlays, where sportsbook margins are 25 to 30 percent rather than 10.2; named, unresolved, and capable of removing the argument entirely. Direction: no reverse causation; both figures are outcomes of completed trades. Ceiling: controlled 7.0 binds below the converging-studies ceiling of 9.0; a context transfer of 2.0 applies for the behavioural gap between venues. The equal-weighting of the published return is handled in the 10 to 31 percent band rather than in P.

The protections that come with a licence

9.9of 100

A state gambling licence carries a package: an exclusion register a person can put themselves on, deposit caps, mandatory funding for treatment, and rules on how the product may be sold. A federal derivatives licence carries none of it. The same wager, placed two ways, comes with two entirely different sets of safeguards.

Value 9 · HealthImpact 2.8Plausibility 4

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is mental health: gambling disorder and the depression, anxiety and suicide risk that travel with it. This site places that in the class it uses for life and health, one step below the top of it because the condition is one most people recover from. What is priced here is the person's own state, not the money they lose, which is counted separately, and not the harm to their household, which nothing available measures well enough to price. That the people concerned chose to open the account does not lower the weight — a disorder is a disorder however it began. The value sits one step below the maximum: the stream is health, and health that can be regained.

One exchange alone traded 1.7 billion dollars of notional volume in June 2026, roughly four fifths of it on sport ^[1]. Across the sector something like 24.1 billion euro a year in sports contracts is a reasonable estimate. At an average annual turnover of about 8,000 euro per active account that is roughly 3 million accounts. Among online sports bettors, 19 percent report problematic gambling behaviour repeatedly, more than twice the rate among gamblers generally, and there is no reason the figure differs on an exchange ^[3]. That gives about 570,000 people. What the licensing package reaches — an exclusion register, deposit caps, an age limit of 21, treatment funded from the tax — is set at 12 percent of them, in a range from 4 to 25 percent: 68,400 people. Each carries a loss of 0.2 quality-adjusted years a year. People drawn in by advertising that no rule currently touches are counted separately below. The Impact is the second largest gain and the heaviest in weight, because the group it reaches is narrow and what happens to them is not.

▸ Show calculation ▾ Hide calculation

	Sports event contract volume across the sector Setting, range 12 to 60 billion dollars a year [1]		24.1 billion euro
÷	Active accounts Setting, range 4,000 to 20,000 euro	8,000 euro of turnover each a year	3 million accounts
×	Showing problematic gambling behaviour repeatedly the rate measured among online sports bettors, more than twice that among gamblers generally [3]	19 %	570,000 people
×	Reached by the licensing protections Setting, range 4 to 25 percent: an exclusion register, deposit caps, an age limit of 21, and treatment funded from the tax [4]	12 %	68,400 people
×	Quality-adjusted years lost per person a year	0.2	13,680 quality-adjusted years
×	Value of the years the value of a healthy life year used across this site	40,000 euro each	547 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	2.74

Score 2.74 Impact × 9 Value × 4 Plausibility ÷ 10 = 9.9 of 100

Every link here is visible and none of the numbers has been measured. The counterfactual is the same customers on a federally regulated exchange with no state protections. That exclusion registers and deposit caps reach problem gamblers is supported by the limit-setting evidence from Europe, though from before-and-after comparisons rather than controlled ones, and none of it was done on an exchange ^[4]. That the rate of problem gambling among exchange customers matches that among sportsbook customers is an assumption, and the loss data now available cut against the benign reading of it: a venue where two thirds of contracts are priced below 10 cents or above 90 is a longshot market, and longshot betting is the pattern most associated with chasing ^[6]. The confounder that matters is selection — the people who seek out an exchange are not a random sample of bettors — and nothing addresses it. The counter-mechanism, that a determined customer moves to an offshore market with no protections at all, is booked as its own argument against this measure. The Plausibility is below the middle: the protections plainly do something, and how much of it reaches this particular population has never been looked at.

evidence basis: Mechanism · P ceiling 5.5 identification: Associational · rung ceiling 5.5 band: Chain closed, unevidenced · P 4–5

Counterfactual: the same customers on a federally regulated exchange without state protections. Design: associational for the harm rate — survey prevalence among online bettors with no counterfactual [3]; mechanistic for the policy link, supported only by European limit-setting studies done elsewhere [4]. Confounder: selection into exchanges by a different kind of bettor, unaddressed; the favourite–longshot pattern in the transaction data [6] suggests the selection runs toward, not away from, chasing behaviour. Direction: reverse causation is unresolved for the prevalence figure. Ceiling: associational 5.5 binds. Band: chain closed but unevidenced — every link named, the offshore counter-mechanism booked as con-2; only the measurement is missing.

Nothing measured argues against the claim; what is missing is any study of problem gambling on prediction markets specifically. The counter-mechanism, customers moving offshore, is booked separately as con-2. Read back: about half the time, state licensing protections reach roughly the share of affected customers assumed here.

Open: State exclusion registers record who has self-banned. Matching those lists against exchange account data — which the exchanges hold — would show directly how many excluded gamblers are trading sports contracts, and could carry P to 6.

Eighteen-year-olds lose access

6.6of 100

Every state that licenses sports betting sets the age at 21. A federally regulated exchange sets it at 18. Three years is a long time at that age, and the age at which someone starts gambling is among the strongest predictors of whether they develop a problem with it.

Value 9 · HealthImpact 2.1Plausibility 3.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is the mental health of people in late adolescence, when the impulse control that keeps most adults out of trouble with gambling is not yet fully in place. It sits in the class this site uses for life and health, one step below the top of it because what is lost can be regained. What is priced is the condition itself rather than the money lost, which for this group is small in absolute terms and large relative to what they have. Nothing here treats an eighteen-year-old as incapable of choosing; what is counted is only that starting earlier makes a problem more likely, which is a finding rather than a judgment. The value sits one step below the maximum: the stream is health, and health that can be regained.

About 13 million Americans are aged 18, 19 or 20. If 8 percent of them would use a prediction market where a sportsbook is closed to them, in a range from 2 to 20 percent, that is roughly 1.04 million young people gaining access to sports gambling three years early. Not all of them are harmed and most are not: an average loss of 0.01 quality-adjusted years a year across the whole group is used, in a range from 0.003 to 0.03, which amounts to assuming that something like one in twenty carries a real cost and the rest carry none. That gives 10,400 quality-adjusted years a year. The figure is deliberately built as an average across everyone rather than as a rate among the affected, because nobody knows which of them are affected. The Impact is comparable to the licensing protections above and reaches a different group, which is why the two are counted separately.

▸ Show calculation ▾ Hide calculation

	Americans aged 18, 19 or 20		13 million people
×	Share using a prediction market because a sportsbook is closed to them Setting, range 2 to 20 percent: every licensing state sets the age at 21, the exchanges at 18 [1]	8 %	1.04 million people
×	Quality-adjusted years lost, averaged across all of them Setting, range 0.003 to 0.03: equivalent to assuming about one in twenty carries a real cost and the rest none [3]	0.01 a year	10,400 quality-adjusted years
×	Value of the years the value of a healthy life year used across this site	40,000 euro each	416 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	2.08

Score 2.08 Impact × 9 Value × 3.5 Plausibility ÷ 10 = 6.6 of 100

That the age of first gambling predicts later problems is well documented and the mechanism is not disputed. What is unmeasured is everything specific to this case. The counterfactual is the same cohort with the exchanges closed to them, and no source counts how many under-21s hold exchange accounts, because the exchanges do not publish it and no regulator collects it. The chain has three links — the age limit differs, young people use the lower one, using it earlier raises later risk — and the middle one has no number attached. The counter-mechanism is real and unanswered: an eighteen-year-old shut out of the exchange has an offshore site and a friend's account available, so the exclusion may be nominal. The confounder in the underlying literature is the familiar one, that people who start gambling young differ in ways that also predict problems later, and the studies establishing the link are observational. Reverse causation does not arise for the policy itself. The Plausibility is low because the number of young people this actually concerns has never been counted.

evidence basis: Mechanism · P ceiling 5.5 identification: Associational · rung ceiling 5.5 band: Chain open · P 3–3.5

Counterfactual: the same cohort with exchanges closed to them; nobody counts under-21 exchange accounts. Design: associational — the age-of-onset link rests on observational studies without exogenous variation; the take-up link has no source at all. Confounder: young starters differing in ways that also predict later problems, unaddressed by the underlying literature. Direction: no reverse causation for the policy. Ceiling: associational 5.5 binds. Band: chain open, because the take-up share carries the whole quantity and the counter-mechanism — offshore sites and borrowed accounts — is unanswered.

The chain is named but the link carrying the quantity — how many 18-to-20-year-olds actually use exchanges because sportsbooks are closed to them — has no source, and the counter-mechanism that they can reach offshore sites anyway is unanswered. Read back: about a third of the time, roughly the assumed share of that cohort gains three years of early access.

Open: Exchanges hold the age distribution of their accounts and file reports with the federal regulator. Publishing it by state, against the state age limit, would settle the take-up link outright.

States collect on it

4.8of 100

A sportsbook hands its state between a tenth and half of what it keeps. An exchange selling the identical wager hands it nothing, because a derivative is not a gambling product. Moving the volume into the licensed channel moves the tax with it.

Value 5 · Public financesImpact 1.8Plausibility 5.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is state and local revenue, priced at the middle of the scale like any other public money. What is counted here is only the receiving end of a transfer whose paying end is the customers' loss at the sportsbook, which is already inside the first argument as the smaller of two loss rates rather than as a separate cost. Booking the whole sum on both sides would say something false about what the measure does. Several states earmark this money for problem gambling treatment, which is not counted separately because an earmark is a budgeting choice rather than a second good. The value is the middle of the scale, the level this site uses for public money whatever it is spent on.

About 24.1 billion euro a year of sports contracts would come under state gambling law, of which 65 percent is assumed to reach a licensed sportsbook — 15.7 billion euro of handle. American sportsbooks keep 10.2 percent of what is staked, and states take about 22 percent of that in tax, which is what the 2025 figures of 16.96 and 3.71 billion dollars imply ^[2]. Applied here that is 350 million euro a year, in a range from 150 to 600 million. The euro carries the standard weight of one for public money. The exchanges' own fee revenue, which is federally rather than state taxed, is not counted. The Impact is the smallest gain here and it is the one figure in the debate that follows almost entirely from published tax rates.

▸ Show calculation ▾ Hide calculation

	Sports event contract volume [1]		24.1 billion euro
×	Share reaching a licensed sportsbook	65 %	15.7 billion euro
×	Kept by the sportsbook 16.96 billion dollars of 166.94 staked in 2025 [2]	10.2 %	1.6 billion euro
×	Taken in state tax 3.71 billion dollars of 16.96 in 2025 [2]	22 %	352 million euro
×	Weight of a euro in a state budget the standard weight for public money on this site	1.0	352 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	1.76

Score 1.76 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 4.8 of 100

The tax arithmetic is not in question: rates are published, sportsbook revenue is reported monthly, and the ratio between them is measured rather than modelled ^[2]. The counterfactual is the current position, in which the same wagers generate no state revenue at all. What is estimated is how much volume actually arrives in the licensed channel, and that is the same number carrying the first argument, so the two move together. The confounder that matters has changed since the first version of this evaluation: it was thought that customers used to a low stated fee would bet considerably less once they saw a sportsbook's margin, but the transaction data show they were already losing twice that margin without seeing it, so a price-driven contraction is less likely than it appeared ^[6]. Reverse causation does not arise. The Plausibility is at the upper end of what a projection can carry: the tax rate is published and only the volume it applies to is estimated.

evidence basis: Projection · P ceiling 6 identification: Definitional · no rung ceiling

Counterfactual: the current position, in which sports event contracts generate no state gambling tax. Design: definitional — tax follows from published rates applied to reported revenue [2]; the estimated element is the volume that switches channel. Confounder: customers betting less once prices are visible, which the transaction data make less likely, since the realised loss rate already exceeded the sportsbook margin [6]. Direction: not applicable. Ceiling: projektion 6.0 binds. Parameter coupling: the 65 percent switching share is shared with pro-4 and con-4.

The advertising comes under rules

3.6of 100

Because this is not legally gambling, none of the rules on gambling advertising apply to it. The football league has banned prediction market commercials outright, alongside tobacco and firearms, so the marketing moved to paid creators: 140 sponsored videos from one platform in ninety days, 101 from the other, plus social posts that are barely labelled as advertising and promo codes offering bonuses.

Value 9 · HealthImpact 1.1Plausibility 3.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is the same one the argument above counts — gambling disorder and what travels with it — reaching a different group of people. It carries the same weight, one step below the top of the scale. What separates the two is who they concern: that argument is about protecting people already trading, this one is about people who would never have started. Nothing here is priced for the deception of an undisclosed advertisement as such, which is a wrong of a different kind and is dealt with by advertising law rather than by gambling law. The value sits one step below the maximum: the stream is health, and health that can be regained.

Roughly 3 million accounts trade sports contracts, and the sector is growing fast enough that acquisition is where the money goes. A quarter of accounts, about 750,000, are assumed to have been acquired in a year through paid creator content or a bonus offer, in a range from a tenth to a half — one platform sponsored at least 140 YouTube videos in ninety days and the other at least 101, alongside social posts that are undisclosed or barely labelled and promotional codes offering credit to new customers ^[7]^[9]. Of those, a fifth would not have started under the advertising rules a state gambling licence carries, in a range from a twentieth to a half: 150,000 people. Applying the 19 percent rate of problematic behaviour measured among online sports bettors gives 28,500 people, each carrying a loss of 0.2 quality-adjusted years a year. That the football league placed these products in the same prohibited advertising category as tobacco and firearms is not evidence of harm, but it is evidence that the people closest to the market do not regard the current arrangement as normal ^[8]. The Impact is the smallest of the health gains and it reaches the one group the other protections cannot: people who are not customers yet.

▸ Show calculation ▾ Hide calculation

	Accounts trading sports contracts [1]		3 million accounts
×	Acquired through paid creator content or a bonus offer Setting, range 10 to 50 percent: one platform sponsored at least 140 YouTube videos in ninety days and the other at least 101, alongside barely labelled social posts and promotional codes [7][9]	25 %	750,000 accounts
×	Who would not have started under gambling advertising rules Setting, range 5 to 50 percent: the Italian advertising ban of 2019 did not reduce gambling, which is why this share is low	20 %	150,000 people
×	Showing problematic gambling behaviour the rate measured among online sports bettors [3]	19 %	28,500 people
×	Quality-adjusted years lost per person a year	0.2	5,700 quality-adjusted years
×	Value of the years the value of a healthy life year used across this site	40,000 euro each	228 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	1.14

Score 1.14 Impact × 9 Value × 3.5 Plausibility ÷ 10 = 3.6 of 100

The marketing is documented and everything downstream of it is assumed. The counterfactual is the same platforms under state advertising rules, which no American jurisdiction has yet applied to them. That the advertising exists at this scale is a matter of record: sponsorship counts are published by tracking services, the undisclosed nature of much of it has been reported, and promotional codes are visible on the platforms themselves ^[7]^[9]. What has no measurement is the link that carries the whole quantity — how many customers advertising actually creates rather than merely reallocates between platforms. The wider evidence on gambling advertising restrictions is discouraging rather than encouraging: Italy banned gambling advertising outright in 2019 and its own parliamentary inquiry found gambling had risen afterwards, which is the strongest counter-argument here and is why this figure sits low. The confounder is that a fast-growing product attracts users through word of mouth and press coverage that no rule reaches. Reverse causation does not arise. The Plausibility is low because the number of customers advertising creates has never been measured and the nearest attempt to restrict it elsewhere did not reduce gambling.

evidence basis: Mechanism · P ceiling 6 identification: Mechanistic · rung ceiling 6 band: Chain open · P 3–3.5

Counterfactual: the same platforms under state gambling advertising rules — not observed anywhere. Design: mechanistic — the marketing volume is documented [7][9], the conversion to new customers is not. Confounder: organic growth through press and word of mouth that no advertising rule reaches; unaddressed. Direction: no reverse causation. Ceiling: mechanistic 6.0 binds. Band: chain open, because the customer-creation link carries the quantity, has no measurement, and the Italian advertising ban is an unanswered counter-indication for the whole mechanism.

The chain is named but the link carrying the quantity — how many customers advertising creates rather than reallocates — is unmeasured, and Italy's 2019 advertising ban, after which gambling rose, is an unanswered counter-indication. Read back: about a third of the time, advertising rules keep out roughly the number of new customers assumed here.

Open: Sponsorship spending and new-account numbers are both known to the platforms and to their creator partners. Account growth against sponsorship volume, across the states that have already blocked these exchanges, would test the link within a year.

Arguments — Against

4 arguments · top 3 shown

The other side of every trade loses the business

12of 100

What retail customers lose on an exchange does not vanish; it goes to the professionals posting the prices and to the platform's fees. Closing sports contracts takes that away from them, and the sportsbooks that inherit the volume take less. Somebody is worse off, and it is mostly people for whom a euro is worth less.

Value 5 · Household budgetsImpact 4.8Plausibility 5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money, priced at the middle of the scale. It is the paying side of the transfer whose receiving side is the first argument: what retail customers keep, somebody else does not get. Those somebodies are professional market makers, the exchange collecting fees, and — for the share that survives the move — sportsbook operators. All of them sit at the top of the income and wealth distribution, which is why the same euro counts for less here than it does in a customer's hand; that difference is in the Impact rather than here. Nothing is counted for the exchanges as businesses beyond the money, because a company is not a person. The value is the middle of the scale, because the stream is money and the distance between its two ends is priced in the Impact.

Of the 15.7 billion euro of staking that moves, the professionals and the platform currently take about 20 percent, or 3.14 billion euro a year ^[6]. After the move, sportsbook operators take 10.2 percent of the same handle, of which 22 percent goes to the state, leaving them about 1.25 billion ^[2]. The net loss to that whole side is therefore 1.89 billion euro a year, in a range from 0.6 to 3.2 billion. The people concerned are market makers, quantitative trading firms and the shareholders of listed exchanges and sportsbooks, where this site counts a euro at half its worth at median income: 0.95 billion euro. What is not counted is the loss to the exchanges' non-sports business, which is a separate argument below. The Impact is the largest against and it is by construction about half the first argument, because the money moves between two ends whose weights differ by a factor of nearly three.

▸ Show calculation ▾ Hide calculation

	Taken from customers on the exchange [6]	20 % of 15.7 billion euro of staking	3.14 billion euro
−	Taken by sportsbook operators after the move [2]	10.2 % less the 22 % that goes to the state	1.89 billion euro
×	Weight of a euro on that side market makers, quantitative trading firms and the shareholders of listed exchanges and sportsbooks sit at the top of the wealth distribution	0.5	0.95 billion euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	4.75

Score 4.75 Impact × 5 Value × 5 Plausibility ÷ 10 = 12 of 100

This figure is the arithmetic complement of the first argument and rests on the same two measured rates. The counterfactual is the current position. What it inherits from the first argument is that argument's central weakness: if switching customers buy parlays rather than straight bets, sportsbook operators take 25 to 30 percent rather than 10.2, and the loss to this side is much smaller than shown — the two arguments move against each other and cannot both be wrong in the same direction. What is specific to this one is the weighting: market makers are unambiguously capital, but the exchange's fee revenue funds staff and technology as well as returns, so treating the whole of it at a capital weight overstates the argument somewhat. Reverse causation does not arise. The Plausibility is at the middle, the same as the argument it mirrors, because it rests on the same two measurements and the same untested behavioural step.

evidence basis: Converging studies · P ceiling 7 identification: Controlled · rung ceiling 7

Counterfactual: the current position, with both channels open. Design: controlled — the arithmetic complement of pro-4, resting on the same two administrative measurements [2][6]. Confounder: switching customers buying parlays, which would raise sportsbook take to 25-30 percent and shrink this loss; the same confounder as pro-4 and moving in the opposite direction, so the two cannot both be wrong the same way. Direction: no reverse causation. Ceiling: controlled 7.0 binds; a context transfer of 2.0 applies as in pro-4. Netting: the transfer is booked once as two legs — customers' gain in pro-4, this side's loss here — with the state's share separated into pro-2.

Some of it goes where nobody is looking

4.4of 100

Prediction markets on sport existed offshore for years before a licensed American exchange listed them, and American customers reached them without difficulty. Closing the regulated version does not close the market. What it closes is the version that files reports.

Value 5 · Household budgetsImpact 2.5Plausibility 3.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money lost where there is no recourse, priced at the middle of the scale like other money. An offshore or crypto-settled market has no age check worth the name, no segregated customer funds, no reporting obligation and no authority a customer can complain to. What is priced here is both the transfer that does not stop and the extra harm of it happening beyond reach. The problem gambling that follows a customer offshore is counted in the argument on licensing protections rather than a second time here, since it is the same person and the same disorder. The value is the middle of the scale, because the stream is money and the absence of protection around it is priced in the Impact.

Of the 24.1 billion euro of sports contract volume, 15 percent is assumed to move to markets outside any American regulator, in a range from 5 to 35 percent — 3.6 billion euro. For that share the licensing protections counted above do not arrive and the state tax does not either; the customers simply carry on somewhere with fewer safeguards than they had before the measure. The loss is set at 14 percent of the migrating volume, which is roughly the difference between an offshore operator's take and what the same customer would have lost on the regulated exchange, in a range from 5 to 25 percent: about 500 million euro a year. Crypto-settled markets make this easier than it was for offshore sportsbooks, because there is no payment processor to block. The Impact is a quarter of the money customers keep under this measure and it eats directly into the protections that are part of the case for it.

▸ Show calculation ▾ Hide calculation

	Sports event contract volume [1]		24.1 billion euro
×	Share moving beyond any American regulator Setting, range 5 to 35 percent: crypto-settled markets have no payment processor to block, but some platforms geofence effectively	15 %	3.6 billion euro
×	Loss per euro that goes there Setting, range 5 to 25 percent: roughly the difference between an offshore operator's take and what the same customer would have lost on the regulated exchange, plus the absence of any recourse	14 %	504 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	2.52

Score 2.52 Impact × 5 Value × 3.5 Plausibility ÷ 10 = 4.4 of 100

The pattern is established and the size is not. Offshore prediction markets on sport operated for years with American customers before a licensed exchange listed the same contracts, which is the precedent this argument rests on, though it is a description rather than a measurement — nobody counted the American volume on those platforms. The counterfactual is the current position with a licensed alternative available. The chain is short and visible: the regulated version closes, the customer wants to continue, an unregulated one is reachable. The counter-mechanism is genuine and only partly answered: several of those offshore platforms geofenced American users under regulatory pressure and some now do so effectively, which would hold the migrating share below the figure used. That is why the range runs down to 5 percent. Reverse causation does not arise. The Plausibility is low because the migrating share has never been measured and the effectiveness of geofencing against it is unresolved.

evidence basis: Precedent · P ceiling 6 identification: Mechanistic · rung ceiling 6 band: Chain open · P 3–3.5

Counterfactual: the current position with a licensed alternative available. Design: mechanistic — offshore prediction markets with American customers are a described precedent rather than a measured one; no source counts the volume. Confounder: geofencing under regulatory pressure, which some platforms now apply effectively; partly answered by the low share used, not resolved. Direction: no reverse causation. Ceiling: mechanistic 6.0 binds below the praezedenz ceiling of 8.5. Band: chain open, because the migrating share carries the quantity and the geofencing counter-mechanism is unresolved.

The chain is named but the migrating share has never been counted, and the counter-mechanism — geofencing that some offshore platforms now apply effectively — is unresolved. Read back: about a third of the time, roughly the assumed share of volume moves beyond any American regulator.

Open: Thirteen states are already enforcing against these exchanges. Comparing crypto-settled market activity from those states against the rest, before and after enforcement, would measure the migration directly.

The forecasts go with the volume

4.3of 100

Prediction markets produce a continuously updated probability for anything they list, and those numbers are used by journalists, researchers and companies. Sport is four fifths of the volume that pays for the platform. Take it away and the markets on elections, interest rates and public health may not survive on their own.

Value 7 · Public debateImpact 1.8Plausibility 3.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is the quality of public information: a number that says how likely something is, produced by people with money at stake rather than by anyone with an argument to make. This site places that in the class it uses for participation and the conditions of public debate, above money and below health. The value is not in any single forecast but in having an independent series at all, which is why removing the volume that funds the platform matters more than removing any one market. Nothing here is counted for the entertainment of following the odds, which belongs with the enjoyment of betting and is not priced in this evaluation. The value sits in the upper middle of the scale, at the level this site uses for the conditions of public debate.

Sport is about four fifths of the volume on the largest exchange, and it is the volume that pays for the technology, the compliance and the market-making that keep the other contracts liquid ^[1]. Removing it does not automatically end the election and economic markets, but it removes the business that supports them. What those forecasts are worth has no market price, so this evaluation sets one: 350 million euro a year, in a range from 50 million to 1 billion. The lower end treats them as a curiosity that better-funded polling and forecasting already supply; the upper end treats them as a genuinely superior instrument whose accuracy in the 2024 and 2026 election cycles is now documented. This is the least grounded number in the debate and it is stated as a price rather than derived. The Impact is small, and it is the one stream here that would be entirely avoidable by writing the measure to exempt the platforms rather than the contracts.

▸ Show calculation ▾ Hide calculation

	Share of exchange volume that is sport [1]	four fifths on the largest platform	80 percent
=	Value set on the forecasting series the volume supports Setting, range 50 million to 1 billion: the low end treats them as a curiosity that polling already supplies, the high end as a superior instrument whose accuracy is documented	350 million euro a year	350 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	1.75

Score 1.75 Impact × 7 Value × 3.5 Plausibility ÷ 10 = 4.3 of 100

Two things would have to hold and neither has been shown. The counterfactual is the exchanges continuing with their non-sports contracts alone, and nobody has established that they could not: the platforms existed before sports contracts were listed, on much smaller volumes, and several operate profitably outside the United States without them. That is a serious counter-argument to the whole chain and it is unresolved. Beyond it sits the question of what the forecasts are worth, which has a literature — prediction markets have repeatedly matched or beaten polling averages — but no method for pricing. The confounder in that literature is that the markets read the polls, so their accuracy is partly borrowed rather than independent, and the transaction data now available show the same prices carrying a systematic favourite–longshot bias, which is a caution about treating them as clean probabilities ^[6]. Reverse causation does not arise. The Plausibility is low: the claim rests on a price nobody has estimated and on a business-model assumption that the platforms' own history argues against.

evidence basis: Plausibility · P ceiling 5 identification: Mechanistic · rung ceiling 6 band: Chain open · P 3–3.5

Counterfactual: the exchanges continuing with non-sports contracts alone — which their own pre-sports history suggests is possible. Design: mechanistic — chain named (sports volume funds the platform → platform sustains other markets → forecasts exist), with the first link contradicted by the platforms' own history. Confounder: prediction markets reading the polls, so their forecasting accuracy is partly borrowed; and the documented favourite–longshot bias in their prices [6], which cautions against treating them as clean probabilities. Direction: no reverse causation. Ceiling: plausibilitaet 5.0 binds because the quantity is a stated price. Band: chain open, since the business-model link is unresolved and the price carries everything.

The chain is named but its first link — that the platforms cannot sustain non-sports markets without sports volume — is contradicted by their own history of operating before sports contracts were listed. Read back: about a third of the time, removing sports contracts costs the public the forecasting series as well.

Open: Several states already block these exchanges. Comparing the liquidity of election and economic contracts before and after those blocks took effect would test the business-model link directly.

An exchange will not throw you out for winning

1.5of 100

A sportsbook that keeps losing to a customer restricts or closes their account, which is legal everywhere and routine. An exchange has no position to protect, so it does not care who wins. It also holds customer money in segregated accounts under federal rules and does not lend against a bet.

Value 5 · Household budgetsImpact 0.8Plausibility 4

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money and the security of money: what a skilled bettor is prevented from earning when a sportsbook closes their account, and what any customer stands to lose if an operator fails holding their balance. It is priced at the middle of the scale like other money. Nothing here is counted for the fairness of being restricted for winning, which is a grievance rather than a stream, nor for the convenience of trading out of a position, which is a feature rather than a harm. The great majority of customers are never restricted and never affected by either, which is why the figure is small. The value is the middle of the scale, because the stream is money and its security.

Two small things sit under this argument. The first is the skilled minority: perhaps 50,000 people trade sports contracts with a genuine edge and would be limited or closed by a sportsbook within months, losing something like 2,000 euro a year each — about 100 million euro, in a range from 20 to 300 million. The second is the safety of customer balances: the exchanges hold them in segregated accounts under federal rules, while sportsbook balances are in several states an ordinary claim against the company. Against perhaps 2 billion euro of balances, an operator failure at one percent a year with half the money recovered gives about 10 million euro of expected loss. Together roughly 150 million euro a year. The transaction data cut against the first half rather than for it: on the exchange, the side posting prices earned minus 9.64 percent, so even the skilled were losing, just less ^[6]. The Impact is the smallest in this debate and it is the one argument for exchanges that survives their own loss data.

▸ Show calculation ▾ Hide calculation

	Customers trading with a genuine edge who a sportsbook would restrict Setting, range 20,000 to 150,000: the measured maker return of minus 9.64 percent suggests the winning group is small [6]		50,000 people
×	Lost to each a year Setting, range 400 to 6,000 euro	2,000 euro	100 million euro
+	Expected loss from an operator failure holding unsegregated balances exchange balances are segregated under federal rules; sportsbook balances are an ordinary claim in several states	2 billion euro × 1 % × half recovered	150 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	0.75

Score 0.75 Impact × 5 Value × 4 Plausibility ÷ 10 = 1.5 of 100

The premises are documented and the sizes are constructed. That sportsbooks restrict winning customers is universal practice, openly acknowledged and legal in every American jurisdiction; that federally regulated exchanges segregate customer funds follows from the rules they operate under. The counterfactual is the current position with both venues available. What has no source is the number of customers with a genuine edge, and the transaction data suggest it is smaller than the argument assumes: the makers' average return was minus 9.64 percent, so the group that this argument treats as winners was on average losing ^[6]. That is a direct counter-finding on the load-bearing quantity. Against it, the makers' figure is an average across a population that certainly contains both winners and losers, and the argument concerns the winners. Reverse causation does not arise. The Plausibility is below the middle: both premises are certain and the number of people who actually benefit is contradicted by the same data that carry the rest of this evaluation.

evidence basis: Mechanism · P ceiling 6 identification: Mechanistic · rung ceiling 6 band: Chain closed, unevidenced · P 4–5

Counterfactual: the current position with both venues available. Design: mechanistic — sportsbook restriction of winning customers and exchange fund segregation are documented practice; the number of customers with an edge is constructed. Confounder: the makers' measured average return of minus 9.64 percent, which suggests the winning group is smaller than assumed [6]; named and only partly answered, since an average conceals a distribution. Direction: no reverse causation. Ceiling: mechanistic 6.0 binds. Band: chain closed but unevidenced — links named, the counter-finding stated, only the count of skilled customers missing.

The premises are documented; what is missing is any count of customers with a genuine edge, and the measured maker return of minus 9.64 percent suggests that group is small. The counter-finding is named and partly answered by the fact that an average conceals a distribution. Read back: about half the time, roughly the assumed number of skilled customers lose access to a venue that tolerates them.

Open: The exchanges hold the distribution of customer returns, not just the average. Publishing the share of accounts with positive lifetime returns would settle the size of this argument outright.

Summary

The industry's central argument for being left alone is that an exchange is cheaper than a bookmaker: it charges a fee of one or two percent where a sportsbook keeps about a tenth of everything staked. Transaction-level data on 313,972 contracts show that is not what happens to the people using it. The average return is about minus 20 percent, and minus 31 percent for those who accept a quoted price rather than posting one — worse than a sportsbook, not better, because two thirds of the contracts are longshots and longshots on this venue lose more than 60 percent. On top of that sits marketing that no rule reaches: the football league bans these advertisements alongside tobacco and firearms, so the platforms buy creators instead, and much of it is not labelled as advertising at all. What holds the balance back from being one-sided is that a fifth of the volume plausibly leaves for offshore markets, and that the same person moved to a sportsbook may buy parlays, where margins reach 25 to 30 percent and the saving disappears. That last point is the one thing that would restore the earlier reading of this debate.

Outlook — effect over time

Better for the future · 0.69 previous scale

Sources

DLA Piper: Legal status at odds: tracking developments in prediction markets and sports betting. dlapiper.com
Sports Handle: US Sports Betting in 2025 Reaches Record Highs. sportshandle.com
National Council on Problem Gambling: National Survey on Gambling Attitudes and Gambling Experiences 3.0. ncpgambling.org
Journal of Gambling Studies, comparative policy review: Limit-setting in online gambling: a comparative policy review of European approaches. pmc.ncbi.nlm.nih.gov
Congressional Research Service: CFTC Issues Proposed Rule Regarding Prediction Markets. congress.gov
Bürgi, Deng and Whelan, University College Dublin: Makers and Takers: The Economics of the Kalshi Prediction Market. karlwhelan.com
Tubefilter: Prediction markets like Kalshi and Polymarket are everywhere, but not at the Super Bowl. tubefilter.com
Front Office Sports: NFL Won't Allow Prediction-Market Super Bowl Commercials. frontofficesports.com
Slate: Polymarket and Kalshi's secret ads are flooding social media. slate.com

Last reviewed by Claude Opus 5 · September 6, 2026 · 2× AI, not yet reviewed by a human

September 6, 2026AI reviewClaude Opus 5re-scored
Re-scored after checking the channel-cost claim: transaction data on 313,972 Kalshi contracts show retail losing about 20 percent, not the stated fee, so v1's largest contra argument was wrong and is withdrawn; advertising added as a separate stream.
September 6, 2026AI reviewClaude Opus 5First evaluation
Created for the English side: the channel-shift cost from an exchange fee to a sportsbook hold is what keeps the balance close.

Evaluations are produced with AI support and reviewed on a schedule for new developments; human passes are marked separately.How we review →