Risk-free is a marketing term. The actual expected value is what changed. A timeline of how a math-literate corner of the internet broke the US sportsbook promo model, what books are doing about it, and what is left for anyone still paying attention.
From 2018 through 2022, the dominant acquisition strategy for newly licensed US sportsbooks was a straightforward formula: give new depositors a risk-free first bet, or a bet-credit match, worth anywhere from $100 to $1,000. The premise was standard customer-acquisition math — spend now to lock in long-run handle. What the books underestimated was how quickly a segment of the user base would treat those promos not as a soft welcome gift but as a systematically exploitable asset with a defined cash-equivalent expected value.
The promo era, 2018-2022
PASPA's 2018 repeal triggered a state-by-state licensing race. Books entering New Jersey, Pennsylvania, Indiana, and Colorado needed depositors fast, and the fastest tool was a generous welcome offer. DraftKings, FanDuel, BetMGM, and Caesars collectively disclosed spending in the range of $1.5-2 billion on promotions between 2018 and 2022 in their quarterly regulatory filings. The model assumed a promo cost-per-acquisition that would be offset by lifetime handle — a playbook borrowed directly from DFS and online poker, both of which had burned through similar acquisition cycles earlier.
The assumption was reasonable on average. The problem was the distribution: the cohort that understood the actual math of a risk-free offer was small but efficient, and it was organizing faster than book risk desks could identify it.
How the math-literate cohort organized
Matched betting and arbitrage communities had existed before US legalization, mostly in UK and Australian markets where legal betting had been running for decades. When US legal markets opened, that knowledge base migrated. Twitter accounts posting US promo expected-value calculations grew to tens of thousands of followers within 18 months of major-state openings. YouTube tutorials walked through the specific mechanics: a $1,000 risk-free bet with a 10% cash-withdrawal-conversion rate was worth roughly $180 in cash-equivalent expected value. Then how to hedge the qualifying side at an offshore book to lock in near-certainty.
Reddit communities dedicated to US promo extraction had 50,000-plus members by 2021. The methodology was transparent, step-by-step, and accessible to anyone who could follow a spreadsheet. For a segment of the depositor base — roughly 5-10% in acquisition-heavy states — promo yield rather than entertainment was the primary motive for opening an account. Books, competing hard for raw depositor count, were not structurally designed to identify or de-prioritize this segment during the acquisition phase.
The 2022-2024 extraction wave
By 2022, the methodology had been refined to a near-institutional level. Organized matched bettors maintained accounts across every state they qualified for, tracked offer calendars through third-party sites that aggregated current promos, and moved capital systematically between books based on reload-offer windows. Reload promos — weekly free-bet offers for existing customers — proved to be the most durable value source precisely because they were harder to eliminate: cutting them risked accelerating churn among recreational customers who also received and appreciated them.
Accounts began receiving restrictions in 2022-2023. Books developed behavioral flagging systems — bet-to-withdrawal ratio, market-selection patterns, account network analysis — and began limiting or closing accounts identified as pure-promo players. But the account restrictions preceded the promo cuts by 12-18 months, because cutting promos broadly required accepting the acquisition cost of ceding share to a competitor still offering them.
| 2021 (peak promo era) | ~38-45% |
|---|---|
| 2022 | ~28-35% |
| 2023 | ~18-25% |
| 2024 | ~12-18% |
| 2026 Q1 trajectory (consensus estimates) | ~8-12% |
The regulatory squeeze
While internal math was pushing books to cut, external pressure arrived from a different angle. Several state gaming regulators began scrutinizing promo advertising language after consumer complaints about the gap between advertised and received value. The phrase 'risk-free bet' became a specific regulatory target: if the bettor receives a bet credit rather than cash on a loss — and that credit carries wagering requirements before withdrawal — calling it 'risk-free' arguably misrepresents the product's structure.
Massachusetts, Ohio, and New York issued guidance language that pushed operators toward neutral terminology. The AGA's responsible marketing code, updated in 2023, added a recommendation that members avoid 'risk-free' language and use 'first-bet bonus' or similar neutral terms. Most major books updated landing pages by mid-2024. The public framing was responsible marketing. The underlying driver was regulatory risk — and the cost of compliance was minimal since promo budgets were already contracting anyway.
What replaced the promo: less math, more opacity
The promo ecosystem has restructured around products with less transparent math and, in most cases, less favorable expected value for the user. Odds boosts are now the primary promotional vehicle: a book takes a $100 parlay that prices at +400 and boosts it to +450, generating a marketing moment without the full cash cost of a free-bet credit. The expected value of a boost is almost always positive relative to the default price, but it is harder to calculate precisely — particularly for multi-leg SGP boosts where each leg's individual price is less transparent.
Profit-boost tokens (multiply your winnings by 1.25x on this bet, one-time use) are another format that defies the standard matched-betting methodology because the boost applies only to a winning outcome, making the hedging calculus different. First-touchdown, first-scorer, and same-game parlays have captured share as primary advertising hooks — formats where the margin is structurally higher and the extraction methodology is less documented in public communities.
The net effect for a bettor paying attention to value: the replacement-era products are marginally harder to evaluate than the 2019-2022 risk-free window, but they are not unmathematizable. An odds boost that takes a price from -110 to +100 on a near-even market adds approximately 4.8% to your expected value on that specific bet. That is a real number, and the calculation is the same implied-probability arithmetic that makes the Lexicon entry on arbitrage work — the only difference is that you need the boosted price and the vig-free fair-value consensus to compute it.
The floor, not the ceiling
The structural shift in promo economics is probably complete. Books are not returning to $1,000 risk-free offers at scale: the acquisition math changed, the regulatory environment changed, and the matched-betting community is now well-identified by book risk systems. What remains is a thinner, more targeted promo layer aimed at specific acquisition windows and reactivation campaigns — not the broad-sweep welcome offers that defined 2019-2022.
For most users, the long-run implication is that the product they are buying today is priced at something close to its actual market rate, which — for standard sides and totals — is still a negative-expected-value proposition for the bettor. The promos were a discount window. The promo-era math-literate cohort that extracted the value was not doing anything the books didn't know was possible; they were doing it at a scale the books didn't price into their acquisition models. The correction arrived predictably. The remaining question is whether the next generation of promotional product will present a similarly well-defined extraction window, or whether the opacity has been baked in deliberately.
My read: it is the latter. The replacement formats were not chosen randomly. They are harder to model, harder to hedge, and harder to compare across books — by design. Whether that represents sound business practice or the low road of confusing your customers is a matter of where you sit. What it is not is a mystery.
Why did sportsbooks run such expensive promos if they knew some people would exploit them?
Early-market acquisition strategy prioritized handle growth and depositor count over short-run profitability. In a licensing race, ceding acquisition share to a competitor was viewed as a larger risk than overpaying on promos. Account-level exploitation was anticipated but underestimated in scale — specifically, the speed at which extraction methodology would diffuse through online communities and the organizational efficiency that emerged from dedicated communities tracking and coordinating offer calendars.
Are there any current promo types that still offer genuine positive expected value for non-professional bettors?
Odds boosts on short-priced selections can offer positive expected value if the boosted price exceeds the vig-free consensus. The calculation: convert the boosted odds to implied probability, compare to the fair-market consensus from two or three books that price the same market, and note whether the boosted price implies a lower win probability than the consensus fair value. If it does, the boost is adding real expected value. The window is smaller than 2019-2022 risk-free offers, but it exists on specific boosts — particularly single-game moneyline boosts on well-priced favorites.
What is the difference between matched betting and arbitrage, exactly?
Matched betting uses a free-bet or bonus credit on one side of a market while hedging the other side with a qualifying bet at an exchange or a different book. The goal is to convert the bonus into cash regardless of outcome — the profit comes from the bonus credit, not from any pricing difference. Arbitrage bets both sides of a market at prices that guarantee a profit regardless of outcome — no bonus required, but the pricing difference must exist simultaneously across books. Matched betting was the primary extraction methodology during the US promo era because free-bet credits are the asset being converted. Arbitrage is rarer and typically smaller margin, but it does not require a promotional window.