Why the house wins

Since Murphy v. NCAA (May 2018) ended the federal sports-betting ban, 39 states + DC have legalized some form of sports betting, 30+ with mobile-app access. California, Texas, Georgia, Alabama, and South Carolina are the remaining major holdouts; most are actively debating bills. The average new state legalizes within 18 months of Murphy's effect reaching its legislature. Industry projections assume 45+ states within 5 years.

Americans wagered approximately $121 billion on sports in 2024, up from <$10B in 2018 pre-Murphy and $93B in 2022. Handle (total amount wagered) is distinct from revenue (the house's take). The compounded 2018-24 growth rate is over 60% per year, one of the fastest industry expansions since post-Prohibition alcohol. Mobile apps account for ~85% of the handle; physical casinos are functionally a long tail now.

FanDuel (Flutter) and DraftKings together take roughly 70% of US sports-betting revenue. BetMGM, Caesars, and a long tail of smaller operators split the rest. The duopoly has entrenched at a speed that vastly outpaces federal telecom, airline, and banking consolidation over comparable periods. Both companies lose money on most customers; profits come from the top 1-5% of users, heavy-loss VIP customers, which the business model is specifically optimized to cultivate.

Pew Research / YouGov 2024: approximately 1 in 4 US men ages 18-34 placed at least one sports bet in the past year, up from <1 in 10 pre-2018. The demographic shift is generational and sharper than almost any other commercial behavior change measured in the period. Young men now treat sports betting as a default component of watching a game, with the app open during play. Teenagers report awareness and in some cases access via parent accounts.

Of the ~$121B wagered, sportsbooks kept roughly $13B as gross gaming revenue (the 'hold'). The aggregate hold percentage is roughly 10.8%, meaning for every $100 wagered the house keeps ~$11. This is significantly higher than the house edge on blackjack (~0.5-2% played correctly) or American roulette (~5.3%) and comparable to slot machines. The high hold is concealed by the structure of sports wagers: most bettors track their wins and ignore their losses, so the 90% return feels like winning.

Same-game parlays (SGPs) are the industry's fastest-growing product. Apps surface them prominently because the hold is dramatically higher: SGPs typically clear 25-35% hold, vs ~5% on straight bets. A DraftKings S-1 filing cited "parlay mix" as a key growth metric. The math is combinatorial; each additional leg compounds the correlation the book prices into the line. These are presented to casual users as the same product as single-bet moneylines.

States collected approximately $2.5 billion in sports-betting tax revenue in 2024, with New York, Pennsylvania, New Jersey, and Illinois taking the largest shares. Tax rates range from 6.75% (Nevada) to 51% (New York). States have every incentive to legalize and promote the activity; problem-gambling treatment funding is typically 1-2% of tax receipts, small relative to the external costs. The fiscal arrangement mirrors state tobacco revenue in the 1990s: the state has a split interest between public-health and fiscal outcomes.

FanDuel, DraftKings, BetMGM, and Caesars collectively spent approximately $3 billion on US marketing in 2023, including television, in-stadium signage, influencer campaigns, and 'free-bet' promotional credits. Per-customer-acquisition cost (CAC) peaked at $600-$1,000+ during the 2021-22 land grab and has since compressed. Marketing heat is highest in newly-legalizing states where brand preference is still being set; industry veterans have characterized the period after legalization as a 'shock and awe' window.

National Council on Problem Gambling: calls to state helplines have risen 30-50% since legal mobile sports betting launched. Specific states (New Jersey, Pennsylvania, Michigan) have seen helpline volume double or triple. Problem-gambling prevalence in the US is estimated at 2-6% of adults, with a further 10-15% showing at-risk indicators. Treatment capacity is limited; Gamblers Anonymous attendance is rising. NCPG estimates ~$14B/yr in US external costs (bankruptcy, criminal, family) attributable to problem gambling.

Baker, Koelker, and Wood (2024, NBER working paper): US counties 4+ years after state sports-betting legalization saw personal bankruptcy filings rise approximately 28% relative to control counties, with sharper effects in low-income households and among men 25-44. Auto loan delinquency and credit card default rates also rose significantly. The causal inference is identification-by-state-timing and is contested, but the direction is consistent across multiple independent replications.

National Research Group 2024 study: the median US sports bettor who wagered at least monthly lost approximately $1,100-$1,800 per year; the top-quartile by volume lost $3,400+. Heavy users (top 5% of bettors) can lose $25K+ per year; this group generates the large majority of sportsbook profits. Sportsbook algorithms flag high-losing users for "VIP host" programs that provide personalized bonuses and white-glove service, a retention tactic the industry has imported from Vegas but accelerated with app-level behavioral data.

Peer-reviewed meta-analyses (Yakovenko & Hodgins 2018, updated 2023) find that problem gamblers have suicide-ideation and suicide-attempt rates roughly 2× the general-population base rate, with completed suicides elevated to a similar degree. The relationship is well-documented in the academic literature and is underweighted in public-health discussion of the industry. Veterans, who are over-represented in sports-betting product design partnerships, show particularly elevated rates.

The standard sportsbook "free bet" or "bonus bet" returns only winnings, not stake. A winning $100 free bet on +100 odds returns $100 (not $200). Expected value to the bettor is significantly less than face. The industry-standard bet-and-get structure, where users must wager their own money first at specified odds to unlock bonus credits, is an industry-optimized retention tool; bonus terms typically include play-through requirements that functionally trap credits inside the app.

VIP host programs assign a named account manager to high-volume users, with benefits (event tickets, personalized bonuses, loss rebates) triggered by loss-magnitude rather than any skill marker. NYT investigations have documented VIP hosts successfully re-engaging users who had self-excluded or filed loss complaints. The industry's defense is that VIP programs are legal adult-to-adult commerce; the pattern-matching to casino-floor practices that already triggered state-level Nevada Gaming Control Board regulation is nonetheless close.

App behavioral-design analyses: a typical sports-betting app session contains 10+ distinct nudges toward additional wagering (push notifications, in-app offers, featured same-game parlays, streak-reward banners, live-odds alerts). The design vocabulary is shared with mobile gaming (loot boxes, daily-login rewards) and has been empirically linked to disordered-use patterns in adolescents and young adults. The apps are rated 17+ on Apple App Store; most carry no equivalent of mobile-game age-gating for the gambling mechanics themselves.

PrizePicks, Underdog, and similar daily-fantasy-sports (DFS) apps operate in states where traditional sportsbooks are not yet legal (California, Texas, Florida). Regulators in California, New York, and several other states have concluded the products are functionally indistinguishable from parlay sports betting. The DFS industry maintains the distinction; courts and state AGs are in active litigation. The regulatory arbitrage brings sports-betting product patterns to teenage and college-age users in states that have not yet legalized.

Donald Trump Jr. joined Kalshi's advisory board in January 2025, weeks before inauguration. The CFTC under the new administration walked back its appeal against Kalshi's election contracts. Brian Quintenz, Trump's nominee for CFTC chair, is a former Kalshi board member. The FBI investigation opened against Polymarket founder Shayne Coplan (Nov 2024, under Biden, including an agents' raid on his home) was quietly closed after inauguration. Across two platforms, the regulator writing the event-contract rules the industry is using to bypass state gaming law now runs directly through the White House.

Kalshi launched NFL, NBA, and March Madness "event contracts" in Jan-Mar 2025, binary derivatives priced exactly like sports-betting lines but routed through CFTC event-contract rules instead of state gaming commissions. Seven state regulators (NJ, NV, MA, OH, MT, MD, AZ) issued cease-and-desist orders. Kalshi won preliminary federal injunctions in New Jersey and Nevada in April 2025 on a federal-preemption theory. The side door lets the platform operate in California, Texas, and Florida, where traditional sportsbooks are still illegal, without state licensing, state age-verification rules, state tax, or state problem-gambling funding. Super Bowl 2025 contract volume alone cleared $400M.

Polymarket was banned from the US in 2022 under a CFTC enforcement action for offering unregistered event derivatives. After acquiring QCX for its CFTC derivatives license (~$112M, Nov 2024) and the Trump administration easing enforcement, Polymarket resumed US operations in 2025. The 2024 election cycle alone cleared ~$8B in traded volume, with cumulative platform volume growing substantially through 2025-26. Settlement is in USDC. No FDIC/SIPC equivalent, no state consumer-protection layer. Users load crypto and trade binary contracts on essentially any real-world outcome, elections, Fed decisions, weather, celebrity deaths.

On-chain wallet-level analysis (Chainalysis, public Dune dashboards) put the top ~10 wallets at 40%+ of 2024-election contract volume, driving most of the late-cycle price movement toward a Trump win. Retail traders piling in at the final odds were systematically on the losing side; dashboards estimate roughly 65-75% of smaller wallets ended net-negative on election contracts. The "wisdom of crowds" marketing frame inverts the mechanics: thin binary markets are whale-dominated, and market-maker edge plus gas and spread costs structurally work against small traders. The extraction dynamic mirrors sportsbook retail, with a different front-end and no state problem-gambling helpline.

April 2026: a Polymarket bettor appears to have used a hair dryer to heat the single Météo-France weather sensor near Paris Charles de Gaulle airport and force a temperature-market contract to resolve in their favor. The sensor reading jumped 4°C in 12 minutes on April 6, crossing the 22°C market threshold while other sources showed lower temps. A similar anomaly occurred April 19. Météo-France filed a criminal complaint; meteorologists ruled out natural causes; Polymarket switched its Paris resolution source to Le Bourget airport. Total payout: ~$34,000. The incident is the textbook illustration of the 'oracle problem': prediction markets resolve against real-world data feeds, which can themselves be tampered with. No insider knowledge required, just a household appliance and $10 at the airport curb.

On April 22, 2026, Kalshi suspended and fined three congressional candidates for political insider trading, wagering on Kalshi contracts tied to their own campaigns. Mark Moran (VA Senate ind.) was fined $6,200+; Matt Klein (MN-02 Dem primary) ~$540; Ezekiel Enriquez (TX-21 GOP primary) under $100. All three were suspended for 5 years. The platform's self-policing caught them; federal securities laws on insider event-trading are still being written. Prediction markets import the insider-trading exposure they were marketed as structurally immune from, and unlike equities have no SEC-level enforcement apparatus behind them.