The In-Play Revolution

Introduction: The Slowest Screen in the Building

On the evening of February 8, 2026, Stats Perform stationed spotters inside Levi’s Stadium with a simple assignment: watch a play happen on the field, then watch it happen again on a screen, and time the gap. Over-the-air television ran about nineteen seconds behind the field. Cable ran thirty-eight. Peacock ran forty-eight, YouTube TV and Hulu fifty-three, DirecTV Stream sixty, and the NFL’s own streaming service sixty-two (Keys). Everyone watching Super Bowl LX at home was watching the past, and the only question was how much of it.

Somewhere in that same minute, a very different clock was running. Sportradar, which holds the official data rights to the NBA, MLB and NHL, has said that it collects and distributes game data to its sportsbook partners in about seven-tenths of a second (Neiburg). Its trading product advertises audiovisual feeds “up to eight seconds faster than any TV broadcaster” (Sportradar, “Live Channel Trading”). Genius Sports, the NFL’s exclusive data partner, puts the ordinary broadcast lag at “45 seconds or more” and sells a betting stream that cuts it to four to eight (Lindenberg). By the time the ball has been snapped on a living-room television, the sportsbook has already priced the play, moved the line, and in many cases settled the bet.

This article is about that gap. It is not about what the live betting interface does to the brain; article 2.1 covered the dopamine story, and found the neuroscience thinner than its reputation. It is about market structure: who sees the game first, who sets the price, who holds the bet before deciding to accept it, and what happened to the one group of people who ever managed to be faster than the house. The in-play revolution is sold as the moment sports betting became a game of watching closely. Its actual architecture is a race the bettor is not permitted to win, run against a counterparty who fired the starting gun half a minute ago.

The word doing the work in “live betting” is “live.” It describes the picture. It does not describe the bet.

The Atomized Game

The scale of the thing first. DraftKings told its investors in March 2026 that 54 percent of its 2025 sportsbook handle arrived while games were in progress, and that it now offers 517 live betting options per game against 124 in 2022 (DraftKings, “2026 Investor Day”). FanDuel’s chief commercial officer was saying in 2022 that live betting was already “around 50% of sports betting in the US” (Mazzeo). An Optimove analysis of 3.8 million bettors across five countries in 2024 put the American live share of bets at 52 percent, and found that the average American live bettor staked $1,583.90 a month against $846.20 for the pre-match bettor (Optimove). In the mature markets of Europe, bet365 has reported in-play at 68 to 75 percent of sports revenue (“Bet365 Sees In-Play Fall”).

Inside that live handle sits the product this article most concerns itself with: the micro-bet, the wager on the next pitch, the next play, the next possession. Simplebet, the automated pricing company DraftKings agreed to acquire in August 2024, reported that in the 2023 baseball season “pitch results” alone accounted for 40 percent of all wagers it processed (Welman). On FanDuel’s Simplebet-powered football product the average user was placing thirty-six bets per Sunday game as early as 2020 (Rybaltowski, “Simplebet Appears Ready”). These are not bets on a game. They are bets on a sequence of events inside a game, each one settled in seconds, each one priced by a machine.

That machine is the subject. Simplebet’s founder described the pricing operation in 2020 in one sentence: “We don’t have traders, that’s the beauty of it. Everything we’ve built is through machine learning and automation, there’s no human intervention” (Rybaltowski, “Simplebet Appears Ready”). The company said its system processes a transaction in 250 milliseconds (Gibbs). nVenue, the micro-market pricing firm in which the NBA took an equity stake in 2023, advertises prediction generation in under fifty milliseconds and more than a million markets a day (nVenue; Seeley). Genius Sports describes the same loop from the other side: “If a receiver makes a 15-yard catch, we instantly adjust player prop bets” (Lindenberg).

Set the two clocks side by side. The price moves in a fraction of a second, on data captured in the stadium. The customer’s screen shows the catch nineteen to sixty seconds later. Everything else in this article follows from that arithmetic.

The Clock the Bettor Cannot See

The delays are not a secret, but they are rarely stated in seconds where a bettor would read them. Table 2.3.1 assembles the two Super Bowl measurements that have been made public.

Table 2.3.1Measured broadcast and streaming delay behind the field, Super Bowls LIX and LX

Platform	Super Bowl LIX, Feb. 2025 (seconds)	Super Bowl LX, Feb. 2026 (seconds)
Over-the-air broadcast	16 to 21 (avg. 17)	~19
Cable	32 to 60 (avg. 43)	38
Peacock	not measured	48
Hulu + Live TV	62	53
YouTube TV	67	53
DirecTV Stream	62	60
NFL+	51	62
Fubo	78	not reported

Source: Phenix Real Time Solutions study of Super Bowl LIX (Phenix; Dachman); Stats Perform and Phenix measurement of Super Bowl LX (Keys; Fletcher, “NBCU’s Peacock”). Phenix sells low-latency streaming and is an interested party; its figures are consistent with independent measurements by the analyst Dan Rayburn (Rayburn; Fletcher, “NBCU’s Peacock”).

Streaming is the direction the audience is moving, and streaming is slower. Amazon, which carries Thursday Night Football, defines its own target as ten seconds from camera to screen and claims to be faster than broadcast (Nickinson), but the FanDuel executive quoted above was blunt about the general case: “Video streaming is 30+ seconds behind traditional television, which makes live betting much harder to do” (Mazzeo). A Fanatics trader put it more plainly still in 2025, describing television latency as leaving “the image on their screen upwards of 30 seconds behind the data feed on their betting app” (James).

Now the other clock. The data that reaches the sportsbook is not scraped from television. It is captured in the venue, by the league’s own systems and by the data company that has bought the exclusive right to carry it: Next Gen Stats chip data and official play-by-play in the NFL, distributed by Genius Sports; Statcast and official scoring in baseball, distributed by Sportradar; tracking data in the NBA and NHL, distributed by Sportradar (Genius Sports, “National Football League Expands”; Sportradar, “Major League Baseball”; NHL Public Relations; Bassam). Sportradar has gone so far as to publish a latency scale for its feeds, grading them from “Low” at under four seconds to “Exceptional” at sixteen and above, with a worked example of a match that “historically delivered at ~5 seconds” (Sportradar, “Latency Indicator”). A sportsbook, in other words, can read on a dashboard exactly how far ahead of its customers it is.

This is the first thing the phrase “live betting” conceals. The bettor is invited to believe he is watching the same game the book is pricing. He is watching a recording of it, and the recording is the slowest instrument in the building.

The Delay Window

A gap of thirty seconds between the book’s information and the bettor’s would be a catastrophe for the book if the bettor could act on it in the other direction. He cannot, and the house rules explain why.

Every major American operator reserves the right to hold a live bet before accepting it. DraftKings: “All bets placed through any DraftKings platform … may be subject to a time delay prior to acceptance, the length of which may vary. Such delay is to be determined by DraftKings in its sole discretion” (DraftKings, “Betting Rules”). Caesars: “For the purposes of security and integrity, Live wagers (meaning events that have already started) are subject to an automated time delay prior to acceptance” (Caesars). Fanatics: “for operational reasons, bets made in-play may be subjected to a delay” (Fanatics). FanDuel: “In-play betting markets may be subject to delays in acceptance and bet requests may not be processed immediately” (FanDuel).

The mechanics of the window are visible in a patent FanDuel’s Edinburgh subsidiary was granted on December 17, 2024, titled “Dynamic wager delays for live contest betting.” The problem it sets out to solve is stated in the patent’s own words: “a user who is in person at the contest while also betting on the contest can place bets based on events occurring in the contest but before the event is communicated to the bookmaker such that the bookmaker can update their offerings appropriately.” The solution is a delay that changes length throughout the game, shortening during timeouts and halftime and lengthening when the ball is in play (“Dynamic Wager Delays”). The delay exists to make sure that no bet is accepted until the book has caught up with the field. It is the book’s insurance against being the slow party. The bettor, who is always the slow party, has no equivalent.

What the window produces is a queue in which the house decides, after the fact, which bets it wants. DraftKings reported to investors that its “live bet success rate” reached 92 percent in 2025, up from 78 percent in 2022, a metric the presentation does not define (DraftKings, “2026 Investor Day”). Read on its face, and the other way round, roughly one live bet in twelve is still refused after the customer has tried to place it, and three years earlier it was more than one in five. The rules describe what happens to the ones that get through at the wrong price. FanDuel “reserves the right to void any bets placed after the outcome of the relevant event or occurrence is known or reasonably knowable” (FanDuel). DraftKings may cancel live bets “effected at incorrect price due to delayed or failing ‘Live’ coverage” (DraftKings, “Betting Rules”). bet365 settles a late bet “at the revised price at the time the bet was placed” and voids anything “where the outcome is already known” (bet365).

Note what is symmetrical here and what is not. When the bettor is behind, which is always, the house accepts his bet at a price it has already updated. When the house is behind, which is rare, the house voids. The one recorded instance of a regulator objecting is instructive precisely because it concerns the book being slow. In September 2024, New Jersey fined bet365 $33,000 for accepting more than $257,000 in wagers on events whose outcomes were already known and then re-pricing them under an “obvious error” clause without the regulator’s approval (Parry). The state’s guidance on such “past-posted” bets requires the book to void them and file an incident report (New Jersey Division of Gaming Enforcement). When DraftKings left Lakers–Nuggets player props at first-quarter prices for thirteen minutes in January 2024, it voided the winning bets in most states and paid them only where regulators insisted; its head of trading told ESPN, “there’s going to be more errors in the future and one of them might financially cripple our company” (Purdum). The system is built so that the book’s errors are reversible and the bettor’s lateness is not.

The house rules also tell the bettor, in a clause almost nobody reads, exactly what this article argues. FanDuel: “Customers should be aware that ‘live’ broadcasts, streams, or data feeds may be delayed, interrupted, or inaccurate” (FanDuel). DraftKings: “There may be a delay in the updating or displaying of information on the DraftKings platforms related to an ongoing game or event giving individuals with more up-to-date information an advantage” (DraftKings, “Betting Rules”). Fanatics: “transmissions described as ‘live’ by some broadcasters may actually be delayed or pre-recorded” (Fanatics). Britain’s Gambling Commission requires operators to say this on the in-play page itself; its technical standard RTS 15A reads, “Information must be made available that explains that ‘live’ TV or other broadcasts are delayed and that others may have more up-to-date information” (Gambling Commission). No American regulator has an equivalent rule. The disclosure lives in the terms of service, where it protects the book, rather than on the betting screen, where it would inform the customer.

The Only Edge That Was Ever Real

There is a control group for the claim that in-play betting rewards watching closely, i.e., the people who watched more closely than anyone, from a seat inside the venue, and transmitted what they saw before the umpire’s tablet had recorded it. They were called courtsiders, and for a few years they were the only in-play bettors on earth with a documented edge.

The practice surfaced publicly at the 2014 Australian Open, when Victoria Police arrested a twenty-two-year-old Englishman named Daniel Dobson and charged him under a new state law against conduct that could corrupt a betting outcome. Prosecutors alleged he had used a device concealed in his shorts, wired to a phone, to send point-by-point results to a betting syndicate in Surrey (Bennett; Farnsworth). His employer, Sporting Data, had been founded by former Betfair employees and, as reported at the time, maintained that nothing about the practice was illegal (“Sporting DataGate”). Six weeks later the charge was withdrawn for want of any reasonable prospect of conviction, and a police spokesman warned that the decision “should not be seen as an invitation” to do it again (Farnsworth; “Charge Dropped”). In the United States the enforcement tool has been the trespass notice, with nineteen suspected courtsiders removed from the 2016 US Open and banned for twenty years, and in 2017 an Estonian was arrested at Arthur Ashe Stadium for violating a stay-away order (“Behind the Shadowy Practice”). In 2020 the tennis integrity authority sanctioned a player for the first time for courtsiding, defining it as “transmission of live scoring data from a match to a third party for betting purposes” (International Tennis Integrity Agency).

What the courtsider had was not knowledge of tennis. It was time. The Associated Press explained the trade in one sentence: “Getting the results first, even by a few seconds, can give bettors an edge over others looking at odds based on the previous point” (“Behind the Shadowy Practice”). The natural home for the trade was the betting exchange, where a courtsider was betting against other customers rather than against a bookmaker, and the exchanges responded with the one tool available to them: a delay. Betfair’s in-play delays run from about one second to as many as nine depending on the sport, and the exchange has described their purpose as overcoming “the transmission lag inherent in broadcasting sports events” (Webb). When Betfair cut its Premier League delay from eight seconds to five in February 2014, it explained that it had been “testing a new radio feed which allows us to get very close to ‘live’” (“Betfair To Reduce”). The exchange had to buy speed to protect its customers from people who had bought it first.

The academic literature on exchange prices during play confirms that the seconds mattered. Alasdair Brown’s study of Betfair’s Wimbledon markets, published in the Economic Journal in 2014, found mispricing between the match-winner and set-score markets roughly ten times larger during play than before it, with an average discrepancy of 5.3 percent, and showed that the win market moved first and the set market followed: traders under time pressure were processing information sequentially, and a faster trader could stand between them (Brown). Croxson and Reade, in the same journal, found that Betfair’s football prices absorbed goals “swiftly and fully” (Croxson and Reade), while Choi and Hui and later Angelini, De Angelis and Singleton found systematic under-reaction to expected goals and over-reaction to surprising ones (Choi and Hui; Angelini et al.). Whichever side of that argument one takes, the point for the bettor is the same. Where any in-play mispricing has been measured, it has been measured on an exchange, in the seconds after an event, by researchers with a data feed. It has never been measured against a bookmaker’s price, and it was never available to anyone watching television.

Courtsiding was, in short, a proof of concept. It demonstrated that the in-play market could be beaten by exactly one thing, which was seeing the game before the market did. The industry’s response was not to close the gap. It was to buy it.

The Sale of the Clock

Between 2019 and 2025 every major American league sold the exclusive right to distribute its official, in-venue game data to sportsbooks, and took equity in the buyer.

The NFL granted Genius Sports exclusivity in April 2021 for warrants amounting to roughly a five percent stake, in a deal reported at up to $1 billion over four years and extended twice since, most recently through the 2029 season (Rybaltowski, “DraftKings-Genius’ Deal”; Genius Sports, “National Football League Expands”). The NBA’s eight-year agreement with Sportradar, effective from the 2023-24 season, was reported at more than $1 billion in cash and stock and gave the league three percent of the company (Bassam). Major League Baseball extended Sportradar’s exclusivity through 2032 in February 2025, in an agreement the company described as delivering “ultra-low latency official data” and paying the league in annual fees and up to 1.86 million shares (Sportradar, “Major League Baseball”). The NHL signed a ten-year deal in 2021 (NHL Public Relations). The NCAA extended Genius through 2032 in April 2025 (Genius Sports, “NCAA and Genius Sports”). An NBA gaming executive had explained the logic as early as 2020: “The U.S. sports betting industry has recognized the value of official data, particularly for in-play betting” (Novy-Williams).

What the sportsbooks bought was not integrity, whatever the press releases said. It was the fast clock. When Genius Sports and Sportradar litigated over British football data in 2022, the settlement gave Sportradar a sublicence to a “delayed feed to be marketed as the Official FDC Secondary Feed,” a product its own lawyers described as carrying “greater latency (i.e. a longer delay) compared to the primary feed” (Genius Sports, “Statement”; Bond et al.). Speed had become a tiered product with a list price. In the NFL’s case the price was reported at about four percent of a book’s gross revenue on pre-game bets and about six percent on in-game bets, several times the previous rate (Rybaltowski, “DraftKings-Genius’ Deal”). The in-play surcharge is the market telling you where the value of the feed lies.

Then the law made the purchase compulsory. Tennessee, the first state to do so, wrote into its statute that “a licensee shall exclusively use official league data for purposes of live betting” unless the league cannot supply it on commercially reasonable terms (Tenn. Code Ann. § 4-51-316). Virginia defines a “tier 2 bet” as one “placed after the event it concerns has started” and requires official data to settle it (Va. Code §§ 58.1-4030, 58.1-4036). Michigan, Arizona, Massachusetts and North Carolina have equivalent provisions (Mich. Comp. Laws § 432.410a; Ariz. Rev. Stat. § 5-1314; Mass. Gen. Laws c. 23N, § 12; N.C. Gen. Stat. § 18C-914). Meanwhile, the ticket in the fan’s pocket forbids him from owning a clock of his own. The 2026-27 terms for Levi’s Stadium, the building where Stats Perform timed the delay, provide that the ticket holder “agrees not to create, transmit, display, distribute, exploit, misappropriate or sell … in any form, any description or account (whether text, data or visual, and including, without limitation, play-by-play data) of the Event” for any commercial purpose, and that the ticket may not be used for gambling without the league’s written consent (San Francisco 49ers).

Put the pieces in order. Courtsiding proved that speed was the edge. The exchanges imposed delays to blunt it. The leagues sold exclusive speed to the sportsbooks and took equity in the sellers. The states required the sportsbooks to buy it. The venues wrote the prohibition into the ticket. Every actor in the chain has now been paid for the clock except the person the product is sold to, who has been sold a picture instead.

What the Exchange Says Out Loud

The prediction markets that article 1.4 examined have made the structure unusually visible, because an exchange cannot hide the delay inside a house rule. It has to write it into the order book.

In December 2025 Crypto.com disclosed to the CFTC that it had placed a three-second delay on every sports order from ordinary users while its professional market makers traded with none, and Kalshi filed for authority to impose order delays of its own (Hamid; “Crypto.com Implements”). Kalshi subsequently signed Sportradar to supply official data and live odds for its pre-game and in-play markets, and Genius Sports for its soccer contracts, with its chief executive explaining that official data would “ensure quicker trade settlements” (Sportradar, “Sportradar and Kalshi”; Genius Sports, “Genius Sports and Kalshi”). The exchange, in other words, arrived at the sportsbook’s architecture within a year: a fast feed for the house side of the book, a speed bump for the customer side. The difference is only that the exchange has to say so in a regulatory filing.

The Wall Street Journal’s account of Kalshi’s first NFL season described a trader who bought a contract during a game and “within 45 seconds” sold it for a $500 profit, and observed that sportsbooks, by contrast, “can suspend wagering or halt bettors’ ability to cash out of winning bets,” which the paper noted “often happens precisely at game-changing moments” (Osipovich and Fernandez). That is the honest version of the difference. An exchange lets you out; a sportsbook decides whether to let you in. Neither lets you be first.

The Price of Going Fast

There is one thing this article cannot tell you, and it should say so plainly. No American operator discloses its hold on live bets separately from pre-game bets, no state regulator breaks in-play handle or revenue out of its monthly reports, and no peer-reviewed study has measured the margin on a US sportsbook’s in-play prices. Assertions that live margins run at some multiple of pre-match margins circulate freely in the trade press and rest on nothing that can be checked. The reader should treat them accordingly, and should treat this article’s silence on the figure as deliberate.

What is on the record is narrower. The chief executive of DraftKings has said of micro-bets that “you don’t want to hold too high on those” (DraftKings, “Q1 2025 Earnings Call”), which is a statement about turnover, not generosity, i.e., a product that is bet thirty-six times a game can carry a thin margin per bet and still take more from the customer than a fat margin taken once. The compounding is arithmetic, not psychology. A bettor who stakes $50 on each of thirty-six micro-bets has put $1,800 through the window on a single Sunday afternoon, and at the 10.16 percent the American industry actually held across all products in 2025 (American Gaming Association) that afternoon costs him about $183 in expectation. The same $50 placed once before kickoff costs him about $5 at the same rate, and about $2.30 at the standard minus 110 price. Hegarty and Whelan’s study of 84,230 European matches found that even at closing pre-match prices, bettors’ realized losses ran about a fifth higher than the posted overround implied, because margins are loaded onto the longer-priced outcomes (Hegarty and Whelan). Whether that loading is steeper during play is unknown, for the reason stated above. What is known is that the live product is structured to be bet often, and that its parlay share is rising: Flutter reported live parlays at 22 percent of FanDuel’s live handle in the first half of 2024 and identified the gap as a “structural hold growth opportunity” (Flutter).

This is the second thing the phrase conceals. The live bettor is not paying a price. He is paying a rate.

Who Bets This Way, and What It Costs Them

The behavioral record is consistent, and it is important to describe it accurately, because the previous article in this series showed how easily the gambling literature gets overstated.

The single most cited account-level finding is LaPlante, Nelson and Gray’s 2014 analysis of 1,440 bwin customers who had completed a clinical screen. After controlling for how many kinds of gambling a customer did, “only Live Action Internet sports betting retained a significant relationship” with a positive screen for gambling disorder (LaPlante et al.). Among 1,813 Australian sports bettors surveyed by Russell and colleagues, 78 percent of those who bet on micro-events met the criteria for problem gambling, against 29 percent of those who did not (Russell et al.). A 2023 Ontario survey of 920 sports bettors found the 223 who bet in-play reported higher problem-gambling severity and more harm in every measured domain than either single-event or traditional bettors, who did not differ from each other (Vieira et al.). A 2026 case-control study found that higher-risk gamblers reported more frequent in-play betting and more use of instant cash-out, with no differences for parlays or props (Farrell et al.). Hing and colleagues found that impulse bets accounted for nearly half of all Australian sports bets and that the most impulsive bettors “favoured betting on in-match contingencies instead of overall match outcomes” (Hing et al., “On the Spur”).

Two honest qualifications belong beside those findings. Every one of them is cross-sectional or observational; none demonstrates that in-play betting causes disordered gambling rather than attracting people already disposed to it. And Australia, the one large jurisdiction that has prohibited online in-play betting since 2001 and closed the “click to call” loophole in 2017, has never formally evaluated whether the ban reduced harm (Interactive Gambling Amendment Act 2017; Podesta and Thomas). The best available evidence under the ban is a 2020 survey in which 34.4 percent of Australian online sports bettors had still bet in-play in the past month, by telephone or offshore, and scored a mean of 8.76 on the problem-gambling index against 3.68 for those who had not (Gainsbury et al.). The prohibition is real; Australia’s regulator found four of the country’s largest bookmakers in breach of it as recently as December 2023 (Australian Communications and Media Authority). Its effect is unmeasured.

The United States restricts none of this for professional sport. The one federal proposal that would, the SAFE Bet Act, would forbid any wager “on any sporting event, once such event has commenced,” and would bar operators from using artificial intelligence to “create a gambling product, such as a microbet” (S. 1033 § 103(b)). It was introduced in March 2025 and has not moved. The only limit actually in force arrived by a different route. On November 9, 2025, federal prosecutors in Brooklyn charged two Cleveland Guardians pitchers, Emmanuel Clase and Luis Ortiz, with rigging individual pitches, alleging that bettors won at least $450,000 on wagers such as whether a given pitch would be a ball or would exceed 94.95 miles per hour (United States Attorney’s Office, Eastern District of New York). The next day Major League Baseball and sportsbooks representing more than 98 percent of the American market announced a $200 cap on pitch-level wagers and their exclusion from parlays (Castrovince). Both players have pleaded not guilty; the case was pending as this article went to press.

The Clase indictment belongs in this article for a reason beyond its news value. A single pitch is the cheapest thing in sport to fix, precisely because the micro-market is priced by a machine that trusts the feed. The book’s speed advantage over the bettor is total; its advantage over the pitcher is nil. The league, which sold the feed, imposed the cap. The regulators did not.

The Mirror Test

The Live Bet

VerdictIllusion

The Mirror Test is this series’ standing verdict: every strategy, product and claim is judged on whether it creates a real edge or only the appearance of one. The verdicts are EDGE, ILLUSION and CONDITIONAL.

The claim under test is the one the product is built to produce, essentially that betting during the game rewards watching it closely, and that a bettor who reads the game well can beat a price that is being set in real time. The ruling is that the claim confuses the picture with the market. The bettor’s information arrives on a broadcast that runs nineteen to sixty seconds behind the field. The price he is offered was set from in-venue data delivered in under a second, by software that repriced the play before his screen showed it, and his bet is then held in a delay window whose only purpose is to make sure the house is never the slower party. Watching closely is not an edge. It is the experience the product sells while the market is settled elsewhere.

The case against

The strongest case against this verdict is that in-play markets have been beaten, and the evidence is on the record. Courtsiders made money for years. The Economic Journal literature shows mispricing during play an order of magnitude larger than before it. Kalshi lets a customer out of a position in forty-five seconds where a sportsbook would suspend the market. If the market is inefficient in the seconds after a play, then the claim is not an illusion at all; it is a real edge that most people are too slow to take.

Why the verdict holds

The verdict survives because every documented instance of that edge belonged to speed rather than to judgment, and speed has since been bought, licensed and prohibited. The courtsider’s advantage was seeing the point before the umpire’s tablet; it ended with trespass notices and a delay on the exchange. The mispricing in the academic record was measured on exchange data, in the seconds after a goal, against other customers, never against a bookmaker’s price. The clock the courtsider used is now sold by the leagues to the books under exclusive contract, required of the books by statute, and forbidden to the fan by the terms printed on his ticket. What remains for the bettor is the one instrument in the chain that nobody has paid to make faster: his television. The house does not need to know more about the game than he does. It only needs to know it first, and it has arranged, at every link, to be paid for the privilege.

ILLUSION The Live Bet

Conclusion: Live Is Not Live

The House of Mirrors has planted three manufactured numbers in this series so far: (i) an illegal-market estimate that traced to a single guess, (ii) a prediction-market volume reported as if it were handle, and (iii) a dopamine story borrowed from cocaine research. The fourth is not a number. It is a word. “Live” describes a broadcast that reached the customer after the play had been priced, the bet had been held, and the market had, in many cases, already been settled. The word is accurate about the picture and false about the bet, and the entire in-play product depends on the customer not noticing the difference.

Nothing in this article requires the reader to believe the sportsbook is cheating. The house rules are explicit; the delays are disclosed in the terms; the data contracts are public; the patent is on file; the statutes mandating the fast feed were passed in daylight. The bettor has been told, in language he is unlikely to have read, that others may have more up-to-date information. What he has not been told is that the gap is not a flaw in the product but its foundation, and that the only people who ever closed it were escorted from the building.

Article 2.1 argued that the house’s durable edge is not its understanding of your brain but the price it publishes. This article narrows the point. In the live market the house’s edge is not even the price. It is the clock. The next article, on the media companies that put the broadcast and the sportsbook on the same screen, examines what happens when the people who control the delay also own the microphone.

Sources

This is 2.3 of House of Mirrors: A Bettor’s Guide to the Illusions of Modern Sports Gambling, a serialized investigation of the American sports betting industry. Next up is article 2.4, The Media-Betting Complex.