A Misscored Walk-Off Double Just Cost Prediction Market Bonders Millions
At approximately 1:15 am Eastern on Saturday morning, Rafael Devers hit a ground-rule double with the bases loaded in the bottom of the tenth inning.
The knock scored the Giants’ automatic runner from second, ending a 7-6 San Francisco win over the Angels. For most people who were still awake to see it, it was a satisfying walk-off in a close game.
For a group of prediction market traders running automated bonding bots on Kalshi and Polymarket, it was the beginning of one of the most chaotic market events the industry has seen involving professional sports.
Obscure MLB Rule Leads to Complete Chaos
The mechanics that caused the chaos are buried in MLB’s official rulebook. Under rule 5.06(b)(3)(D), a walk-off ground-rule double is not treated as a standard double. When a batter hits a ball into dead-ball territory with runners on base in a walk-off situation, only enough runners score to end the game. The remaining baserunners are not awarded two bases. The play is effectively re-scored as whatever outcome produces exactly the winning run. In this case, with the automatic runner on second, Devers’ ground-rule double scored exactly one run for a 7-6 final, not two runs for an 8-6 final.
The initial official MLB scorer, apparently unaware of or misapplying this rule in the moment, recorded the play as a standard two-run ground-rule double. The official score initially displayed as 8-6. Major sports data feeds, including Fox Sports and ESPN, picked up and propagated that 8-6 score. Prediction market bots, which trade algorithmically against final score data feeds, saw 8-6 and bought the “sure bet” accordingly.
Bots Turned a Scorer’s Error Into a Multi-Million Dollar ‘Rinsing’
Bonding bots are automated market-making programs designed to profit from pricing inefficiencies on prediction market exchanges. They work by offering to buy or sell contracts at prices that imply a statistical edge, operating continuously and at volume too high for any human trader to replicate manually. When a game ends, these bots bond markets to their settlement prices at close to 99 cents for contracts that will resolve YES and close to 1 cent for contracts that will resolve NO, locking in profit from the spread.
With the scoreboard showing 8-6, the bots bonded three markets to their apparent settlement prices: over 13.5 total runs at 99 cents, Giants team total over 7.5 runs at 99 cents, and Giants -1.5 run line at 99 cents. Under an 8-6 final score, all three settled YES. The bots bought aggressively, and prices snapped to 99 cents.
Human traders who understood the walkoff scoring rule more quickly than the data feeds corrected themselves and spotted the opportunity. According to DataBasedBets, insane action came in to unbond the above markets almost immediately, with prices going haywire and touching 1 cent and 99 cents multiple times as the market wrestled with the competing data. The chaos lasted until the feeds were corrected.
Prediction Markets’ Inconsistent Settlements in the Same Game
What happened next is the most analytically interesting part of the story, and the part most directly relevant to how prediction markets operate as regulated exchanges.
Kalshi settled the totals and team totals markets while at least one official score source still showed 8-6. Under the data it had at settlement time, over 13.5 runs and Giants team total over 7.5 both settled YES, rinsing the traders who had sold those contracts expecting the score to correct. Polymarket US settled the Giants -1.5 run line the same way, also at an 8-6 implied final.
The run line market on Kalshi, however, remained open longer. During that window, the official score sources corrected to 7-6. Market volume on the spread ballooned from approximately 11,000 contracts at the start of the bottom of the tenth inning to five million contracts by final settlement, as traders piled in to take the now-obvious Angels +1.5 position. Kalshi settled the spread at 7-6, making Angels +1.5 a winner. Polymarket International also settled Angels +1.5 as a winner.
The result was a settlement contradiction within the same game: Kalshi settled the totals as though the final score was 8-6, then settled the run line as though the final score was 7-6. Traders who held Giants -1.5 positions expecting consistency with the totals settlement were rinsed. Traders who correctly anticipated the run line would settle on the corrected score collected.
DataBasedBets estimated several million dollars in total losses across exchanges for traders who assumed the 8-6 final score, with Kalshi volume on the incorrectly settled markets alone exceeding 1.1 million on the over-under 13.5 and approximately 60,000 on the Giants team total.
Sportsbooks Have More Experience and Settled More Cleanly
Licensed sportsbooks handled the situation differently and, from a consumer protection standpoint, more cleanly. Sportsbooks appear to have initially paid out based on the 8-6 score, but then resettled all affected markets based on the corrected 7-6 final once the official score correction came through. That resettlement mechanism, standard practice in regulated sports betting, allowed sportsbooks to correct an error after the fact.
Prediction market exchanges do not work the same way. As CFTC-designated contract markets, they operate under rules that govern financial derivatives rather than sports wagers. Those rules prioritize the finality and certainty of settled contracts, because the integrity of a derivatives market depends on settled trades being settled. A futures exchange that routinely reopens and resettles contracts after the fact would undermine the basic confidence that makes the market functional. Chairman Selig made exactly this argument when he blocked Michigan from voiding executed Kalshi trades last week.
The CFTC’s logic, that settled contracts must be honored, produces a clean result in most circumstances and a messy one here. Kalshi settled two contracts based on data it had at settlement time, data that was subsequently corrected by the official scorer. It then settled a third contract based on the corrected data. The settlements were each technically defensible under the exchange’s CFTC rules at the moment they were made. They were also inconsistent with each other.
Refunds Are Unlikely, And Bots Have No Way to Adjust in Real-Time
The Giants-Angels incident is an extreme version of a structural vulnerability that prediction markets share with any market that settles against real-time data feeds. A scorer’s error, a data transmission delay, or a feed discrepancy between sources: any of these can create temporary price divergences that automated bots will exploit before the underlying data is corrected.
Sportsbooks have settlement processes designed around human review, giving operators time to catch and correct data errors before final payouts are made. Prediction market exchanges, operating as automated derivatives markets with settlement rules tied to data sources rather than manual review, are faster but less forgiving. When the data is wrong, the settlement can be wrong, and unwinding a settled derivatives contract is structurally different from resettling a sports bet.
Whether Kalshi will issue any refunds or adjustments for the incorrectly settled total market is unclear. DataBasedBets, who documented the events in real time, said they would not expect a refund since the settlements appear to have complied with the exchange’s CFTC rules. That may be the legally correct answer. It is also the answer that crystallizes the difference between a sportsbook’s customer-facing settlement practices and a derivatives exchange’s regulatory settlement obligations.
For the bonders whose bots were running on autopilot at 1 am Eastern on a Friday night, the distinction carries a seven-figure price tag.
Players trust our reporting due to our commitment to unbiased and professional evaluations of the iGaming sector. We track hundreds of platforms and industry updates daily to ensure our news feed and leaderboards reflect the most recent market shifts. With nearly two decades of experience within iGaming, our team provides a wealth of expert knowledge. This long-standing expertise enables us to deliver thorough, reliable news and guidance to our readers.
Latest News
