The Automated Ball-Strike challenge system did not replace the umpire. It sat a robot beside him. The human still calls every pitch; but the batter, catcher, or pitcher can tap their helmet and ask Hawk-Eye to settle it. Each team gets two challenges. Win one and you keep it; lose one and it is gone.
Across 643 games and 190,813 pitches (March 25 to May 13, 2026), players initiated 2,629 challenges, and about 53% were overturned. That number alone is the whole tension of the system: a coin flip with the game on the line. The rest of this is about turning that coin flip into a decision.
1 · The bottom-edge problem
Umpires get roughly 6.6% of called pitches wrong versus the ABS zone. The misses are not random — they cluster on the edges, and the bottom edge most of all. A low strike at the knees, dropping through the zone, is the hardest call in baseball, and the heatmaps make that obvious. This is where challenges live.
2 · A challenge is geometry
The single most important variable is distance from the zone edge. Push a pitch two inches off the plate and a missed call is a near-certain overturn. Bring it to the edge and the overturn rate collapses toward a coin flip. Everything downstream — who should challenge, and when — is a negotiation with that curve.
3 · The catcher’s eye is real, and measurable
Catchers are the best challengers (about 60% success), ahead of batters (~46%) and pitchers (~37%). But aggregate rates hide the skill. We fit each catcher a perception sharpness — a sigma, in inches — describing how close to the edge they can still tell a ball from a strike. The sharpest, like Will Smith (~76% overturn), are not luckier; they simply see the margin better.
4 · Winning teams decide differently
Net challenge value — runs gained on overturns minus the cost of wasted challenges — spans nearly a full win between the best and worst teams. The gap is rarely about eyesight. It is allocation: weak teams burn an early challenge on a low-stakes call and have nothing left when a game-deciding pitch is missed in the eighth.
5 · The cost of a challenge is not a flat number
The popular shortcut treats a lost challenge as a flat ~0.2-run cost. It is not. Losing your first challenge in the first inning is expensive — you forfeit eight innings of optionality — worth closer to 0.5 runs. Losing your second is cheap. So the confidence you need to justify a challenge should fall as the game goes on, from roughly 47% in the first inning to ~21% in the ninth. Conserve early, spend late.
6 · The leverage twist
Stakes bend the threshold further. Take the same pitch — bases loaded, 3-2, two outs, tied. In the top of the first, the break-even confidence is about 12%: the game is long, hold your shot. In the bottom of the ninth, it is under 1%: challenge almost anything. And counter-intuitively, a sharper-eyed catcher carries a higher bar — because their challenges are more valuable, a wasted one costs more.
The prize: ~60 wins on the table
Add up every umpire miss that went unchallenged and had positive expected value, across the league, and it comes to roughly 596 runs — about 60 wins — sitting on the field, unclaimed. Not from better eyesight, but from better detection and timing. That is the opportunity this whole project is built to quantify.
Want the receipts? Every claim here is something you can poke at in the explorers, and the model is laid out in the methodology.