Colorado Bought Its ABS Edge in Bulk

Colorado’s hitters won 75 of 178 challenges, 42%. Hitters league-wide won 49%. Colorado’s catchers and pitchers won 134 of 208, 64%, against 58% for everyone else. The club still finished +7.2 in Runs Above Average, second only to the Minnesota Twins (+8.7). That doesn’t add up on its face, and I wanted to know why.

ABS is new to the majors in 2026. A hitter, catcher or pitcher can challenge a called ball or strike, and an automated zone makes the final ruling. The zone is 17 inches wide and runs from 27% to 53.5% of the batter’s height, judged at the middle of the plate, and any part of the ball touching it counts. Each club gets two challenges and keeps one each time it wins. To find out which of Colorado’s two faces is the real one, I went back through every called pitch and every challenge of the season and charted where each one crossed the plate.

Short on time? Read the one-page brief (PDF), written like an internal research summary. It is not a Rockies document.

To separate the two stories, I split each club’s gain into two parts. Volume is what a club earned by challenging more (or less) often than an average club would on the same pitches. Selection is the rest: how valuable each challenge was compared with an average one. To define “average,” I built a model of how often a typical club challenges a pitch, given where it crossed the plate, the count, the situation and the stakes. It held up on games it hadn’t seen (details are in the methodology).

Colorado’s hitters challenged 178 times where an average club would have challenged about 130. That volume was worth +5.2 runs (+2.6 to +8.3), the most in MLB. Their selection cost −2.8 (−6.1 to +0.5), 29th of 30. The catchers and pitchers did both: +2.6 from volume and +3.5 from selection (+0.1 to +7.0), fourth in MLB.

My own version of the measure puts Colorado third rather than second. It leaves out Savant’s penalty for failed challenges and uses my own definition of an average club, so the rankings shouldn’t match exactly. The pieces I rebuilt track Savant’s published numbers almost perfectly (a correlation of about 0.99 across the 30 clubs for expected challenges, runs gained and expected runs gained), so I’m describing the same thing Savant scores.

Three things stood out, and one of them is a caution. First, the bottom edge. Hitters’ challenges nearest the bottom of the zone succeeded 25 of 73 times, 34%, against 49% for other clubs’ hitters (−2.5 standard deviations from the league rate). Colorado’s catchers and pitchers, by contrast, did unusually well at the same edge: 52 of 73, 71% against 59%. I looked at six combinations of role and edge, so one at 2.5 deviations is suggestive, not established. As a check I looked at 2026 spring training, where Colorado hitters went 4 for 9 at the bottom edge to others’ 2 for 6. That’s far too small a sample to confirm or refute anything.

Second, where hitters place their bets. Colorado’s hitters challenged more called strikes that were correct by two to three inches (13% of their challenges against 8%) and fewer that were clearly balls by one to two inches (10% against 16%). That’s where the weaker pitch selection comes from.

Third, a caution. 683 of the calls against Colorado were wrong and challengeable (the club had a challenge in hand, ABS was running and no position player was pitching), which exactly matches Savant’s count. 682 of them have full game context, and those are the ones the numbers below use. Colorado overturned 209 of them, 31%, against 25% league-wide. That leaves 473 on the table, which makes for a tempting headline. But 318 of them (67%) were within an inch of the line, a distance no player can judge from the box or from behind the plate. Only 42 were two inches or more off, and not one of those was worth 0.30 runs or more. Savant’s unrealized-value number is real, but most of it was out of anyone’s reach.

The cost side is less obvious. Winning a challenge costs nothing, but losing one costs you a future option. Savant’s documentation uses a rule of thumb of 0.2 runs. I estimated it from the data instead, by comparing how clubs fared in games where they started an inning holding two challenges versus one (or one versus none). That works out to about 0.047 runs per lost challenge, or 0.043 using only the 4,331 of 4,754 lost challenges with enough comparable games. For first-inning losses, which have nothing to compare against, I assumed one challenge is worth 60% of two. This is a comparison, not proof of cause and effect: clubs that kept their challenges may differ in skill or opportunity. The average club ends a game with 1.0 challenges unspent (Colorado: 0.9), which is why the number comes out small. At a cost nearer 0.2, failed challenges would matter far more, so everything below depends on this estimate. Using it, the best “skip the low-stakes ones” rule I could find for hitters, tuned on the other 29 clubs, added 0.03 runs per club in the sample and 0.05 runs for Colorado’s hitters (range −0.17 to +0.28). That’s close to nothing.

The lab below shows the break-even for any situation: how sure a player has to be before challenging pays off. It can’t tell you how sure he actually is, and that gap is the whole problem.