Andrew Armstrong

Run Value/100 replicates MLB's own published methodology: a count-conditioned run-expectancy model (base-out state × ball-strike count, same RE24 approach as the baserunning model but extended to the pitch level) values every single pitch, not just PA outcomes — a called strike moving the count from 1-1 to 1-2 has real value even though nothing "happened." Doesn't use Trackman at all, just play-by-play + pitch sequencing, so it isn't restricted to Trackman-covered games the way most of this section is. The Fielding row is a range-runs estimate: batted balls are bucketed by position + Trackman-derived difficulty (hang time/distance/exit velo, same buckets as the baserunning model), and the fielder — parsed directly from that specific play's PBP description, not a per-game starting lineup — is credited or debited the run-value gap between the bucket's expected out rate and what actually happened. Barrel% uses MLB's actual published EV/launch-angle lookup table — a ball must be hit 98+ mph, with the qualifying launch-angle window widening as EV climbs (a tight 26-30° band right at 98 mph, up to a wide 8-50° band at 116+ mph); Solid Contact % is a CCBL-only complement using that same table shifted 10 mph lower (88+ mph and up), since MLB's absolute 98+ mph Barrel threshold leaves ~70% of CCBL hitters flat at 0%. Hard-Hit % is exit velo alone, no launch-angle requirement: 95+ mph off the bat. xBA/xSLG/xwOBA use a k-nearest-neighbor model on this season's own batted balls, not MLB's own published model or league-wide MLB data — those, Fielding, and Run Value should be read as CCBL-relative, not compared directly to MLB's own advanced tracking numbers.