Stockyard grades use machine learning to predict the expected run value (xRV) of each pitch based on its physical characteristics. Pitches that prevent more runs grade higher. Because xRV captures strikeouts, walks, and contact quality in a single metric, the grades reflect total pitch effectiveness — not just whiffs.
Across all pitch types, tunnel distance — how similar a pitch looks to other pitches at the batter's decision point (~23 feet from the plate) — is consistently the highest-weighted feature. Pitches that are indistinguishable from the fastball until the last moment force late adjustments and generate both whiffs and weak contact. This is why the per-pitch descriptions below emphasize what makes each pitch type tunnel differently, rather than restating tunneling's universal importance.
Standard deviation measures how spread out values are from the average. If you lined up every pitch thrown in a season by quality, most would cluster around average (100). A few standout pitches land far from the pack. One standard deviation (15 points) above average means the pitch is better than ~84% of all pitches — only about 1 in 6 are that good or better. Two standard deviations (130) means top ~2% — truly elite. This 'bell curve' shape is how natural variation works: most pitches are ordinary, and extreme quality is rare in both directions.
100 = league average. Every 15 points represents one standard deviation. A grade of 115 means the pitch is in the top ~16% of all pitches.
Fastballs are graded primarily on tunneling and approach angle, with velocity mattering less than expected. Under xRV, sinkers and cutters get full credit for the weak contact they produce — not just whiffs.
Tunneling (weighted far more than velocity) + optimal approach angle = hard to barrel
Breaking balls are graded on tunneling first, movement second. For sliders, more horizontal sweep means better run prevention. For curveballs, vertical drop barely matters — it's all about making the pitch look hittable before it breaks.
Tunneling + horizontal sweep (sliders) or timing deception (curves) = whiffs and weak contact
Offspeed pitches are graded on tunneling and timing deception. For changeups, velocity differential is critical: 10-14 mph slower is far more effective than minimal separation. Splitters generate more whiffs through sharper drop.
Tunneling + optimal velocity differential (10-14 mph for CH) = swings over the top
