The Battle Is Won
Before the Ball
Leaves the Hand.

You have seen this before. A pitching prospect is lighting up Double-A. Mid-90s fastball, strikeout rate pushing 30 percent, ERA in the low twos. The reports say the stuff is electric. The stat line says he's ready. Then he gets to Triple-A, or the majors, and suddenly the fastball nobody could touch is getting fouled off. The breaking ball that buried minor league hitters is getting spit on. The swing-and-miss is gone, even though the radar gun still says 96.

Nothing changed. Except the hitters.

Modern hitters are not losing to velocity. They are losing to prediction errors. The pitchers who succeed at the highest level are the ones whose arsenals consistently break those predictions - not just once, but on every single pitch, across a full at-bat, in a way that compounds over the course of a game. Understanding why requires going deeper than strikeout rates or Stuff+ grades. It requires understanding how the brain processes a baseball in flight.

Every swing starts as a prediction problem. A major league fastball travels from the pitcher's release point to the catcher's mitt in roughly 400 milliseconds. A hitter's neuromuscular system needs at least 150 to 200 milliseconds to initiate and complete a swing. That leaves somewhere between 150 and 200 milliseconds to make a decision - a window so small that true reaction is essentially impossible at high velocity.

What hitters are actually doing is pattern recognition. Within the first 100 to 150 milliseconds after release, the brain extracts information from release height, arm angle, extension, spin direction, and early trajectory. From that data, it constructs a prediction: where will this pitch cross the plate, and when? The swing launches to intersect that predicted location. The hitter is not reacting to the ball at contact - he committed 200 milliseconds earlier.

Pitchers win when the ball does not end up where the prediction said it would.

This is not a new insight, but its implications for how we evaluate pitching prospects are more radical than most analysis acknowledges. A prospect with a 97 mph fastball and a 30 percent strikeout rate in Double-A may be winning primarily because minor league hitters are building poor predictions from inadequate exposure to his release. Against major league hitters with more reps, better pattern libraries, and video rooms full of data, the same arsenal may be immediately readable. The stuff has not changed. The prediction quality has.

A baseball thrown with backspin - the way a four-seam fastball is thrown - experiences an upward Magnus force as it travels toward the plate. Gravity is simultaneously pulling it down. The interaction between these two forces determines how the ball actually moves through space.

Induced Vertical Break measures how much the pitch resists gravity relative to a theoretically spinless ball following a pure ballistic path. A four-seam fastball with 18 inches of IVB does not literally rise - that would violate physics. What it does is fall significantly less than the hitter's brain expects based on the early flight trajectory. The brain sees the ball leave the hand, calculates a descent curve based on the perceived speed and early path, and generates a prediction. The high-IVB fastball falls off that prediction curve. It arrives higher than expected.

The practical effect is striking. Hitters trained to match their swing path to the expected descent of a typical fastball consistently swing underneath high-IVB pitches at the top of the zone. The contact that does occur is often on the bottom half of the ball - weak pop-ups, harmless flyouts, and the occasional miss entirely.

This is why the swing-and-miss rate on four-seam fastballs is heavily correlated with IVB. A pitcher with 18+ inches of IVB and consistent release height can locate the same fastball to the same zone and generate whiffs repeatedly, not because hitters do not know what is coming, but because the ball repeatedly ends up outside the range of their pre-committed swing path.

Vertical Approach Angle describes the angle at which a pitch enters the strike zone. A steep VAA means the pitch is descending sharply as it crosses the plate. A flat VAA means it is arriving on a more horizontal plane.

Hitter swing paths are not horizontal - they angle upward through the zone to match the expected descending plane of the pitch. This is optimal for pitches that enter the zone at the angles hitters are expecting. But a fastball with both high IVB and a flat VAA creates a specific kind of problem: the pitch stays on a near-horizontal plane for longer than expected before arriving, and when it does arrive, it is higher in the zone than the swing was designed to intersect.

Flat VAA is primarily a function of extension and release height. Pitchers who release the ball closer to home plate (greater extension) reduce the angle at which the pitch descends before crossing the plate. Pitchers with high release heights generate naturally flatter approach angles at the top of the zone. Overhand arm slots tend to produce steeper VAA; lower three-quarter slots produce flatter angles.

The interaction between IVB and VAA is more important than either metric alone. A pitch with elite IVB but steep VAA still arrives at a predictable angle and can be accounted for. A pitch with moderate IVB but very flat VAA creates a different kind of deception - the ball stays on a plane the hitter is not prepared for. The combination of both - high IVB and flat VAA - is the theoretical ideal for a strikeout fastball at the top of the zone.

IVB and VAA describe individual pitches. Tunneling describes how pitches interact as a set.

The concept is straightforward: pitches that share early flight path before diverging late force the hitter into a specific kind of error. If a fastball and a curveball emerge from the same release point and follow the same initial trajectory for the first 20 to 25 feet of flight, the hitter cannot distinguish between them until the decision window has essentially closed. The fastball continues toward the top of the zone. The curveball drops below it. Both looked identical until it was too late to adjust.

Good tunneling is not accidental. It requires matching three things: release height and angle, early trajectory, and perceived velocity window. A curveball thrown with the same arm action as the fastball, from the same release height, at a velocity close enough that the early speed signal does not immediately distinguish it - that pitch will tunnel well. A curveball thrown from a lower release point with dramatically different arm action will telegraph itself regardless of movement quality.

This is where the gap between minor league and major league performance often lives. Minor league hitters have limited exposure, less advanced pitch recognition, and frequently face pitchers whose arsenals work on individual pitch quality alone. A 97 mph fastball and a loopy curveball with dramatically different arm action can still generate strikeouts in Double-A because the hitters simply have not seen enough of the pitcher to build an accurate prediction model. Major league hitters with video rooms, advance scouting, and thousands of reps against similar stuff solve that problem in a fraction of the at-bats a minor league hitter would need.

The question for evaluators is not whether the fastball misses bats in Double-A. It is whether the fastball and the breaking ball share enough early trajectory that the pitch sequence creates decision problems. Whether the changeup mirrors the fastball's initial flight before introducing horizontal or downward movement. Whether the pitcher's release consistency is good enough that multiple pitches can actually share the same tunnel window.

No amount of pitch quality matters if the delivery is inconsistent. Release point variation is one of the clearest signals of a pitcher who will struggle against major league hitters. When the release point moves - even by a fraction of an inch - the early trajectory changes, the tunnel breaks down, and hitters receive more distinguishing information earlier in the flight. A pitcher who tunnels beautifully when fresh and falls off his release point by the fourth inning is effectively showing hitters two different arsenals.

Extension adds a separate dimension. Every additional inch of extension toward home plate has two effects: it reduces the time the hitter has to process the pitch, and it flattens the VAA. For a pitcher releasing at 6.5 feet of extension versus 6.0 feet, the difference in perceived velocity is roughly 1 to 1.5 mph - not from actually throwing harder, but from delivering the same velocity from a point closer to the plate. That compression is meaningful at the major league level where fractions of a mph in perceived velocity affect swing timing decisions.

The combination of elite extension, consistent release point, and pitches that actually share a tunnel window is rarer than the prospect industry typically acknowledges. Finding a pitcher who has all three is finding a genuine major league arm - regardless of what the radar gun says.

A 30 percent strikeout rate in Double-A is meaningful. It is not proof of a major league arm. The distinction matters enormously for prospect evaluation, and collapsing the two is one of the most common analytical errors in amateur scouting.

Minor league hitters are solving a different problem than major league hitters. They have less exposure, less advance information, and are frequently facing pitchers in their first sustained run through a level. A pitcher who is winning those at-bats on raw stuff - velocity alone, or a single elite pitch without an effective tunnel partner - may generate impressive strikeout totals against opponents who simply have not seen him enough to build an accurate prediction model.

The more useful question is: does this pitcher's arsenal create genuine decision problems, or is it creating recency problems? The difference is whether the pitch combinations force hitters into prediction errors on the physics of the pitches, or whether hitters are simply failing to recognize a pitch they have not seen much of. The first is a structural advantage that scales upward. The second is a temporary advantage that disappears.

Metrics that help answer this distinction: SwStr% on individual pitches (is the swing-and-miss coming from genuinely good pitch shapes or from hitters swinging at pitches they cannot see?), release point consistency over the course of outings (does the pitcher maintain his tunnel window deep in games?), and pitch mix distribution (does he use his secondary pitches enough to establish them as threats, or does he lean on one offering and dare hitters to lay off everything else?).

Abstract principles are only useful when they connect to real players. Here is how the tunneling and IVB framework applies to specific pitching prospects currently tracked in the model or relevant to the 2026 class.

Even a perfectly tunneling two-pitch combination has a fundamental problem: with only two pitches, hitters face binary prediction. They are either right or wrong. In a binary system, even partial information is useful - a hitter who eliminates one pitch from the possibility space has a 50/50 guess on every pitch, plus whatever directional information the partial signal provides.

A three-pitch arsenal reduces that to a third-probability problem even before the tunnel starts providing information. A four-pitch arsenal with strong tunnel relationships between multiple pairs - fastball and changeup tunneling in one direction, fastball and curveball tunneling in another - creates a prediction problem that cannot be solved with available information. There are simply too many branches in the decision tree for the available processing time.

This is why pitch mix distribution matters as much as individual pitch quality in prospect evaluation. A pitcher who throws his fastball 70 percent of the time has told hitters that the slider or curveball appears roughly 30 percent of the time. Hitters can sit fastball, look for the breaking ball as a change-of-plan, and function at a reasonable success rate. A pitcher who distributes pitches more evenly - 45 percent fastball, 30 percent slider, 25 percent changeup - forces hitters into a much harder prediction problem before any single pitch has even been released.

This is one of the reasons Corbin Burnes' reinvention was so effective. By moving away from the fastball as a primary pitch and building a cutter-heavy attack with multiple complementary offerings, he transformed a binary prediction problem into a multi-branch one. The individual pitches are excellent. But the distribution and sequencing compound that excellence into something that resists adjustment in a way that raw stuff alone cannot achieve.

The velocity obsession in prospect evaluation is not irrational - velocity correlates with success because it compresses reaction time and makes individual pitches harder to identify. But velocity is a floor, not a ceiling. A 97 mph fastball that tunnels poorly, moves predictably, and is released from the same clear window as a curveball with dramatically different arm action can be solved by good major league hitters in under a full game's worth of at-bats.

The traditional scouting framework grades individual pitches on a 20-80 scale. A 60-grade fastball, a 55-grade slider, a 50-grade changeup. The grades describe each pitch in isolation. They do not capture whether the 60-grade fastball and the 55-grade slider actually share enough early trajectory to create a tunnel. A pitcher with a 70-grade fastball and a 45-grade curveball with poor tunnel relationships may project worse than a pitcher with a 60-grade fastball and a 55-grade curveball that shares the first 25 feet of flight perfectly.

The analytical frontier in pitcher evaluation is moving toward pitch-level trajectory data that measures tunnel quality directly. How similar are the first 15 feet of flight between the primary fastball and the primary breaking ball? How consistent is the release point within and across outings? What is the decision-point location for each pitch pair, and how much movement has occurred at that point relative to the total movement?

None of those questions are answered by a radar gun reading or a raw strikeout rate. But they are increasingly answerable from the Statcast and TrackMan data that player development departments have access to - and that the most analytically sophisticated organizations are using to make development decisions that look unintuitive from the outside.

The battle is won before the ball leaves the hand because by the time the ball is in flight, the pitcher's entire arsenal has already either set up or failed to set up the deception. Release point consistency has already determined whether multiple pitches can share the same tunnel window. IVB and VAA have already determined whether the fastball will arrive where the hitter's brain expects it to. Extension has already compressed or expanded the window the hitter has to distinguish between pitch types.

Evaluating pitching prospects through this lens changes the conversation materially. It shifts focus from what any single pitch does in isolation to what the arsenal does as a system. It demands asking whether the strikeout rate reflects genuine prediction problems or temporary exposure gaps. It requires distinguishing between pitchers who will scale because their arsenal creates structural deception and pitchers who are generating stats against underprepared competition.

That distinction is the core of accurate pitcher projection. And it is consistently underweighted in how the prospect industry talks about arms - because individual pitch grades are easier to communicate than pitch interaction quality, and because minor league strikeout numbers are immediately satisfying in a way that tunneling percentages are not.

The pitchers who survive the jump to the majors are usually not the ones who were the most impressive in Double-A. They are the ones whose arsenals kept creating the same prediction problem at every level - because the problem was structural, not situational.