The Pitch Tunneling Atlas

What Tunneling Actually Is

A batter facing a 95 mph pitch has roughly 400 milliseconds from release to plate arrival. But the decision to swing happens earlier — by the time the ball is about 24 feet from the plate, the batter must commit. Everything after that is physics and reflexes.

Tunneling is what happens when two different pitches pass through nearly the same point at that commit moment, then diverge on the way to the plate. The batter makes a decision based on where the pitch appears to be going — and then gets something completely different.

Tap through the steps below to see how Craig Kimbrel's 4-seam fastball and knuckle-curve look at each stage of the flight:

Kimbrel's fastball (red) and knuckle-curve (blue) at three points along the trajectory. At the decision point, they're 1.8 inches apart. At the plate, 15.1 inches apart. The batter committed when they looked the same.

That 1.8-inch gap at the decision point is smaller than a baseball. The batter is making a swing-or-take decision on a pitch whose final destination could be 15 inches in either direction. This is what elite tunneling looks like.

Tunneling vs Anti-Tunneling

Not every pitcher tunnels well. Some do the opposite — their pitches are farther apart at the decision point than at the plate. The batter can already see what's coming before they need to commit.

Left: Kimbrel's pitches diverge from decision to plate — the gap widens. Right: Taylor Rogers' pitches converge — the gap narrows. One is deceptive; the other telegraphs.

Kimbrel's pitches diverge: they spread apart between the decision point and the plate. Rogers' pitches converge: they're farther apart at the decision point than at the plate. In Rogers' case, the batter actually gets more information about pitch type as the pitch approaches.

We measure this with a single number: divergence = plate separation − decision-point separation. Positive = deceptive (pitches spread apart). Negative = anti-tunnel (pitches converge). Kimbrel: +13.3 inches. Rogers: -1.5 inches.

The Question Nobody's Answered

Everyone in baseball talks about tunneling. But nobody has tested the fundamental question across the full league: does tunneling actually predict outcomes, or is it just geometry?

Baseball Savant publishes pitch movement. FanGraphs discusses tunneling qualitatively. Nobody publishes a systematic metric that decomposes deception into its components and tests whether it predicts whiff rate beyond raw stuff.

We built that model. 654 pitchers, 739,820 pitches, full trajectory data. And the answer is more nuanced than the hype suggests.

What Actually Matters: The Decomposition

We decomposed pitch deception into two components:

• Plate separation — how different your pitches end up at the plate (pitch diversity)
• Decision-point tightness — how similar they look at the commit point (pure tunneling)

Both matter. But they don't matter equally.

Incremental R² predicting whiff rate. Velocity, spin, and movement explain 7.3%. Adding plate separation adds +9.0%. Adding decision-point tightness adds +0.8% more. OLS regression across 621 qualified pitchers.

Plate separation — having pitches that end up in different places — explains 11 times more whiff-rate variance than decision-point similarity. The tunneling signal is real (the decision-point coefficient is negative and significant at p=0.016, meaning closer at the decision point = more whiffs), but it's the supporting actor, not the lead.

The pitchers who miss the most bats aren't just the ones with the best tunnel. They're the ones whose pitches end up in the widest range of locations at the plate while also looking similar on the way there. Diversity at the destination matters more than similarity on the journey.

The Deception Leaderboard

We score each pitcher on divergence: plate separation minus decision-point separation. Higher divergence means the pitcher's pitches spread apart more between the decision point and the plate — the batter sees something similar and gets something different.

Top 10 most and least deceptive arsenals in 2025 by divergence (plate separation − decision separation). Tap any pitcher for the component breakdown. Blue = starters, clay = relievers.

Craig Kimbrel leads at 13.3 inches of divergence. His 4-seam and knuckle-curve share a tight window at the decision point (1.8 inches apart) but land 15.1 inches apart at the plate. The batter is committing to a pitch that could arrive in vastly different locations.

Bryan Abreu is the purest tunneler in the top group: his pitches are just 0.9 inches apart at the decision point — the tightest in the top 20. His plate separation is more modest (12.5 inches), but the tunnel itself is elite.

At the bottom, Taylor Rogers has negative divergence (-1.5 inches): his pitches are actually farther apart at the decision point than at the plate. His arsenal converges rather than diverges — the opposite of deception.

Reliever caveat: 18 of the top 20 are relievers. This is partly real (relievers optimize 2-3 pitch combinations) and partly methodological (fewer pitch pairs = less dilution). Among starters with 4+ pitch types, Spencer Strider, Blake Snell, and Pablo Lopez lead.

Which Pitch Pairs Tunnel Best?

Tunneling is pair-specific. Some pitch combinations naturally produce high divergence; others are physically constrained to look similar throughout the entire trajectory.

Average divergence (inches) for every pitch-pair combination across the 2025 league. Green = high divergence (more deceptive). Red = low divergence. Only includes pairs where 20+ pitchers throw both types at 30+ volume.

The best pair in baseball: 4-seam + knuckle-curve (9.8 inches average divergence). The fastball rides; the knuckle-curve drops. At the decision point they're 3.2 inches apart. At the plate they're 13.0 inches apart. That 9.8-inch growth is the most of any common pair.

The worst pairs are within the same movement family: slider + curveball (1.6 inches), slider + sweeper (1.7 inches). These pitches move in similar directions and don't diverge much — a batter who reads one can handle both.

The sinker-slider combination (SI-SL) has an interesting profile: the tightest decision-point separation in the league (3.0 inches) but only moderate plate separation (10.5 inches). It's the best pure "tunnel" — the two pitches look almost identical longer than any other pair — even though the plate divergence isn't extreme.

Does Tunneling Predict Outcomes?

The validation question. If tunneling doesn't predict results, it's just geometry.

Left: Divergence vs whiff rate (r=0.355, partial r=0.319 controlling for velocity+spin). Right: Decision separation vs plate separation, colored by whiff rate. The ideal position is bottom-right: tight at the decision point, wide at the plate.

The correlation between divergence and whiff rate is r = 0.355. After controlling for velocity, spin rate, and movement magnitude, the partial correlation is r = 0.319. Tunneling adds real predictive value beyond raw stuff — but it's modest, not dominant.

The full regression model tells the clearest story:

Plate separation (p<0.001) and decision-point tightness (p=0.016) are both independently significant. The plate separation coefficient is ~1.6x larger. Spin rate and movement magnitude — the usual "stuff" metrics — are not independently significant once tunneling components are included.

Important limitation: this signal holds for whiff rate only. CSW% and xwOBA-against show no significant tunneling effect in this framework. Deception helps generate swings-and-misses, but its path to broader run prevention is indirect at best.

The Batter-Handedness Asymmetry

Some pitchers tunnel well from only one side. The same arsenal can produce elite divergence against right-handed batters and poor divergence against left-handers — or vice versa.

Kyle Gibson has the largest asymmetry: 10.3 inches of divergence vs RHB but just 0.3 inches vs LHB. His arsenal is effectively deceptive from one perspective only. Elvin Rodriguez shows the reverse: 17.5 vs LHB, 7.2 vs RHB.

This has platoon-advantage implications that go beyond traditional handedness splits. A pitcher who tunnels well only against opposite-hand batters may be losing deception in same-side matchups — a vulnerability that lineup construction could exploit.

What This Means

Tunneling is real, but the pitch-design community has the emphasis slightly backwards. The dominant component of deception is plate-level pitch diversity — how different your pitches are when they arrive. The tunneling component (similarity at the decision point) adds a real but smaller edge on top.

For pitching coaches: the highest-leverage design target isn't making two pitches look identical out of the hand (though that helps). It's maximizing the plate separation between your most-used pitch pairs while keeping the release point tight. The tunnel is the icing; the cake is diversity of outcomes.

For fans: the next time a broadcaster says a pitcher "tunnels well," look at where the pitches end up, not just where they start. The pitchers at the top of this leaderboard — Kimbrel, Ginkel, Megill, Williams — all share one trait: their pitches arrive in locations that make the batter's committed swing catastrophically wrong.