Depth Without Definition: The Flawed Foundation of Statcast’s Contact Point Data
©2025 By Ken Cherryhomes
Statcast’s newest metric, contact depth, is being treated like the Rosetta Stone of hitting. Analysts are quoting how many inches in front or behind the plate hitters make contact, as if that number alone explains power output or approach efficiency.
It doesn’t.
Measuring contact depth without integrating spray angle, pitch location, and the degree of barrel turn is like measuring altitude without knowing terrain. It is context-free, and that makes it wildly misleading.
But context is only half the problem. The other half is comparison.
Statcast attempts to anchor contact depth using two reference points—the front edge of home plate and the hitter’s center of mass. But neither is stable. One shifts with the hitter’s position in the box. The other changes with stance width, posture, and launch mechanics. So, when two hitters make contact at the same location in space, the system can still report different depths relative to these shifting anchors.
This makes cross-hitter comparisons unreliable at best and anecdotal at worst. The same collision can be labeled deeper or farther forward depending on who the hitter is, how they’re built, and the shape of their swing arc. Without a fixed spatial anchor, contact depth becomes a floating metric, even with Statcast’s attempt to reconcile this using dual measurements from the plate and the body. And without knowing how the barrel is oriented at the moment of contact, the number itself lacks interpretive value.
In short, Statcast is measuring with precision, but comparing without equivalence and interpreting without context.
The Illusion of Meaning in Averaged Contact Depth
Recent coverage around contact depth appears to me to be leaning heavily on mechanical conclusions drawn from averaged measurements. A hitter shows a tendency to make contact deeper in the zone, and so it is assumed that this trait explains their ability to drive the ball to all fields or generate power under constraint.
Even if we assume that contact points are being segmented by spray angle, and this is a reasonable assumption, it still does not solve the deeper issue. Categorizing data more precisely does nothing to resolve the instability of the reference points. Without a fixed spatial anchor, the same contact point in space can still be reported differently depending on the hitter’s posture, stance width, or launch position. And swing approach further complicates the picture. A hitter with a consistent opposite field approach will naturally accumulate deeper contact points, not as a mechanical trait, but as a byproduct of pitch selection and intent. The average shifts, but the cause is often misread.
These layered variables, body geometry, launch position, and swing intent make the aggregated data inherently unstable. Even within cleanly segmented buckets, the contact depth values are shaped by factors external to mechanics. Yet once the numbers are published, they become raw material for cross hitter comparisons. And that is where the breakdown begins.
The problem is comparison. These depth measurements, even when bucketed by field direction, still collapse into false equivalencies when analysts start comparing one hitter’s deep contact to another’s. The moment they tie those measurements to outcomes, and then attempt to reverse engineer swing traits from them, the interpretation slips back into anecdote.
This is the same pattern seen when launch angles and exit velocities were first popularized. Objective data was collected, then quickly folded into mechanical narratives without accounting for variables like timing, swing arc shape, or intent. Once again, results are being used to define presumed causes.
That is not measurement. That is mythology.
Statcast’s Two Reference Points—And Why Both Fail
Statcast thinks it solved the problem of variability by measuring contact point from two references:
- The front edge of home plate
- The hitter’s center of mass
But if comparison is the objective and standardizing contact depth is the goal, this method fails outright. They have created two flawed anchors instead of one.
The plate based measurement is skewed by where the hitter stands in the box. Move forward or back, and the number shifts, even if the collision point is identical. The body based measurement, in front of the hitter’s COM, is even worse. COM is not fixed. It shifts with:
- Base width
- Height and leg length
- Stride length
- Launch posture
- Back side loading
Some hitters launch with their COM back near the rear leg. Others launch more centered. A 6 foot 5 hitter with a wide base will show deeper contact than a 5 foot 9 hitter with a narrow stance, even if they hit the exact same pitch at the exact same depth in space. The only time these measurements offer a stable point of reference is when comparing a hitter to themselves. Across hitters, they are meaningless.
You cannot normalize contact depth with a floating anchor.
By contrast, I have selected a fixed, stable, invariant reference point within the swing, one that does not shift with stance or stride. I am not disclosing it, but that is why comparisons in my system hold across hitters, pitch types, and approach styles. Without a universal coordinate, you are not comparing hitters. You are comparing postures.
And While They Were Guessing
I have been measuring contact points and contact depths for years, long before Statcast entered the picture. I have built both standardized and context guided models that account for spray angle, timing, barrel orientation, and swing arc geometry. I use a fixed, stable, and comparable measurement point, one that avoids the shifting variables introduced by stance width, posture, or launch mechanics.
But more than that, I have measured swing time to each of those contact points. My system captures a single swing and uses a proprietary algorithm to extrapolate time to contact across 25 mapped locations in the strike zone. These values are not inferred from results. They are calculated from the geometry and temporal characteristics of the swing itself.
This work is not new. I have written extensively about these issues, built the technologies to measure them, and designed progressive modeling systems that account for layered variables and situational adjustments. My systems do not rely on inference or reverse logic. They are built on measured inputs, not assumptions.
What They’re Not Telling You: Spray Angle Is Tied to Barrel Turn
The problem is not that contact points are being measured. The problem is how they’re being measured, and worse, how they’re being interpreted.
Statcast’s anchoring system is flawed. We’ve already established that the two reference points, home plate and the hitter’s center of mass, shift with position, posture, and body type. So even before analysis begins, the depth values are unstable. But instead of correcting for that instability, analysts are drawing conclusions from the raw numbers as if they represent universal truths without accounting for the variables that define them.
What is missing in this layer of analysis is any understanding of spray angle mechanics and batter tendencies, not where the ball goes, but how it gets there.
Spray angle is dictated by barrel orientation at the moment of collision. The bat moves on a three-dimensional arc through the hitting zone. The degree of barrel turn when the bat meets the ball is what determines where it launches. Even swings with reduced arc circumference still follow this geometry.
- Same pitch, hit at different depths = different spray angles
- Same depth, different barrel angle = different launch directions
- Pulled contact = out front
- Oppo contact = deeper
Spray angle is not an outcome to be backfilled. It is a function of collision geometry. And contact depth, without knowing where the barrel is in its arc, tells you nothing.
That is why I built a 25-point contact map that accounts for both the spatial and temporal relationship at every location in the zone. It is not a chart of outcomes. It is a dynamic map of when and where—the intersection of swing time, barrel path, and contact point. It explains what happened, not just what resulted.
The Dangerous Simplicity of “Deep Contact = Good”
The analyst class is mistaking correlation for causation.
Deep contact is not exclusive to technique or swing arc. It is usually a byproduct of approach and pitch location. Suggesting that all hitters follow a deep contact model ignores basic physics and overlooks the role of physical strength, which is not evenly distributed across all hitters. It treats contact depth as a static trait rather than a dynamic outcome shaped by intent, pitch type, and barrel path. The result is a model that rewards correlation and mislabels it as cause.
A perfect example: Aaron Judge making deep contact on a pitch driven to centerfield. But this is not a blueprint, it’s a physical outlier. Judge succeeds with deep contact because his size and strength compensate for a reduced arc in which the barrel turns over sooner than other, less physically gifted hitters. For most hitters, that same collision point would produce a routine fly ball or, at best, reflect warning track power.
When you meet the ball too deep:
- Angular momentum hasn’t peaked
- The barrel hasn’t fully accelerated
- Exit velocity is reduced
- Force transfer is compromised
When you meet the ball farther out front:
- Arc is expanded
- Energy transfer is maximized
That is why the farthest hit balls in baseball, according to Statcast data, occur somewhere around 12 to 18 inches in front of the plate, not two inches behind it.
Deep contact is not a blueprint. It is a workaround for certain physical profiles and batted ball objectives. In the past, reduced swing arcs were the mark of control hitters, all fields hitters, not power hitters.
The Fix: Integrate Barrel Orientation, Swing Arc, and Timing
My suggestion is to stop treating contact depth as an isolated variable and start anchoring it to the elements that actually give it meaning. That begins with a fixed spatial reference point, one that holds across hitters, stances, and swing types. Without that, depth measurements are nothing more than floating numbers. With it, they become comparable. Patterns emerge. Interpretation becomes possible.
From there, integrate the rest: barrel orientation, swing arc geometry, and time to each zone location. Spray angle should not be inferred after contact. It should be understood before it happens, mapped in advance by knowing the arc and the bat’s position in space. Even if contact depth is bucketed by spray angle, to what end? Without knowing what caused the ball to go there, or whether it was even the intended result, categorizing it tells you nothing about the swing that produced it.
This is where most systems fall short. Because they treat post-contact results as clues to reverse engineer intent, they never establish causal structure. But from a stable reference point and a mapped arc, you can shift the model from static comparisons to actionable analysis. If a hitter consistently pulls inside pitches foul, you do not guess. You identify the collision point, the orientation mismatch, and, as I do in my system, the precise timing pattern tied to that zone location. I am not inferring results. I am mapping them, then correlating them to known optimal outcomes across both space and time.
Take the example below. A swing and miss is labeled a spatial miss, but the bat was actually on the correct arc.
The issue was not where the swing was aimed. It was when the swing arrived. The targeted contact point was right. The timing was late. Most coaches would prescribe a mechanical fix. But with access to the hitter’s historical contact map, we can see that the circled point aligns with optimal outcomes. Now we have precision. We do not speculate. We isolate the temporal error and prescribe a timing correction, not a mechanical guess. That is what I do.
From that data set, you can ask the right questions. Was it a deviation in barrel orientation? Was the arc shape altered? Did bat speed deviate from historically optimal outcomes to that depth, indicating hesitation?
This is what makes the difference between surface level correlation and mechanical insight. Contact depth is not diagnostic without the temporal and spatial layers behind it. Analysts treat contact point as a clue. I treat it as a consequence.
Statcast is measuring where hitters make contact. That is progress. But without explaining why, or anchoring it to anything stable, it is just surface level data dressed in scientific vocabulary.
Closing Thought
If your model treats contact depth as a static trait, you’ve misunderstood the swing.
Contact depth is a consequence of timing, of arc, of barrel orientation. It’s not a cause. And until models account for that reality, they’ll keep offering precision without meaning.