MLB Pitcher
Strikeout Distributions
and Projections
See the full distribution, not a point projection.
KSplit models strikeouts at the plate appearance level, building full probability distributions from pitcher tendencies, hitter handedness, and expected workload. The result is a system that explains how strikeout outcomes form, not just what the number might be.
Most models give you a number.
KSplit gives you a structure.
The strikeout market is saturated with single-number projections built on similar inputs. KSplit approaches it differently — modeling the shape of each outcome, not just its center.
- A single projected totalOften expressed as a mean or median with no information about how that number is distributed around uncertainty.
- Team-level K rate inputsOpponent strikeout rate against righties or lefties, applied uniformly regardless of lineup structure or order.
- Fixed workload assumptionsBF or innings estimates treated as constants rather than as a distribution of their own, ignoring game-state variance.
- No right-tail visibilityNo measure of how likely extreme outcomes are, or how accessible the upside actually is given the line.
- A full probability distributionEvery K outcome from 0 to 13+ carries an explicit probability. You see not just the median, but where the mass is concentrated and where the tail runs.
- Plate appearance-level modelingEach batter slot in the lineup is modeled individually using pitcher K% splits against their handedness, then aggregated across the confirmed batting order.
- Workload-weighted distributionsExpected batters faced drives the shape of the distribution. A pitcher projected for 5 innings looks structurally different from one expected to go 7.
- Right Tail Mass (RTM%)Every projection includes the probability of exceeding the line by 2+ strikeouts, calculated against the modeled distribution and corrected for truncation.
Where the mass sits matters more than the median
Two pitchers with the same projected median can have completely different distributions. One might be tightly centered, the other skewed right with accessible upside. KSplit shows you both and tells you which one.
Right tail mass as a structural signal
RTM% measures how much probability sits at least 2 strikeouts above the market line. It is not a pick. It is a structural read on whether the upside of the distribution is accessible given where the line is set.
Not all high-K projections are the same
KSplit classifies every game's distribution by ceiling profile — Low/Centered, Mid/Tail-Supported, High/Tail-Driven — based on where probability mass accumulates above the median. The label tells you the shape before you read a single number.
Built around the outcome, not the estimate
Plate Appearance-Level Inputs
Each projection is built from the ground up using per-batter K probability against the confirmed lineup. Pitcher splits against handedness, hitter zone rates, and CSW% combine at the slot level before aggregation. No team averages, no shortcuts.
Built for Today's Slate
Distributions update daily using confirmed lineups, handedness matchups, and current workload expectations. Every projection reflects the specific matchup in front of you, not a seasonal average transplanted onto a different game context.
Performance Evaluated on Distributions
Every projection is archived with the full probability table and actual outcome. Accuracy is measured using CRPS and bucket-level calibration across the entire distribution, not just whether the over hit. No selective screenshots. No cherrypicked results.
Why Strikeouts?
Strikeouts occur one plate appearance at a time. They are independent of team defense, park dimensions, and most of the noise that makes other outcomes hard to model structurally. Each batter either strikes out or doesn't, and that binary event happens enough times per game to produce a meaningful distribution.
That structure makes strikeouts uniquely well-suited to probability modeling. You can assign a K probability to each plate appearance based on observable, repeatable characteristics: pitcher strikeout rate against handedness, hitter contact and swing tendencies, CSW rate. Aggregate those individual probabilities into a full game distribution and the line becomes a threshold inside that distribution, not an external benchmark to be manually compared.
KSplit is built entirely around this. The strikeout market is the one place in baseball where a well-constructed distribution model has a genuine informational edge over a point estimate.
Ready to see the full distribution?
Create a free account to browse historical strikeout projections and archived results. Upgrade anytime to unlock live slates, probability tables, full distributions, ceiling profiles, and RTM% across every game on the slate.
Create Account