Glossary
Definitions for every metric and term used across the KSplit framework.
For a walkthrough of how to use these metrics on the live board, read our How to Guide for Today's Dashboard.
LineMarket strikeout total
The strikeout total offered by the market for the pitcher. This is the reference point for all probability and comparison metrics. Market lines may move over time.
Over / Under OddsMarket pricing including vig
The market odds for both sides of the strikeout line. These include sportsbook margin (vig) and are shown for comparison purposes only. Odds may change as markets update.
Mean KsAverage modeled outcome
The expected strikeout total based on the modeled distribution. This represents the average outcome across all possible results — not a guaranteed projection.
Median KsDistribution midpoint — Expected Ks
The midpoint of the modeled strikeout distribution. Half of modeled outcomes fall above this value, half below. This is the Expected Ks figure displayed on the board.
Mode KsMost likely single outcome
The most frequently occurring strikeout outcome in the modeled distribution. This is the single most likely discrete result — one specific number, not a range.
Model Over / Under ProbabilityFull distribution over/under estimates
The modeled probabilities of the pitcher finishing over or under the listed strikeout line. Derived from the full outcome distribution, not a single estimate.
Distribution Shape Score (DSS)How much upside exists relative to the line
Measures how much a pitcher's strikeout projection is influenced by right-tail upside rather than a tightly centered outcome. DSS is line-relative — it may shift when the market line moves, even if the underlying distribution does not change.
Low scores indicate a centered, median-driven projection. High scores indicate the distribution is being pulled upward by meaningful upside mass.
Best EdgeLargest model vs. market divergence
The difference between the model's probability and the market's de-vigged implied probability. Highlights where the model and market disagree most significantly.
SideWhich side has the largest probability gap
Indicates which side of the line corresponds to the largest probability difference between model and market. Shown for clarity — not instruction.
ConfidenceQualitative strength label
A qualitative label reflecting the strength of the probability difference between the model and market. It does not imply certainty or guarantee an outcome.
Probability +1 from LineP(K ≥ line + 1)
The modeled probability that the pitcher exceeds the listed line by at least one strikeout. Reflects moderate upside scenarios.
Example: if the line is 5.5, this is the probability the pitcher reaches 7.
Probability +2 from LineP(K ≥ line + 2)
The modeled probability that the pitcher exceeds the listed line by at least two strikeouts. Reflects higher-end outcome scenarios.
Example: if the line is 5.5, this is the probability the pitcher reaches 8.
Edge +1Model vs. market gap at +1 threshold
The difference between the model's probability that a pitcher records one more strikeout than the market line and the market-implied probability for that outcome.
Edge +2Model vs. market gap at +2 threshold
The difference between the model's probability that a pitcher records two more strikeouts than the market line and the market-implied probability for that outcome. Values are typically smaller than Edge +1 due to lower tail probability.
Fragile Tail EnvironmentUpside present but unreliable
A distribution state where right-tail outcomes require significant variance or favorable conditions to occur. Tail probability may exist mathematically, but conversion reliability is low due to unstable shape or weak supporting structure.
These environments carry elevated risk for upside-focused projections.
Low Tail EnvironmentLimited right-tail mass
A distribution with limited right-tail mass where extreme strikeout outcomes are statistically unlikely but still possible. The model indicates a heavier concentration around median outcomes, reducing ladder-style upside.
These profiles generally favor conservative expectations over aggressive ceiling outcomes.
Balanced Tail EnvironmentModerate, centered distribution
A middle-ground distribution where outcomes cluster near the center and tail behavior is neither strongly suppressed nor strongly emphasized. Upside exists but is not a defining characteristic of the projection.
These environments typically reflect stable median expectations with moderate dispersion.
Balanced Tail (Accessible)Moderate upside without extreme variance required
A neutral distribution shape where upside exists and remains proportionate to the center of the distribution. Right-tail outcomes are reachable without requiring extreme variance, but the model does not classify the environment as strongly tail-driven.
These profiles generally represent controlled risk with moderate upside accessibility.
Tail-Friendly EnvironmentMeaningful upside, moderate stability
A distribution state with meaningful right-tail potential, but slightly less stability than a Stable Tail Environment. The tail can realize, but outcomes rely more on game flow or volatility factors.
Upside is present — conversion to extreme outcomes is less consistent than fully stable tail profiles.
Stable Tail EnvironmentStructurally supported, high-conversion upside
A distribution state where right-tail outcomes are both structurally supported and accessible. The model shows strong tail probability with high conversion reliability — elevated strikeout outcomes occur without requiring extreme variance.
These profiles typically combine strong tail mass with stable distribution shape and are the most dependable high-upside environments.
Right Tail Mass %How much of the range lives above the mean
How much of the outcome range lives well above the mean. More tail mass means more ways for the over to win big — not just barely clear the line.
Expected CRPS (xCRPS)Distribution stability vs. historical behavior
Classifies how stable or unpredictable a strikeout environment is, based on how well similar distributions have matched real outcomes historically. Derived from CRPS (Continuous Ranked Probability Score), which evaluates the accuracy of the full distribution — not just a single projection.
Lower xCRPS means outcomes tend to stay closer to the median. Higher xCRPS means outcomes are more spread out, with wider result ranges and more right-tail events.
Ceiling Conversion Context (CCC)Will the environment let upside survive?
Measures how reliably a pitcher's existing strikeout upside holds through real game flow, based on workload stability, control, and volatility. CCC does not create upside and does not change the Ceiling Profile — it only answers: if the upside exists, how often does it actually convert?
Ceiling ProfileHow usable is the upside itself?
Describes the shape of the strikeout distribution's right tail via DSS, independent of game context. It is structural, not situational.
Tail WatchlistCurated slate — right-tail opportunity filter
A curated view that surfaces pitchers with meaningful right-tail probability, supportive workload, and favorable distribution shape. This is a focus tool — it prioritizes upside without changing any underlying projections.
KSplit models ranges, not outcomes. Use the glossary as a reference for how each metric behaves; no single metric guarantees a particular result.
Ready to see these metrics live? View Today's Board or preview Yesterday's Board.