KSplit Analytics

How to Guide for Today's Dashboard

KSplit models ranges of outcomes, not single predictions. Follow the board left to right — from baseline expectation to market comparison to probability to upside structure.

Need definitions for any metric? Open the Glossary.

Quick Flow
Median→Edge→Tail Profile
1
Start With Leash and Baseline Ks
Workload context and baseline expectation

Begin with workload context, then read the three baseline projections in order:

1stMedian Ks — most stable expectation
2ndMean Ks — influenced by upside
3rdMode Ks — most likely exact outcome
If the mean is greater than the median, the distribution is being pulled to the right — upside outcomes are carrying more weight.
2
Compare to the Main Line
Model vs. market pricing
LineThe market's strikeout total.
Over / Under OddsMarket pricing including vig.
Best EdgeLargest model vs. market divergence.
Best Edge signals where the largest probability difference exists — not a prediction of outcome. It shows disagreement between model and market, not a recommendation.
3
Move Into Probability
Distribution range and ladder upside
DistributionWhere outcomes cluster and where risk exists.
+1 / +2 LaddersHow often the distribution extends beyond the main line.

As you move up the ladder, outcomes become less likely and more sensitive to workload and matchup.

4
Evaluate Upside StructureInsight
Stable upside vs. fragile upside

The right side of the board explains why upside exists — and how reliable it is.

Right Tail Mass %How much of the range lives above the mean.
DSSUpside presence relative to the line.
CCCWhether the game environment lets upside survive.
Split InfluenceHow much handedness splits are shaping the projection.
Ceiling ProfileHow usable the tail structure is.
Tail EnvironmentOverall distribution classification.
These tools separate structural upside from variance-dependent upside. A high DSS with a fragile tail environment reads very differently than a high DSS with a stable one.
5
Use Yesterday's Board for Context
Results within the distribution

Yesterday's Board shows what was expected, what happened, and where results landed within the distribution.

Variance is expected. The goal is long-term process consistency — not outcome-by-outcome accuracy.
KSplit models ranges, not outcomes. No individual projection guarantees a result. The framework is designed to evaluate process — where probability sits, where the market disagrees, and whether upside is structurally supported.