Veritas Baseball Index // D1 Rankings

// Matt DePrey's VBI

Why RPI, KPI, and DSR all SUCK

RPI measures schedule strength more than team quality. Since 75% of the formula comes from opponent and opponent-opponent winning percentage, teams inherit value from the schedules they play rather than fully earning it on the field. The system can reward teams for simply surviving difficult schedules, even when their actual results are mediocre.

KPI attempts to fix this by rewarding "quality wins," but that creates a different problem: the rankings become dependent on arbitrary cutoff tiers and constantly shifting definitions of what counts as a "good win." Teams are judged relative to constructed categories instead of the full network of games.

DSR goes even further toward prediction. It incorporates win expectancy, contextual adjustments, and margin-based modeling to estimate how strong a team is going forward. That can be useful for forecasting, but predictive systems inevitably inject more assumptions into the rankings. A 10-run win becomes fundamentally different from a 1-run win, and teams begin optimizing for model behavior rather than simply winning games.

VBI takes a different approach. Using the Colley Matrix, every team's rating is solved simultaneously through the entire network of wins and losses across Division I baseball. No score margin. No quality-win buckets. No predictive modeling. The system only asks:

   * Who did you play?
   * Did you win?
   * Where was the game played?

Instead of hardcoding schedule strength, quality tiers, or predictive assumptions into the formula, VBI allows strength to emerge naturally from the interconnected results of the entire season. Every game matters. Every opponent matters. Every team influences every other team through the same mathematical system.

// the equation

C · r = b

Three variables. One system. 308 unknowns solved simultaneously.

C · r = b

the matrix

A 308×308 matrix representing every Division I team. Each team's main value is based on its weighted games played, while every matchup against another team creates a negative connection between the two teams. When the system solves the equation, all of those connections work together so every team's rating is influenced by the strength of its opponents.

the ratings

What we solve for. One number per team. Higher = better. No team's rating can be solved without knowing all others — that interdependence is the whole point.

the record vector

The b vector measures a team's game results. It starts every team at a neutral baseline of 1, then adds weighted wins and subtracts weighted losses based on game location. Teams that win difficult games on the road build a stronger input value before the matrix adjusts for opponent strength.

This is not adjusted winning percentage. Solving C·r = b creates a web of interdependence across the entire sport. Xavier's rating is shaped by Arkansas's rating, which is shaped by every team Arkansas played, and so on — simultaneously, across all 308 teams. Schedule strength isn't an input. It's an output.

// location weighting

Where you play matters.

Every game result is multiplied by a location weight before entering the system.

×1.1

ROAD

Win away from home: +1.1 to b[i]. Lose on the road: −0.9.

×0.9

HOME

Win at home: +0.9 to b[i]. Lose at home: −1.1.

// see the full 308×308 matrix

Every cell is a game connection. Drag the slider to watch it fill up day by day across the 2026 season.

// built by matt deprey · veritasbaseballindex.com · updates nightly at 2am est · 2026

// try it yourself

How does one game move the needle?

Xavier · Big East · 8 games in · pick what happens next

VBI RATING

0.5000

record: 4-4 games: 8

C[i][i] 9.60 = 2 + Σ(games)

Anchor weight. Bigger = harder to move.

b[i] 0.0000 = 1 + Σ(W×wt) − Σ(L×wt)

Weighted record. 4W 4L = at baseline.

r[i] 0.5000 = solve(C · r = b)

Solved with all 308 teams at once.

Pick a game to see the math move.

// game log

// game 9 — what happens?