How the BANE Score Predicts March Madness Upsets

A 7-factor algorithm validated against 40 years of NCAA tournament data. No black boxes - here's exactly how it works.

Every March, brackets get busted. A 12-seed knocks off a 5-seed. A 15-seed shocks the world. The question isn't whether upsets will happen - it's which ones.

The BANE Score (Bookie Adverse Narrative Engine) is our answer. It's a composite algorithm that weighs seven research-backed factors to generate a single 0–100 score for each tournament matchup. Higher score = higher upset probability.

But here's what makes it different: we show you every piece of the math. No "proprietary model" hand-waving. No "trust our AI." Full transparency.

The Seven Factors

Each factor is weighted based on how predictive it's been in backtesting against 40 years of tournament results. The exact weighting is proprietary - it's the result of extensive regression analysis that took months to calibrate. Here's what each factor captures:

1. Seed History

The most heavily weighted factor. Some seed matchups produce upsets far more often than others. The classic 12-vs-5 matchup has a 35.3% upset rate over 40 years - more than one in three. The BANE Score uses these empirical rates directly, not theoretical estimates.

2. Defensive Efficiency

The second strongest predictor. Defense travels. Teams with elite defensive efficiency tend to perform closer to their baseline in single-elimination games. When the underdog has strong defensive metrics relative to their seed, the BANE Score increases.

3. Three-Point Variance

Three-point shooting variance is a major equalizer in tournament basketball. A team that lives and dies by the three can overperform their talent level on any given night. The BANE Score measures the gap between the underdog's 3PT reliance and the favorite's 3PT defense.

4. Tempo Control

Slow-tempo underdogs compress the number of possessions in a game, reducing the sample size and increasing variance. The BANE Score rewards tempo mismatches where the underdog is the slower team.

5. Turnover Vulnerability

When a favored team has a high turnover rate against a ball-hawking underdog, possessions get redistributed. Every turnover is roughly 1 point of expected value changing hands.

6. Conference Strength

Not all seeds are created equal. A 12-seed from a power conference is fundamentally different from a 12-seed out of a mid-major. The BANE Score adjusts based on the underdog's conference strength relative to their seed expectation.

7. Momentum

The lightest factor - but still contributes. Teams entering the tournament on a hot streak carry real psychological and strategic advantages. Conference tournament champions and teams on winning streaks get a momentum boost.

How Scores Are Calculated

Each factor produces a sub-score based on the underlying data. These are then combined using our proprietary weighting system - calibrated through regression analysis on 2,518 tournament games. The composite is normalized to a 0–100 scale and classified into tiers:

Validation Results

We backtested the BANE Score against every NCAA tournament game from 1985 to 2024 - 2,428 games across 40 seasons. The headline: at BANE ≥65 in Round of 64 games, 49.5% resulted in upsets, nearly double the 26.1% base rate. In the modern era (2003+) with full box score data, BANE ≥65 catches upsets at a 53.4% rate.

The model excels at 5-12 through 8-9 seed matchups and is transparent about what it can't do: 15-over-2 and 16-over-1 upsets are black swan events that no pre-game model can reliably predict.

Read the full backtest report →

Importantly, the BANE Score is not designed to predict every upset - that's impossible. It's designed to identify games where the conditions for an upset are strongest, so you can make informed decisions about where to differentiate your bracket.

What the BANE Score Is Not

The BANE Score is a decision support tool, not a crystal ball. It doesn't tell you "this team will win." It tells you "the conditions in this game are historically favorable for an upset at rate X%."

We don't make picks. We don't guarantee accuracy. We show you the math and let you decide.

That's the BookieBane way.