NCAA Tournament Picks: How to Use Model Data to Research March Madness

Adjusted Efficiency Is the Most Important Number

Raw win-loss records are misleading in March. A team with a 25-5 record from a weak conference may have a worse adjusted efficiency profile than a 20-12 team that played the toughest schedule in the country. KenPom's adjusted efficiency margin — the difference between a team's adjusted offense and adjusted defense per 100 possessions — is the single most predictive metric for tournament outcomes. Teams in the top 30 of KenPom's efficiency margin historically outperform their seed in the tournament. Teams ranked outside the top 50 are prone to first-round exits regardless of seeding.

Barttorvik's T-Rank is a free, publicly accessible alternative with similar adjusted efficiency methodology plus additional shot-quality metrics. Both should be in your research toolkit for every round.

Compare Adjusted Efficiency Margins

Start by looking at the efficiency margin gap between teams. A game where the favorite is +15 in adjusted efficiency margin against an opponent at +5 is a significant talent gap that should translate to a meaningful spread. If the posted spread feels too small for that gap, look at the over on the favorite or consider the spread directly.

Adjust for Neutral Court

Remove the home court advantage from your model. KenPom's home court adjustment is approximately 3.75 points. If you're using team efficiency numbers that include home games, the neutral-court projection will be slightly different than what you'd calculate from raw stats.

Factor in Tournament Experience

Teams making their first tournament appearance — especially mid-majors — sometimes underperform their efficiency ratings due to the elevated pressure environment. This effect is real but small and shouldn't override a significant efficiency gap. It matters most in close matchups where both teams have similar adjusted ratings.

Use Our Game Model for Independent Projections

Our CBB game model projects spreads and totals for each tournament matchup using adjusted efficiency, pace, and schedule context. When the model's projected spread diverges from the market by 2+ points, that gap warrants a closer look at the specific matchup.

Pace Mismatch Creates Value

When a fast-tempo team (top 50 in possession rate) faces a slow team (bottom 50), the resulting pace will typically land between their averages — but the total may be set at the slow team's expected pace, creating over value if the fast team controls the game tempo. Check both teams' possession rates on Barttorvik before taking totals.

Offensive Efficiency vs Defensive Efficiency

Calculate the expected score for each team by applying their adjusted offensive efficiency against the opponent's adjusted defensive efficiency. Add the two projected team scores for a raw total estimate, then compare to the posted total. A 3-4 point gap between your model and the posted total is meaningful in a tournament context where every point matters.

Analytically Supported Upsets

A 12-seed with a top-40 KenPom efficiency margin facing a 5-seed with a top-60 margin is a genuinely competitive game. The market often undervalues these because casual bettors back the higher seed out of habit. When the efficiency gap is small, look at the spread as potential 12-seed value.

Pure Variance Upsets

A 15-seed beating a 2-seed is typically a variance event — the 2-seed had a bad day, the 15-seed hit unsustainable shooting percentages. These are not predictable and not worth chasing. The 2-seed cover and spread line remains the expected-value play even when upsets occasionally happen.

First Round: Exploit Mismatches, Avoid Blowout Props

The first round features the widest talent gaps of any round. 1-seeds vs 16-seeds and 2-seeds vs 15-seeds are near-automatic covers historically. The value lies in identifying competitive games (5-12, 6-11, 7-10 matchups) where efficiency data suggests the spread is set incorrectly.

Second Round: Efficiency Data Most Valuable

By the second round, upset-prone teams are mostly eliminated. Every remaining team is legitimately capable. This is when KenPom and T-Rank data is most actionable — teams are evenly matched enough that the efficiency details matter.

Sweet 16 Through Championship: Market Efficiency Increases

The later rounds attract the most betting handle and the most sharp money. Lines are tighter and harder to beat. Focus on player props rather than game-level bets in the later rounds — individual matchup advantages are more identifiable than team-level spread edges.

What is the best strategy for March Madness picks?

Use adjusted efficiency ratings from KenPom or Barttorvik as your baseline. Compare the model-implied spread to the posted market number. Focus on games where the efficiency data shows a larger or smaller gap than the market spread implies. Apply the same research process in each round rather than changing approach based on narratives.

How accurate are efficiency ratings in the NCAA Tournament?

Adjusted efficiency ratings are the most predictive metric available for tournament outcomes over large samples. Individual game predictions are still uncertain — variance is high in single elimination — but teams with superior efficiency margins cover more often than the market implies, especially in the first weekend.

Do 12-seeds really beat 5-seeds that often?

The 12-5 upset is historically the most common bracket upset, occurring roughly 35% of the time since the tournament expanded to 64 teams. It happens because 5-seeds from major conferences often carry weaker adjusted efficiency ratings than their seeding implies, and the 12-seed from a strong mid-major can be genuinely competitive.

Should I use the same research process for all 67 games?

Yes. The same analytical process — efficiency comparison, pace evaluation, matchup context, line comparison — applies to every game. The later rounds require more specific matchup analysis because team quality is more even, but the framework stays the same.