How the NBA title odds and Super Bowl odds are calculated

The NBA odds come from a team rating that knows what a roster looked like before the season started; the NFL odds come from point differential. Both are Balladex model estimates, not sportsbook odds.

By Beau Thoutt · Used on NBA power rankings, NFL power rankings, NBA playoff picture

Inputs to the Title and Super Bowl odds

NBA: last completed season's measured net rating per team (data/nbaTeamRatings.json, a full 82 games), the current roster's per-player scouting grades weighted by minutes (with a production-based estimate for players who have no grades), current injuries, and this season's net rating from ESPN's regular-season standings.
NFL: total season point differential from ESPN standings.
When ESPN rate-limits the site, the most recent committed standings snapshot is used instead.

How NBA championship odds are calculated

Prior: last season's net rating, regressed toward zero (multiplied by 0.70), because teams revert.
Talent: the current roster's quality, from each player's scouting grades (offense 50%, defense 40%, clutch 10%) weighted by minutes, converted at 1.6 net points per standard deviation of roster talent. Injured players' minutes are redistributed to the healthy roster.
Preseason rating = 0.55 × prior + 0.45 × talent. This season's net rating then takes over in proportion to games played: its weight is games ÷ (games + 20), so it starts near zero and dominates by mid-season. Ratings are capped at ±12 net points, the most any team has sustained over a season.
Each team's chance to win its conference is a softmax over the ratings inside that conference (temperature 1.15). A Finals win probability comes from the ratings gap (spread 5.5 net points). Title odds = P(win own conference) × Σ over the other conference [P(they win it) × P(beat them)]. The board sums to exactly 100%.
Publishing floor: odds publish only for the current season, and only once teams average about 10 games played. Before that, pages show the preseason rating order and say when odds will appear.

How Super Bowl odds are calculated

For each team, compute e^((season point differential − best differential) ÷ 60).
Divide by the sum over all 32 teams and multiply by 100, rounded, with a 1% minimum.
From September on, the current season's standings are used. Earlier in the year, the last completed season's are used.

Known limitations

NBA scouting grades are editorial judgments; the talent term is only as good as they are.
The NBA model does not simulate the schedule or the play-in; conference odds come from ratings, not from remaining games.
NFL odds use point differential only — no talent, injuries or schedule — and every team is shown at no less than 1%, so they can sum to a little over 100%. In the first weeks one result moves them sharply; treat them as provisional.
Neither league's odds are calibrated against betting markets.

Title and Super Bowl odds: frequently asked questions

How does Balladex calculate NBA title odds?

Each team gets a rating that blends last season's net rating (regressed toward zero), the current roster's talent weighted by minutes, and this season's results weighted by games played. Those ratings are turned into conference-winning chances and head-to-head Finals chances, so the title odds sum to exactly 100%. They publish once teams have played about 10 games.

Why does Balladex show no NBA title odds at the start of the season?

Its publishing floor: no title odds until teams average about 10 regular-season games, so a hot first week cannot read as a title favorite. Until then the power rankings show the preseason rating order.

How does Balladex calculate Super Bowl odds?

It takes each team's season point differential and applies a softmax with a temperature of 60 across all 32 teams. Each team is shown at a minimum of 1%.

Are these sportsbook odds?

No. They are Balladex model estimates. They are not betting lines or betting advice.

Other Balladex models

Balladex Impact index — A 0–99 composite of an NBA player's skill ratings plus a capped bonus for major honors.
MVP odds — NBA and NFL MVP ladders: a production-based score per contender, turned into probabilities across the field.
Power rankings and team grades — Teams ordered by point differential, with an A–F grade from win percentage.
Fantasy projections — Forward-looking fantasy points, value over replacement, tiers and a model draft round for NFL and NBA.
Compare verdict — Head-to-head verdicts: who leads in more shared stat categories, and which pairs get a page.
NFL production index — A cross-position NFL rating: each player's production as a percentile of his position, curved to 20–99.

Source. Transcribed from lib/nbaTeamRating.js (teamRatings, titleOdds); lib/nbaModel.js (getStandings); lib/nflStandings.js (getNflStandings); lib/season.js (modelReady) in the Balladex codebase. Found an error? Email hello@balladex.com. See the full methodology, including data sources and the corrections policy.