Power Rating

What is Power Rating?

Power Rating is a Bayesian estimate of a team's true scoring strength at any point in the season. It represents the system's best guess at how many points a team would score in a typical week, accounting for recent performance, sample size, and league-wide context.

Unlike raw points per game, Power Rating corrects for early-season noise and year-over-year scoring inflation. It is the foundation of the Playoff Odds and Championship Odds simulation engines — every simulated game draws from the Power Rating model.

How Power Rating is calculated

Recency-weighted scoring history

The system looks at every game a team has played and weights them by recency using an exponential decay with a 14-week half-life. A game from last week has full weight. A game from 14 weeks ago has half weight. A game from 28 weeks ago has a quarter weight.

Games from the prior season are further penalized during the early weeks of the current season, then gradually trusted more as more current-season data becomes available. By about Week 5, prior-season data reaches its natural decay weight with no additional penalty.

Bayesian shrinkage toward league average

Early in the season, a team might be 2-0 and averaging 140 points per game. But two games is a tiny sample — that team could be elite, or they could have gotten lucky twice. Power Rating handles this by shrinking each team's observed average toward the league-wide average, as if every team had played 5 additional games at the league mean before the season started.

As a team plays more real games, the shrinkage effect fades and their actual performance dominates. By mid-season, Power Rating is almost entirely driven by the team's real results.

The formula:

Power Rating = (w x Team Average) + ((1 - w) x League Average)

Where w increases with sample size: w = games_played / (games_played + 5)

• After 5 games: w = 0.50 — half team, half league
• After 10 games: w = 0.67 — two-thirds team
• After 15 games: w = 0.75 — three-quarters team

Variance floor

Each team's scoring volatility is also estimated and floored at 80% of the league-wide standard deviation. This prevents the model from being falsely certain about low-variance teams. Even a consistent team can have an unexpectedly high or low scoring week, and the model accounts for that.

Inflation adjustment

In leagues that have changed scoring rules over time (e.g., switching from standard to PPR), raw points per game aren't comparable across years. Power Rating is normalized by an inflation rate — the ratio of the current year's league-average scoring to the first year's. This keeps multi-year Power Rating trends meaningful.

How Power Rating is used

Interpreting Power Rating

Power Rating is expressed in the same units as fantasy points. A Power Rating of 120 means the system expects that team to score about 120 points in a typical week under current league scoring.

Season and career Power Rating averages are shown in the Team Stats section, often alongside percentile tiers:

• Elite: Above the 90th percentile of historical Power Ratings
• Good: 75th to 90th percentile
• Average: 25th to 75th percentile
• Below Average: Below the 25th percentile

See Power Ratings in your league

Import your league for free to see how every team's Power Rating has evolved week by week across your league's history — and how it predicts the rest of the season.