How We Make the Call

How we make predictions

PrepRank predicts every game and ranks every team using historical results, opponent strength, recent performance, and other factors. Our model is updated continuously as new games are played.

Two numbers, never blended

PrepRank Index. Our predictive power number. It drives the rankings and every game and playoff projection on this site, and it is the number every accuracy and calibration claim below attaches to — including the playoff-qualify band that locks at ±5 percentage points by Week 9. It is PrepRank's own metric, built from full-season opponent strength.

Official LHSAA Rating. A separate view that reproduces LHSAA's official rating exactly — the seeding math from the LHSAA handbook (game result plus play-up plus opponent strength, averaged over counting games). It is an exact arithmetic figure, not a calibrated probability, and we show it only for completed, published seasons. In-season, LHSAA's official ratings update weekly and finalize at season end.

The two are never blended: the PrepRank Index is never labeled “official,” and the Official LHSAA Rating is never a PrepRank prediction.

How to read confidence. Every PrepRank projection comes with a confidence level — High, Moderate, Low, or Very low. This is a statement about our data, not about the game. Confidence reflects how much historical evidence stands behind a projection: specifically, how tightly we can pin down the win probability given how many comparable games we've seen at this projection strength in this sport. A High-confidence projection rests on a deep historical sample, so the margin of error around our number is small. A Very-low-confidence projection rests on a thin sample, so the margin of error is wide.

Confidence is not the same as how lopsided the matchup is. A team favored at 75% can carry a Low confidence label if we have little historical data at that projection strength — and a team favored at just 54% can be High confidence if we've seen that situation many times. Confidence never changes who we project to win. It only tells you how much to trust the number.

Tier definitions

Every PrepRank prediction comes with a confidence range. We label each prediction based on how wide that range is — the wider the range, the less certain we are.

When we hold back

When we don't have enough data to be confident, we say so. Rare matchups, brand-new teams, and prediction ranges where we've seen few historical examples all reduce our confidence. We'd rather be honest about what we don't yet know than overstate certainty.

Per-sport accountability

We check each sport's predictions against actual game outcomes. Status reflects whether our predictions matched observations, plus how much historical data the sport has.

Playoff probability calibration

PrepRank's Football playoff-qualify probability is shown from early in the season, with an honest confidence band that tightens as games are played and locks at ±5 percentage points by Week 9 of the 10-week regular season (when we say 70%, it's 65–75%). Earlier weeks carry wider bands, labeled plainly. Deeper-round predictions (Round of 16, Quarterfinals, Semifinals, Finals, Champion) are shown directionally; calibrated validation arrives with multi-season holdout data in v1.1.

Why the band tightens: more games played → power ratings stabilize → the playoff simulation gets sharper, so the confidence band narrows week by week.

Coverage reflects the rateable division universe. The smallest classes (Class B, Class C, Division V) are validated as data coverage expands in v1.1.

Source-data note (Baseball)

Baseball games on lhsaaonline.org are sometimes recorded with only the winning team's runs. PrepRank's Baseball forecasts use win/loss outcomes (which aren't affected), and Baseball game cards carry an additional source-data note alongside the prediction.