The Preseason Issue

Start from each team's current rating, the one earned across every match through the 25-26 season. Rosters turn over in the offseason, so the engine hands back part of every team's distance from average before a new season starts, and it does not do it blindly: it regresses a team that lost most of its squad harder than one returning a young core. That is the same adjustment the live engine will apply in August; this page just runs it early. (The exact weights stay behind the official curtain, as always. The newsletter explains the structure.)

How it measures that turnover now depends on whether a school has posted its 26-27 roster. Where one is up and looks complete, the page counts the players who actually came back, matching last season's squad name by name and requiring each to advance exactly one grade. Where it is not, it falls back to the old proxy, the share of last season's roster that were seniors and therefore eligible to leave. The difference is not cosmetic: measured turnover runs to about 41 percent of a squad, where graduation alone accounts for roughly 24. Teams lose more than the senior class, and until now the projection did not see it. As of this run, these numbers are read from the projection itself.

Then the postseason: for each class we take the sixteen strongest projected teams, seed them, and play the bracket ten thousand times, with every result drawn from the same calibrated win-probability scale the head-to-head tool uses.

The projected rating is an estimate, and the simulation now says so. Each simulated season draws every team's rating from the spread that projections have actually missed by: measuring 6,108 girls team-seasons since 2010, the real end-of-season rating lands about 61 points from where the offseason regression put it, and about 80 points for the teams good enough to contest a state tournament. Eighty points is roughly the whole gap between the top two seeds in AAA. That is why no favourite here clears fifty percent. An earlier version of this page held the ratings fixed and ran a steeper win curve than the engine's own, and it read Bearden at 71 percent. That was too confident, and the number moved to 43 once both were corrected.

Three limits worth naming. The 26-27 rosters are not out. What the regression reads is the 25-26 roster, so it knows who graduated, not who came back, and several programs have published nothing for the coming season yet. Retention, transfers, incoming freshmen and coaching moves are all invisible in July, so expect these numbers to move once squads are posted. And the real TSSAA draw is geographic, not seeded, so read these as strength odds, not bracket forecasts.

Every one of the girls programs with a 26-27 projection is here, not just the contenders. Type a school name to see where the model has them, and what share of ten thousand simulated seasons they reach the state tournament in. A team on the bubble is the interesting case: qualifying is the hard part, and most of the field never gets a title number worth reading.

Every class on one ladder, ordered by projected rating. The "returning core" column is the share of last season's roster that graduated, read straight from the 25-26 rosters: lower means more of the team comes back.

Regression is not even-handed, and that is the point. A team that just graduated half its starting eleven should not keep its whole rating over the summer; a team that dominated with sophomores should. These are the biggest projected rank moves among the current top 60, before anyone plays a match: