Seeding
Points are category wins plus half a point per tie, which is what Yahoo's Pct column is built from. The top two seeds sit out the first round.
| Seed | Manager | Team | Points | Banked | This week | All-play | Last 8 | Last 4 |
|---|
The Bracket
Title Odds
Twenty thousand runs of the whole bracket. Each week is played by drawing a week each team actually had this season and scoring the twelve categories against each other, so the odds carry a team's real range rather than its average. Rounds Yahoo has already decided are not simulated.
| Seed | Manager | Reaches semi | Reaches final | Wins it | Odds |
|---|
This Round
Each matchup predicted three ways — the whole season, the last eight weeks, and the last four — plus a blend that weights them . When the three disagree, that disagreement is the story: a team that is much better over four weeks than over twenty is either genuinely transformed or on a heater about to end. Click any of the four boxes to switch the whole card to that window — the headline odds, the win/tie/loss bar and every category bar all move together — and watch which categories are actually doing the work.
Any Matchup
Reseeding means the second round is not knowable until the first is played, so here is every pairing that could still happen.
How the Prediction Works
A week is drawn whole. To play one matchup, the model picks a week each team actually had and scores its twelve categories against the other's. Drawing the whole week matters: a team's ERA, WHIP and K/9 all move together with how its starters went, and a model that drew each category on its own would happily deal a team its best ERA of the year alongside its fewest strikeouts — a week that has never happened and never will.
The innings rule is simulated, not smoothed over. A drawn week under 50 IP forfeits all five pitching categories, exactly as Yahoo scores it. Managers who habitually run short carry that risk into their odds rather than getting credit for tidy rate stats posted on 35 innings.
Ties go to the better seed, which is the league's rule. A 6–6 week is not rare here, so it is worth knowing which way it falls before it happens.
Odds are American moneyline. Negative is the favourite — −150 means risk 150 to win 100; positive is the dog — +200 means 100 returns 200. A matchup is priced three ways — win, tie, loss — with the tie quoted next to its probability under the bar, because a 6–6 week is a real result here and not a push.
What it does not know. Injuries, call-ups, who is actually setting a lineup in September, and the two-start weeks that decide most pitching categories. A two-week window is eleven weeks of noise short of meaning anything on its own, which is why it is shown next to the season rather than instead of it.
The regular-season race that produced this bracket is preserved on the week 22 scenarios page.