How it works
The poll
- Every bot ranks every team in all leagues, using regular-season weeks 1–11.
- One scale. Each bot's rating is turned into standard deviations from the average team (z-score, capped at ±3).
- Best and worst dropped. For each team, its single best and single worst bot are thrown out (BCS style); the rest are averaged. That average is the Score: 0 = average team, +1 = one standard deviation better.
- Top 25 = the 25 highest Scores. Ties: Score, then average bot rank, then Scoreboard.
- First-place votes = how many bots rank the team #1.
- Others receiving votes = outside the top 25 but in at least one counted bot's top 25.
- Missing data is bridged, not left blank. A bot missing a week's input uses a straight line between the nearest weeks with data, or the nearest week's value when there is only one side; those ballots are marked * on the poll page. A bot with no data at all for a season sits out (shown –) and the other bots carry the poll.
- Final poll (after the postseason): postseason games count as extra games for Scoreboard; Vegas, Bandwagon, SOR, Headliner, Hammer, Bagman, Professor stay at the end of the regular season.
- Postseason bumps added to Score per win: playoff rounds +0.15/+0.25/+0.45/+0.85/+1.65, conference title game +0.15, bowl +0.15, NIT +0.05 (× share of rivals outscored).
- Final-four backstop: a semifinal or final winner always ranks directly above the team it beat.
The bots
Scoreboard SCB
“I only care about the standings.”
Measures: Exactly the playoff rank: wins, then points for
How: Wins (ties = half), then official total points as the tiebreaker - the league's real playoff order. Final poll: postseason games are added as extra games.
Vegas VEG
“Strip out the schedule luck and tell me who is actually good.”
Measures: Expected wins vs. every other team, every week
How: Each week, the share of all 180 teams you outscored (not just your opponent). Summed over the regular season; points for breaks ties.
Bandwagon BAND
“Whoever is hot right now.”
Measures: All-play over the last 3 weeks, weighted 3:2:1
How: All-play share for the last 3 weeks, weighted 3:2:1 (most recent heaviest); season all-play breaks ties.
SOR SOR
“Anyone can win. Win where a contender would stumble.”
Measures: Strength of Record: how hard your record would be for a top-25 team to match
How: Benchmark = the top 25 teams by season all-play. Each game: the share of that week's benchmark scores that beat your opponent's score. Combined = the chance a benchmark team matches your record or better against your exact schedule; the lower, the better. Regular season only (frozen at week 11 for the Final poll).
Headliner HEAD
“Star power: the most points this roster could have put up.”
Measures: Best possible lineup per game
How: Average weekly points of the best legal lineup from your own roster (what you could have scored).
Hammer HAM
“Close wins don't impress me.”
Measures: How hard you win (and lose), ranked against that week's games
How: Each game: a win scores where its margin ranks among that week's winning margins (biggest blowout +1, median win +0.5, a squeaker about 0); a loss scores minus the same for that week's losing margins; a tie 0. Season value = the average. Regular season only.
Bagman BAG
“Follows the money: what the market says your starters are worth.”
Measures: Redraft market value of the best starting lineup (FantasyCalc)
How: FantasyCalc redraft value of your best legal starting lineup, from the latest value snapshot. Before 2026 (no market history): Sleeper's projected starter points for the next week.
Professor PROF
“Grades on a curve that shifts as the semester goes on.”
Measures: Starts with preseason roster strength, rewards quality weeks midseason, finishes on the resume
How: Three blocks, each built from percentile ranks across all 180 teams; the block weights shift with games played (table below).
Prior
| Preseason starter value (redraft value of the best lineup) | 30% |
| Week-1 projected starter points | 50% |
| Preseason dynasty team value | 10% |
| Last season's final rank | 10% |
Set once at the start of the season and never refreshed. Last season's final rank = the final poll, or the standings if there was no poll. A team missing an input (no preseason values that year, a new league) gets the other inputs' weights scaled up to 100%.
Quality
| All-play win % vs all teams, weekly | 50% |
| Points per game | 30% |
| Weeks above the all-team weekly median | 20% |
Resume
| Win % | 70% |
| Wins over opponents in the top half of Quality that week | 20% |
| Division / conference standing | 10% |
Block weights by games played (straight line between rows)
| Prior | Quality | Resume | |
|---|---|---|---|
| Preseason | 100% | 0% | 0% |
| Week 1 | 70% | 25% | 5% |
| Week 3 | 40% | 40% | 20% |
| Week 6 | 15% | 45% | 40% |
| Week 9 | 8% | 32% | 60% |
| Week 11 | 5% | 25% | 70% |
Every measure is a percentile rank across all teams that season. Not used: lineup efficiency, max points, points against, week-to-week spread.