Across leagues and seasons
We combine the imported regular-season records from your Redraft and Keeper leagues. You qualify once that total reaches 26 games.
Example: 14 games in one league + 12 in another = enough to qualify.
Winning matters.
Context does too.
Your Sleeper score combines wins, all-play scoring, opponent strength and championships.
It uses an explicit weighted algorithm, with separate rankings for each league type.
01 / THE STARTING LINE
Two leaderboards. Two entry rules.
In a single dynasty league
You need a full completed regular season in at least one Dynasty league. Several partial seasons do not satisfy that requirement.
Example: Two half-seasons do not qualify. One complete season does.
What does NR mean?
You have not met the entry rule or the core-evidence requirement for that leaderboard. Your rank is blank, and your Sleeper score cannot exceed 30. Your imported record still appears.
The two groups stay separate: Dynasty games do not count toward Redraft / Keeper qualification, or vice versa. Qualification uses the records we have imported. A published rank also requires 13 rated games, 13 complete scoring weeks, and at least 80% coverage of known eligible matchup records, scoring weeks and league records. Meeting the game threshold alone does not prove sufficient evidence.
02 / WHAT COUNTS
More than your win–loss record.
Four signals determine your score. Each has a fixed weight: missing data never increases the weight of the others. Related statistics remain visible, without giving the same result several bonuses.
| Metric | What it tells us | Redraft / Keeper | Dynasty |
|---|---|---|---|
| Winning | Win percentage from the imported career record shown on the leaderboard; ties count as half a win. | 30% | 40% |
| All-play scoring | How often you outscore other teams in the same league-week. Only weeks with every expected roster score enter the score. | 40% | 30% |
| Opponent strength | The strength of opponents before each matchup. Future results cannot improve an earlier opponent’s rating. | 20% | 20% |
| Championships | Verified titles divided by known completed league-seasons. Missing brackets earn no title credit until verified. | 10% | 10% |
| Total | 100% | 100% |
Performance, achievements, confidence.
Championship rate contributes 10% of the score. Titles and playoffs also have their own achievement index. PF/PA, margins, quality wins, consistency and league difficulty remain useful context, with no extra performance bonus. Confidence describes the evidence and never multiplies the performance score.
One matchup doesn’t tell the whole story.
You score the second-most points in a 12-team league, but happen to face the top scorer. Your actual record that week is 0–1. Your all-play record is 10–1: you outscored ten of the eleven other teams.
All-play adds context to scoring. It does not add wins to your actual record. Ties count as half a win in all-play percentage.
Did you score more than you allowed?
If you scored 1,200 points and opponents scored 1,000 against you, your PF / PA ratio is 1.20. You scored 20% more points than your opponents in those imported matchups.
Average winning margin is also shown on the leaderboard. It is a display statistic; the scoring-margin metric includes both wins and losses.
03 / FROM METRICS TO A SCORE
Three steps. One Sleeper score.
- 01
Measure the available results
Turn wins, all-play and opponent strength into 0–100 values, then add verified championship credit. If a core metric is missing, do not publish a score. If coverage is too incomplete, withhold a rank and apply the NR cap to any available score.
- 02
Combine them using the weights
Multiply each value by its share, then add the contributions. For example, an all-play value of 60 at a 40% weight contributes 24 points to the raw index.
- 03
Convert the index to the final scale
A fixed conversion turns the raw index into your Sleeper score. An index of 45 becomes 80; 55 becomes 88; 70 becomes 95. The scale does not change just because a new manager takes the lead. Unranked scores are then capped at 30.
Your score is not your win percentage.
A score of 88 does not mean an 88% chance of winning, or that you beat 88% of managers. Your rank is your position among qualifying managers in the observed group.
What would it take to score 100?
Perfect metric values, a verified complete career and a win in every game, with no losses or ties. Dynasty also requires every league to reach at least 65/100 difficulty.
Current imports cannot verify complete career coverage, so their maximum is 99.9. These are scoring rules, not a claim that any imported profile has perfect evidence.
Show the exact conversion and formulas
The raw index maps to these fixed anchors, with straight-line interpolation between them:
0 → 0 · 20 → 35 · 35 → 65 · 45 → 80 · 55 → 88 · 70 → 95 · 85 → 98 · 100 → 99.9
Championship and playoff indices: 50 × observed successes / expected successes, capped at 100. Expected title rate is one divided by league size; expected playoff rate is playoff spots divided by league size. Only complete, resolved championship brackets contribute postseason evidence. The separate achievement index is 65% title index plus 35% playoff index; it is unknown without postseason evidence and is displayed separately. The score itself includes a 10% championship component: 100 × verified titles / known completed league-seasons. Missing brackets remain unknown and earn no title credit.
Scoring-margin index: 50 + 100 × average((PF − PA) / (|PF| + |PA|)) across eligible matchups. The PF/PA index uses the same normalized comparison on aggregate totals. Both are bounded to 0–100 and retained as descriptive statistics, not performance-score inputs.
Quality wins require an opponent rating of at least 1550 and at least 20 recorded games before the matchup. Elo supplies opponent context; it is not the final Sleeper score, and there is no seasonal regression toward an average.
The weights and conversion are explicit product choices. They have not been validated as predictions of future performance.
VALIDATION / WHAT WE KNOW
Tested for honesty. Still experimental.
We froze all data before each held-out season and compared v5 with the preserved v4 algorithm on the same 129 supported matchups. Higher-score accuracy was 43.0% for v5, 41.9% for v4 and 43.0% for a simple win-percentage baseline. Ties receive half credit. Only 56 comparisons had both managers eligible for a v5 rank.
This small, incomplete sample repeats managers and leagues. It does not demonstrate predictive superiority. We kept the existing score conversion unchanged: it remains a display scale, not a calibrated probability. The release improves missing-data handling, avoids overlapping bonuses, and keeps historical league strength free of later results.
Read the backtest results ↗How confidence and league-rank eligibility work
Confidence combines known matchup coverage, complete scoring-week coverage and record coverage with sample depth. Sample depth uses distinct season-weeks (up to 52), league series (up to 3) and seasons (up to 3). Concurrent leagues in the same week do not count as independent weeks. Limited network connections reduce confidence. This is an evidence index, not a statistical confidence interval.
A league needs at least six observed weeks and 80% of expected teams observed. A published league rank also needs five qualifying leagues in the same format group and season. Smaller groups return a null rank with an insufficient-evidence status; the observed position remains available as context.
04 / KNOW WHAT YOU’RE LOOKING AT
Your history is still coming into view.
Discovered is not fully evaluated
Finding a username does not mean we have imported every league or season. Counts and ranks reflect the data currently available, not the entire Sleeper population.
Records and scoring coverage can differ
Career records include available regular-season standings in the relevant group. Some leagues may count toward the record without being eligible for the scoring model. Playoff games are not added to the regular-season record; verified titles and playoff appearances are separate metrics.
Provisional is different from unranked
You can meet the entry rule and still have a provisional score. That means evidence confidence is below 75 or the underlying league connections remain limited. More evidence can change your score and position.
Missing games are not zero games
If your record looks short, search for your username on the rankings page and request your full career history. Imports follow available Sleeper seasons and linked previous leagues. A completed import still cannot guarantee every lifetime result is available.
COMMON QUESTIONS
Why am I ranked here?
Why am I behind someone with fewer wins and a lower win percentage?
Wins are part of the score. All-play results, opponent strength and championships also count. Someone can win fewer games but score better against the rest of their league or face tougher opponents. Open either profile to see how each part adds up.
Do championships count?
Yes. Championship rate is 10% of your score in both groups. Two titles in five known completed league-seasons gives a 40% title rate and adds 4 points to the raw index before it becomes your Sleeper score. We only credit titles confirmed by a resolved championship bracket.
Why not just count total wins and rings?
Playing in more leagues gives you more chances to win games and titles. Rates let us compare people who play different numbers of leagues. Your total wins and rings still appear on the leaderboard.
Can someone with fewer championships rank above me?
Yes. Titles matter, but they are 10% of the score. Better results across the other metrics can outweigh a lower title rate. The same rules apply to everyone.
What if my championship is missing?
Search your username and request your full career history. Until we have a verified bracket, that title earns no credit. Known completed seasons still count in the denominator, so missing results can lower your score. We show unknown title totals as unknown, not as verified zero titles.
Do playoff appearances, PF / PA and winning margin count?
They appear on your profile and leaderboard. Playoff appearances also count toward the separate achievement index. They do not currently add points to your Sleeper score. All-play already captures how well you score against your league.
What does all-play mean?
It asks how you would have done against every other team that week. If you outscore 8 of the other 11 teams, your all-play record is 8–3, even if you lost your actual matchup.
Why am I unranked or capped at 30?
Redraft and Keeper need 26 imported games. Dynasty needs one full completed season. Both also need enough matchup, scoring and league data. Until those requirements are met, you have no rank and your score is capped at 30.
Does high confidence give me extra points?
No. Confidence tells you how much evidence supports the result. It does not add points. A provisional manager can rank above you, but their result has more uncertainty.
Is this my rank across everyone on Sleeper?
It is your rank among qualifying managers we have evaluated in your group. Finding a username is only the first step. We still need to import their results. The goal is to cover the full user base, but coverage is not complete yet.
The published backtest covers career-v5, before championships were added. Career-v6 has not been validated for predictive accuracy.