The Guillotine League Companion
Your league, survival curves, and next move. Keep your team above the weekly chop line.
League facts come from Sleeper. Odds and curves are simulations, not live scores.
Choose a league to get started.
Paste a Sleeper league link or ID. Read-only access; no Sleeper login needed.
Every bar is a team. The taller the bar, the likelier the chop.
Connect a league below and the ladder fills in.
Compare this week’s starters, free agents, and survival odds using your league’s scoring.
Connect a league to begin.
Current Sleeper starters, not automatically optimized opponents. Rank is probability of surviving this week. The score range is a modeled 80% interval. Click a team to investigate.
Pregame forecast, not in-game win probability. Injury designations of Out, IR, PUP, suspended or inactive score zero; questionable players retain Sleeper’s projection. Correlation uses a historical same-team estimate; opponent and negative teammate correlations are not modeled.
Tests legal combinations against the same opponent simulations. Players whose games started cannot be moved into or out of your starting lineup. Suggestions do not submit changes to Sleeper.
This is your adjustable spending limit, not a prediction of winning bids. Default 15% is a planning assumption. Increase it when survival matters more than preserving cash.
Only completed Sleeper waiver transactions. We cannot observe losing bids. Historical ranges describe winning bids at the same position; they are not player-specific clearing-price predictions. A separate league’s history is never silently pooled with yours.
Holds current weekly player strength fixed, applies known future byes, and reselects legal starters each week. No future acquisitions, changing player roles, or injuries are forecast. This scenario illustrates roster exposure; it is not a calibrated championship forecast.
The daily job freezes the public league’s current-starter probabilities before the first kickoff, inside a 24-hour gate. It never backfills a missed prediction. Actual completed-week scores determine grades; ties follow Sleeper’s chopped rules. No player names or ownership identities are stored in this ledger.
Historical projection errors estimate uncertainty, but historical Sleeper projections can be revised. They do not prove predictive accuracy. The prospective ledger measures the current model going forward. Every player’s expected points use this league’s scoring weights; no season-total-to-weekly division.
One elimination per week. Sleeper uses total season points to break ties; Week 1 ties eliminate the team with the better draft position. Roster-empty elimination inference is shown explicitly and can be overridden below.
Open selected league in Sleeper · Sleeper chopped rules · Dated player inputs
Each slice is a team, sized by that team’s modeled chop risk for the coming week. Spinning draws from that distribution — it is a way to feel the odds, not a second opinion about them.
Pick the team you think goes this week and lock it in. Once the week is scored the page marks it right or wrong on its own and keeps your running record. Saved on this device only — nobody else in the league sees it, and it does not follow you to another phone.
Checking this week’s first kickoff…
Your focus team is a private preference, not verified Sleeper roster ownership. Pick any team from the ladder above or the Full Board.
Each line is one team’s modeled chance of still being alive, week by week. Tap a name above to overlay it.
Turns on once two completed weeks exist — the same gate as the weekly odds.
Everything above averages thousands of seasons. This runs one and prints how it finished — press it again and it will differ, which is the point.
Two structural facts drive guillotine FAAB: supply rises every week (each chop dumps a full roster, eventually stars, into the pool) while the money dies with the chopped (unspent budget leaves the market). Talent gets cheaper all season; the only reason to pay early prices is that surviving September is worth the premium. Your branch depends on your margin over the chop line, not the standings.
| Chopped player quality | Wk 2–4 | Wk 5–8 | Wk 9–12 | Wk 13+ |
|---|---|---|---|---|
| Top-15 overall talent | 30–52% | 18–38% | 12–30% | 3–9% |
| Every-week starter | 12–25% | 6–16% | 4–11% | 1–3% |
| Floor patch / matchup | 3–8% | 1–5% | 0–3% | 0–1% |
| Placeholder / bye cover | 0–1% | 0–1% | ~0% | ~0% |
Ranges are a share of the full starting budget. This is a dated editorial framework — calibrated to one 17-team room’s observed 2025 winning bids (top bids by week: Wk 2 52%, Wk 3 21%, Wk 8 48%, Wk 13 2% of budget) and published championship case studies — not a bid engine, not model output, and not a promise about your room.
Modeled from completed weekly team scores. This is not a live Sunday leaderboard.
Injury, bye, positional-depth and star-reliance components are roster-derived and work before the season starts. Observed volatility and floor need two completed weeks.
Six weeks of 90% survival is not 90%. It is 53%. That compounding is the whole reason a weekly “On the Block” flag is not enough.
Paste a league ID. Sleeper's read-only API needs no login and no token, so there is nothing to authorize and nothing that can write to your league.
The modeled weekly minimum across surviving teams, and your margin above or below it. Quoted as a median with an 80% band, because a single number oversells what a simulation knows.
Monte Carlo over each team's completed-score baseline for one weekly stage. It does not ingest current in-game scores.
Dollars appreciate as the pool shrinks: ~$18,000 across 18 teams in week 1 is nearly worthless; ~$400 across five teams in December is not. Reads every team's remaining budget from your league.
Open Weekly decisions for player-level start/sit comparisons, weekly survival estimates, and FAAB acquisition planning using this league’s scoring.
Every elimination dumps a full roster into the pool. Who is unrostered in your league and what they actually scored in your scoring — observed points, not projections.
Reads a Sleeper draft read-only and strikes the Guillotine Board as picks land. Set your slot and the snake math counts picks until you’re up. The board itself is a dated editorial re-rank over the site’s MV snapshot — opinion, not projection.
Survival odds are a simulation from completed-score history and its variance — they assume the next week plays like a typical past week. Before two completed weeks exist the page runs in projection mode instead: Sleeper weekly player projections, scored with this league’s own settings, build each roster’s best projected lineup and every team carries an assumed ±21-point spread (if that dated weekly file is unreachable the page falls back to season projections divided by games and says so on the card) — byes, injuries and schedule are not modeled there, and nothing in that mode is observed. Once the week’s first game kicks off both modes switch to live scoring: every starter counts the points he has actually banked plus his projection scaled by how much of his game is left, the assumed spread shrinks with the clock, and the board re-polls Sleeper every 90 seconds until the last game is final. The clock itself is approximated — Sleeper’s read-only API exposes no game clock, so a game decays linearly over 190 minutes from kickoff, which is exact before kickoff and after the whistle and roughest mid-game. It still cannot see an inactive before his game starts, or know that a spike was garbage time. Your weekly vote is graded on this device: once a week has been scored the page compares your pick against whoever actually posted the lowest score that week and keeps a running record. It is still LOCAL to the device — not immutable, not server-persisted, not cross-device, and not scored with Brier history — so clearing site data clears the record. Universal Data Vault are Planned. Weak Spots reads injury designations and byes from Sleeper’s season projections and the site’s own schedule, and measures depth and star reliance off projected lineups — it cannot see depth-chart roles, snap share or a coach’s intent, and none of it is a prediction. The Long Game’s season Monte Carlo rolls the weekly model forward with rosters frozen at their current mean and spread — it cannot see waivers, injuries, byes, or the talent redistribution every chop causes, those omissions compound with every simulated week, and its finishing places cover the simulated teams only. The observed FAAB module encodes a framework about dollar appreciation, not a guarantee. The Waiver Plan sequences targets and prices them as ranges: each step assumes the steps above it were won, so the ordering carries more information than the dollars, and the dollars are modelled and ungraded rather than a bid engine. The FAAB War Plan is a dated editorial framework, not a bid engine.