Survivor & Matchup Lab · Experimental
Play the matchup.
Explore the possibilities.
Simulate two NFL teams with today's saved ratings, or plan a no-repeat survivor path through the season.
NFL scoring estimates—not fantasy points. Injuries, roster changes, and schemes are not yet modeled.
Head to head
Who wins this matchup?
Expected scoring & uncertainty
SD (standard deviation) describes the spread of simulated game outcomes—not how precisely we know the average. The middle 80% range runs from the 10th to the 90th percentile; outcomes outside it are still possible.
Outcome distribution
Scroll inside the chart to see the full scoring range.
How might the game finish?
Current-season results & team strengths
Season averages describe completed games, not this matchup's prediction. Before a team has played, those averages stay blank.
This week's betting lines
How these game simulations work
Expected scoreboard points come from the existing opponent-adjusted team model. We add paired scoring errors from forecasts made before each historical week, using the prior two regular seasons and giving recent games more weight. Errors are centered around zero; both teams' errors stay together. A neutral field removes home advantage and treats the error roles symmetrically.
Scores are rounded and cannot be negative. Decisive score scenarios are sampled separately from a small historical tie component, so rounding does not create excessive ties. This approximates final scoreboard outcomes; it does not simulate drives, overtime rules, touchdowns, yards, or individual player box scores. Probabilities can differ from the survivor matrix's simpler normal-margin calculation.
20,000 trials use a repeatable seed. The ranges reflect historical game variation, not uncertainty in team ratings, and have not been calibrated as prediction intervals. Injuries, starters, weather, schemes and roster changes are not included yet. Future games hold current team strength fixed. These estimates are research tools, not proven betting edges or guaranteed outcomes.
Lines are attached only to the same upcoming fixture in the current schedule week (within eight days). Dated sportsbook and exchange pairs older than 48 hours are withheld. Undated schedule reference lines are labeled separately and are not live quotes. Betting lines do not influence these simulations. Reloading reads saved files and makes no sportsbook API call.
Your pool
Set the planning window
Saved only in this browser.
Teams you have already used
Mark previous picks here. A used team cannot be selected again. Marking one removes it from your planned picks.
The season at a glance
Weekly survival matrix
Select a cell to pick that team for the week; select it again to remove it. Swipe sideways for later weeks. Percentages use the selected tie rule.
One pick per week
Your planned path
Filling empty weeks keeps your existing picks and reserves every team only once. Picks outside this window also reserve their teams. This is a planner, not a pool-entry service.
Team offense and defense ratings
Points above a league-average team after accounting for opponents and home field. Positive offense adds scoring; positive defense prevents it. These are scoreboard-based team ratings, not individual-player grades.
Know the assumptions
A baseline we can measure and improve
Team scoring: Opponent-adjusted offense and defense fitted to regular-season scoreboard points from the prior three seasons and completed current-season games. Recent results get more weight (365-day half-life); ratings are pulled toward league average. Neutral sites have no home advantage. Scores are treated as completed only when present and at least six hours past kickoff.
Win estimates: Expected point margin is compared with historical margin variation. A small, smoothed historical tie rate is included. Conditional win estimates are bounded at 2.5–97.5%. These experimental probabilities are not calibrated guarantees; future weeks use today's team strengths.
Planning and simulations: The suggested path maximizes the product of weekly survival probabilities with no team reuse. Weeks are assumed independent. Simulations sample one shared result per game, including ties, and compare your path against a fresh best path under these same assumptions. They do not model other pool entrants, popularity, adaptive picking, or the chance of winning the whole pool.
Survivor path market comparison: Only complete opposing moneylines from the same licensed sportsbook, timestamped within 48 hours, for games within eight days. Bookmaker margin is removed per pair, then the median share is shown. It is conditional on a non-tie, not a guarantee or an input to this model. Missing coverage stays blank; exchanges and undated reference lines are excluded from this path comparison.
Next model layers: Current-season efficiency, verified injury/availability adjustments (starting with quarterbacks), roster and coaching changes, then pass/run tendencies and defensive personnel. Individual defenders and scheme adjustments are not included yet. No injury data does not mean a team is healthy.
Data is refreshed through the site's existing four-times-daily update workflow. Reloading this page reads a saved snapshot; it does not spend sportsbook API credits. Schedules can change—confirm kickoff and your pool's lock rules.