86.2%
Same core team set
Within a season, two winning entries shared 17.24 of their 20 teams on average.
Loading DraftADynasty…
Loading DraftADynasty…
NFL survivor / historical research
DraftADynasty analyzed 24 complete winning paths and 480 selections from the five 20-leg Circa Survivor seasons from 2021 through 2025. The strongest recurring signal is that winners used similar team inventories but timed them differently.
measured patterns
86.2%
Within a season, two winning entries shared 17.24 of their 20 teams on average.
41.9%
The same winner pairs matched on only 8.38 exact segments. Assignment mattered more than team discovery.
76.3%
The modal winner choice was more concentrated on constrained Thanksgiving and Christmas legs.
48.5%
Endgame choices depended on each entry's remaining inventory rather than one universal best team.
These are outcome-conditioned winner-path descriptions, not causal estimates. The 2024 winner-versus-other-survivors comparison is stronger evidence than winner histories alone, and current team names never become future-season priors.
core result
Two winners in the same season shared 17.24 of 20 teams on average, yet matched on only 8.38 exact segments. That makes Survivor a constrained assignment problem: choose when to spend each viable team, preserve alternatives, and re-solve when news changes the board.
planning policy
The planner should discount farther weeks, compare multiple near-optimal paths, and retain a hard Smartypants Week 18 reserve constraint even when Week 18 lies beyond the active horizon. A preseason route is a scenario, not a commitment.
measured contrarianism
In 2024, 65% of winner selections were first or second in contest popularity, while only 3.75% sat outside the top five. Lower ownership is useful only when the reduction in shared elimination exposure justifies the survival probability surrendered.
Smartypants translation
Holiday winner convergence shows that constrained slates need explicit planning. Smartypants extends that idea: reserve health must evaluate one regular Week 18 option plus five tiebreak teams, their quality, replacement options, and joint outcome distribution.
feature registry
What is true now: survival probability, ownership, confidence, and field state.
market_win_probability_no_vigrequired-liveCurrent survival probability remains the dominant weekly input.
market_probability_dispersionderivedSource disagreement should lower confidence instead of being hidden inside one consensus number.
pool_specific_ownershiprequired-liveLeverage is created by the actual field's concentration, not generic public sentiment.
ownership_source_typerequired-liveSmartypants-specific ownership, sharp-pool proxies, public proxies, and unavailable data must remain distinct.
conditional_evderivedA useful weekly choice depends on both its own survival chance and the projected surviving field after outcomes.
data_freshnessrequired-liveA probability or ownership estimate without a timestamp cannot be represented as current.
How spending one team changes the best feasible assignments over the rolling horizon.
rolling_horizon_path_valuederivedWinning team sets converged far more than their exact weekly assignment; the path must be re-solved as information changes.
future_value_discountedderivedFarther weeks matter but should be discounted because team strength, injuries, and roles become less certain.
team_replacement_costresearch-nextThe cost of spending a team is the path value lost when it is removed from future feasible assignments.
path_flexibilityresearch-nextNear-optimal alternate assignments protect an entry when news invalidates the preferred path.
bottleneck_week_coveragederivedConstrained segments produced materially higher winner convergence and must be reserved explicitly.
How ownership concentration and shared picks change total wipeout and leverage exposure.
ownership_concentration_hhiderivedConcentration changes the equity impact of a popular-team loss even when individual win probabilities do not change.
chalk_elimination_massderivedOwnership multiplied by upset probability approximates the field mass exposed to one popular loss.
entry_pick_correlationderivedMultiple entries should be evaluated as one interdependent portfolio rather than isolated cards.
portfolio_wipeout_probabilityderivedComplete elimination risk is a first-class portfolio metric, with the outcome-dependence assumption stated.
marginal_diversification_gainresearch-nextA different choice is valuable only when the reduction in shared risk exceeds the survival probability surrendered.
Entry-specific availability, opponent scarcity, split equity, and the six-team Week 18 package.
week18_six_team_reserve_healthderivedSmartypants requires one regular Week 18 choice plus five additional unused-team tiebreak selections.
reserve_redundancyresearch-nextSix unused teams are insufficient when several are weak or highly correlated; alternate package quality matters.
availability_scarcity_vs_fieldresearch-nextLate value comes from holding useful teams that fewer opponents still have available.
solo_win_probabilityresearch-nextIn a top-heavy pool, the chance of avoiding a split can differ from the chance of merely surviving.
Quarterbacks, offensive lines, injuries, market reaction, weather, rest, and incentives.
qb_status_deltarequired-liveStarting-quarterback changes can materially reprice a matchup and its future utility.
coach_qb_continuityrequired-liveContinuity is especially useful as an uncertainty input before early-season ratings stabilize.
offensive_line_availability_deltarequired-liveClustered line absences can change both current survival probability and downstream team value.
injury_market_gapderivedNews should affect the model only to the extent it is not already reflected in current market probability.
line_movementrequired-liveMovement is evidence that the market's view changed and should trigger re-evaluation when material.
weather_deltarequired-liveWeather belongs only when it is close enough to lock to change current probabilities or uncertainty.
rest_and_incentive_uncertaintyrequired-liveLate-season rest and playoff incentives can invalidate otherwise strong historical team ratings.
myth and bias controls
Winner histories are vulnerable to survivorship bias. Road status and division-rivalry labels are already incorporated into market prices and do not receive blanket penalties. Extreme upset weeks become scenario stress tests, not automatic underdog bonuses.
always_fade_chalkalways_save_elite_teamalways_burn_low_future_value_teamwinner_team_frequency_as_current_priorstatic_full_season_pathmissing_ownership_equals_zerogeneric_road_team_penaltygeneric_division_rivalry_penaltyunderdog_bonussource ledger
Twenty-four winning entries and 480 one-team-once selections.
Open source →Popularity, current spread, and future-value differences.
Open source →Operations-research treatment of survivor path optimization.
Open source →Plain-language explanation of repeated rolling-horizon planning.
Open source →