Matchups

For every investigator who has logged at least 5 games against a particular Ancient One, we compute the win rate, shrunk toward the global mean (59.8%). Best matchups (n ≥ 30) lead; worst follow.

Win rate by matchup

Color = win rate after Bayesian shrinkage. Hover a cell for n and the raw rate.

N= 96,488 HOW?

Hover a row below to spotlight that matchup in the heatmap above; click to pin.

Best matchups (n ≥ 30)

N= 317
  • Agatha vs Rise of the Elder Things87.9% (48 games)
  • Bob vs Rise of the Elder Things83.6% (51 games)
  • Jacqueline vs Antediluvium83.3% (86 games)
  • Agatha vs Shub-Niggurath83.3% (44 games)
  • Ursula vs Hypnos82.6% (88 games)

Worst matchups (n ≥ 30)

N= 562
  • Michael vs Cthulhu33.8% (55 games)
  • Finn vs Cthulhu36.2% (106 games)
  • Harvey vs Cthulhu37.5% (62 games)
  • Mark vs Yig37.9% (307 games)
  • Vincent vs Cthulhu38.0% (32 games)

The cost of a matchup

A 60% win rate against Cthulhu feels different from a 60% win rate against Yig if one of them leaves the table littered with corpses. Win rate alone misses the bodies.

Empty chairs by Ancient One

Each bubble is one Ancient One. X = win rate. Y = average investigators lost (defeated or devoured) per team member. Bottom-right is the dream: you win, and everyone goes home. Top-left is the nightmare: you lose, and you lose everyone.

N= 21,132

Modifiers: preludes that flip a matchup

Picking the right prelude can swing a matchup by more than picking the right investigator. Each cell shows how the win rate against a given Ancient One changes when a particular prelude is in play vs. when it isn't.

Prelude × Ancient One swing

Difference in win rate, with prelude minus without, for every (prelude, AO) pair with enough games on both sides. Green cells = this prelude makes that matchup easier; red cells = it makes it harder. Showing the 30 preludes with the biggest total swing across matchups; sparse cells are dimmed.

N= 9,306

Partnerships that change the matchup

Some investigator pairs win more often together than either of them does apart. Below: the pairs whose joint win rate most beats (or trails) their solo win rates. Anything to the right of the dashed 1.00× line means the pair is greater than the sum of its parts.

Pairs that lift each other

Top synergy pairs (n ≥ 20 shared games). Lift = win rate together divided by win rate apart. Whiskers are 95% confidence intervals — green whiskers mean the lift is clearly above 1×.

N= 409

Pairs that drag each other down

Bottom synergy pairs (n ≥ 20 shared games). Red whiskers mean these two investigators win less together than they do separately — possibly fighting for the same items or covering the same ground.

N= 421