Ranked War Intelligence · v0.1
Scout your next Ranked War
before the first attack lands.
A deterministic prediction engine for Torn.com faction wars. We model actual roster matchups — not just totals — to estimate who wins, why, and what needs to happen for the other side to flip it.
WIN PROBABILITY
0.02 – 0.98
logistic on weighted diff
MATCHUP MATRIX
N × M
pairwise attackability
TARGET COVERAGE
per player
% of viable enemy
ROSTER RETENTION
last 5 wars
dep. contribution excluded
TOP-HEAVY DETECT
scarcity
elite bottlenecks
CONFIDENCE
HIGH → V.LOW
data-driven, not vibes
signal
Target coverage, not just totals
A player with elite battle stats but only 2 viable targets contributes less than a mid-tier player with 40 viable targets. The model treats target access as a first-class feature.
signal
Real roster reconstruction
Historical performance is only projected forward for members still in the faction. Departed contributors show up in the war record but never inflate current strength.
signal
Transparent probability
A logistic model over weighted features — historical output, matchup, target access, participation, stability. Every number is explainable, and every weight is tunable.