Population results
AWAITING RUNRun a population test to measure Hero’s performance across independently generated opponents.
Simulate any fight
Build your matchup. Test your loadout. Understand the odds.
BETA
Current catalog and all listed ranked bonuses, with a new stateful combat model. Evidence limits and assumptions are listed in the register.
Saved comparisons
Different seeds sample the same configured population independently. Results with different population IDs use different generation rules, stat bounds or engine versions.
One generated matchup
Matchup configuration
Fight analysis
AWAITING RUNRun a simulation to see one complete fight here, with body markers and an action-by-action log.
Run a simulation to see statistics for the whole batch.
MECHANICS / THE SCIENCE
The Science.
Explore how battle stats shape hit chance, damage mitigation and raw damage.
Benchmarked equations
These equations have been benchmarked against Proxima’s published formulae. Crimson points show published reference values.
01 · DAMAGE MITIGATION
Strength versus defense
How the Defense / Strength ratio reduces incoming damage.
The percentage reduction from battle stats alone, before armour and other effects.
Damage mitigatedPublished reference points
02 · HIT PROBABILITY
Speed versus dexterity
How the Speed / Dexterity ratio determines your chance to land a hit.
The main curve uses a 50-accuracy weapon. Add another accuracy to compare both curves.
Enable the comparison to add a second curve.
Hit chance · 50 accuracyHit chance · 65 accuracyPublished reference points
03 · UNMITIGATED DAMAGE
Strength versus raw damage
How effective strength increases nominal damage, up to one quadrillion.
Nominal head-hit damage for a weapon damage stat of 10, before mitigation, damage variance and integer rounding.
Head damage · ×1. Published points use the same normalisation.
Unmitigated damage · headPublished reference points
04 · THE PLAYER MODEL
Every hit has a destination.
Hover, tap or focus a body region to see its share of landed hits and its damage modifier.
Percentages are conditional on landing a hit.
Regular locationCritical location
Left and right refer to the target’s body. Overlay boundaries are visual guides.
Player artwork © Torn · Original image ↗
Share of landed hits
Damage vs arm / leg
ALL 14 LOCATIONS
How the player model works
The default critical rate is 12%. Critical hits split 80% to the head, 10% to the throat and 10% to the heart. The remaining hits use the regular-location weights. Paired limbs are listed individually.
Damage is normalised to an arm or leg hit (×1). The strength graph above uses head damage (×3.5) as its baseline, so the inspector also shows each location relative to a head hit. Values exclude armour and location-specific perks such as the throat-damage education bonus. This describes ordinary location rolls; special attacks and weapon effects can behave differently.
These are community-derived probabilities and damage multipliers. Sources: Proxima’s player-model guide ↗ · Torn wiki: Attack ↗
ONE WEAPON. ONE FIXED TEST.
Evaluate your weapon.
Measure its damage, reloads and bonus activations against a silent, unarmoured Villain.
Fixed fighter setup
| Setting | Hero | Villain |
|---|---|---|
| Base strength / defense / speed / dexterity | 1,000,000,000 each | 1,000,000,000 each |
| Merit stat passives | +30% each (10 × 3%) | +30% each (10 × 3%) |
| Life / level | 7,500 / 100 | 7,500 / 100 |
| Education | All combat education | All combat education |
| Merits / experience | Level-10 mastery, 10 critical-rate merits, 100% experience | Stat passives and 10 critical-rate merits; no attacks |
| Property / company / faction | +2% property damage; no company or faction perks | No property, company or faction perks |
| Ammunition / mods | Standard ammo / no mods | No weapon |
| Armour / distraction | None / none | None / none |
Reloads are automatic and use a turn. Trials stop on a Hero win, the 25-turn limit, complete ammo exhaustion or Hero’s death; there is no switch to another weapon or fists. Both fighters start fresh in every trial. The fixed seed makes repeated evaluations reproducible. Hero’s 7,500 life also allows self-damage effects to be tracked.
The weapon
MAX EXPERIENCEDAMAGE DISTRIBUTION
Damage per turn
Each trial contributes its actual damage divided by Hero turns used, including reloads and preparation. DOT and zero-damage trials are included. Damage counts only life removed from the 7,500-life target.
WEAPON BONUSES
Bonus activations
There are no incoming attacks, armour or distraction. Defensive bonuses cannot trigger, and armour- or distraction-dependent damage benefits are absent.
Example trial
FINAL TRIALHero’s actions from one trial in this batch. Villain never takes an action.
TRANSPARENCY BY DESIGN
Know the model.
Public evidence, implementation status and the limits of the combat model. Checked through patch #455 (28 October 2026).
52Ranked bonuses modeled
74Armour pieces catalogued
185Weapons & temporary items
READ FIRST
“Modeled” means implemented against public descriptions and synthetic tests. It does not mean Torn’s private engine has been reproduced exactly. See per-mechanic evidence limits and unresolved assumptions below.
Mechanics inventory
Model assumptions & source disagreements
AUDITED 01 NOVEMBER 2026
The equipment library.
Current public catalog, corrected weapon slots, intrinsic effects and source references.
Loading catalog…
Items excluded from ordinary PvP loadouts
THE NEXT PROXISIM
A stronger foundation.
A private research build with an audited catalog and explicit combat state.
A traceable baseline
The publicly served Proxima / 811 engine, weapon data, armour coverage, attachments and company settings have been recovered. The existing calculations are preserved as a baseline, with ranked effects added separately.
What works now
Two configurable combatants; 185 weapon and temporary-item records; 74 armour records; all 52 listed general bonuses; intrinsic combat effects; multi-hit actions; real-time temporary effects; full-set and location-specific armour; reproducible background simulations and scenario import/export.
What comes next
Validate the public model against real attack logs. Resolve published range disagreements, Rage’s hit-count distribution, precise timer and DOT ordering, armour overlap geometry and uncommon status interactions. NPC phases, group-attacker scheduling and stealth are outside the current duel model.
Keep your setups
The current matchup is saved on this browser. Export JSON for a portable copy. Importing only reads scenario data; no Torn account access or API key is needed.
Future sharing can build on the same versioned scenario format.