One number: luck +0.45
Here is the appraisal problem Lucky sets you: every other trait in the game changes how a hero decides, and this one changes how the dice land. The combat-AI table shows a single coefficient, luck +0.45, and nothing else there. No discipline, no courage, no kill hunger, no threat. The combat AI scores its options exactly as it would for a blank-slate hero and picks exactly the same actions; the trait then leans random outcomes in the hero’s favour after the choice is made.
That +0.45 is enormous by trait standards. Only Glory Hunter’s +0.5 kill hunger and Lone Wolf’s −0.5 protectiveness carry more raw weight anywhere in the table, and both of those buy a behaviour you can watch and plan around. Lucky’s number buys no behaviour at all, which is why the game’s second-biggest positive coefficient is also its hardest to value. The only other trait touching the same parameter is Sure-Footed, whose +0.12 luck comes bundled with a recklessness damper: a discount version with a personality rider attached. Lucky is the pure article at nearly four times the dose.
The places you can see it
Since the 2.12 patch Lucky has one effect you can watch. When a Lucky hero rolls a natural 1 on a hazard or puzzle check, the game throws the die again, once per delve, and the delve report says whether the luck held. A second natural 1 still fails. On the twenty trap and puzzle rooms, where a natural 1 can put a hero on the floor, that reroll is the most concrete thing the trait does. Jinxed, one of the 2.12 traits, is its mirror: it rethrows its own natural 20.
The reroll is not the only visible effect in the 2.1.8 code:
- +2 on every trap and puzzle roll the Lucky hero makes, the joint largest bonus on those rolls (Trap-Wise adds +2 on traps, Methodical +2 on puzzles, Quick-Witted +1 on both).
- As leader at a fork, if the party goes into a sealed door or a treasure side room without scouting it first, a Lucky leader makes the sealed door’s trap 3 easier and raises the gold found there to 125%. A Clumsy leader does the reverse. Careful leaders scout first, and scouting cancels both effects.
- The Stranger’s Dice. In this treasure-room event a Greedy, Glory Hunter, Reckless or Hot-Headed hero gambles with a stranger. Normally they win 45% of the time; with any Lucky hero in the party they always win.
Why you cannot see the rest of Lucky working
Outside those rolls, you will never catch the combat bias in the act. Read a delve report where a Lucky hero shrugs off a blow, lands the improbable strike, slips the curse that caught everyone else. Any one of those outcomes could have happened to anyone. A bias is not a guarantee; it is a lean on the distribution, and single events tell you nothing about a distribution. The evidence for Lucky is actuarial: across a season of delves, this hero’s ledger of close calls tilts visibly in one direction, and only in aggregate.
Two structural facts make the lean worth far more than its invisibility suggests. First, permadeath. In a game where heroes die for good, the rolls that matter most are the tail outcomes: the one save that decides whether a career continues. A standing bias in the hero’s favour is cheapest exactly where the downside is a funeral. Second, exposure. The trait pays out per random outcome, so its value scales with how many dice the hero is actually exposed to: long delves, attrition-heavy conditions that stretch fights across extra rounds, a schedule that sends the hero out constantly. A luck bias on a bench-warmer is a coefficient with nothing to multiply.
Who carries Lucky best
Role barely enters into it, and that is unusual enough to say twice: Lucky is role-agnostic, because it modifies outcomes rather than priorities. A front-liner, a finisher and a healer all roll dice, and all of them roll better with it. The sharper question is whose dice you most want loaded, and there are two defensible answers. Load the hero who eats the most randomness, typically the front line, standing in the largest volume of enemy contact per round. Or load the hero whose death would cost you most: the bloodline founder, the star your promotion push is built on. Under permadeath, Lucky on your best hero is the closest thing the game sells to insurance.
The interaction with risk-appetite traits deserves one careful sentence. Recklessness decides which gambles a hero volunteers for; luck decides how gambles resolve. The two operate on different layers, so Lucky does not fix a Reckless hero. It just wins slightly more of the coin flips that hero was always going to take. Softening the cost of a habit is not the same as curing it. Note also that you cannot build around Lucky: named party synergies fire off who stands with whom, and a trait with no behaviour gives a composition nothing to hook into. It compounds any party quietly and anchors none.
Pricing a pure bias
Appraising Lucky splits the market into two wrong camps: managers who pay a legend premium for the name, and managers who dismiss what they cannot see in a report. The rational price sits between them and scales with exposure. Pay real money for Lucky on a hero who will delve constantly for years, and pay nothing for it on a rotational squad player. It ages perfectly, since a bias neither decays with attributes nor dulls with the hero’s reflexes; a Lucky veteran is exactly as lucky as a Lucky rookie, which is more than you can say for any attribute you ever paid for.
At oath time, the trait is pleasantly cheap to keep. Kill hunger builds glory and glory inflates renegotiations, but luck generates no highlight reel of its own; the hero’s wins look ordinary on paper. When you are deciding whether the premium recoups, run the numbers on what a season of delving actually returns through the dungeon-yield tool; a bias only compounds on top of throughput that exists. Pass on Lucky only when the hero will not see the dice often enough to collect.
Around the hall
- Jobs. Bounty and Gear Hunt contracts both list Lucky among their liked traits, adding 5% to the hero’s suitability.
- Expedition finds. Lucky adds +0.3 to the hero’s “finder” score on expeditions (Curious +1.1, Pathfinder +1.0). See Pathfinder for how finds work.
- “Dice by the Fire”. When the guild’s dice night gets out of hand, a Lucky or Gambler hero can run the table: +4 morale for them, +2 for the hero who was losing, and +1 for the whole guild, at no cost in gold.
- Flavour. On pair duty the hero speaks with the “luck” voice shared with Gambler, Streaky and Superstitious.
Lucky can never appear with Jinxed.
Change log since v1.9.6
Luck +0.45 has not moved in any build this wiki tracks, v1.9.6 through v2.1.8. The natural-1 reroll arrived in 2.12 on top of it, and the skill-roll, fork, event and job hooks above are read from the 2.1.8 code. The most mysterious coefficient in the game is also one of its most settled.