Make every rank conditional
- Step 1: Write the exact use case at the top of the comparison: early game, boss, exploration, group combat, or a specific biome.
- Step 2: Filter the candidates to the same material tier and upgrade level, then compare them against the same target with the same food and repair assumptions.
- Step 3: Craft the top candidate and one fallback, test both on the intended route, then write down the winner, fallback, and reason the winner can be maintained and replaced in your current world.

A tier list should not hide the situation behind a permanent S or A label. The winning weapon changes with progression, target resistance, repair access, and the amount of material the world can replace.
- Write the activity and progression point beside the ranking before comparing candidates.
- Use the same food, upgrade, target, and repair assumptions for every candidate.
- Choose a primary and fallback only after both survive a short test on the route where they will be used.
Practical weapon ranking by situation (not permanent DPS)
| Situation | Ranked recommendation | Reason to choose it |
|---|---|---|
| Cover physical damage types | 1. / · 2. + · 3. + elemental arrows | The material recommends a polearm or hammer and sword for pierce, blunt, and slash coverage, with a bow and elemental arrows as extra coverage. |
| Skeleton-heavy route | 1. or · 2. · 3. Pierce only when needed | The material states that skeletons are weak to blunt and resist pierce. |
| Parry-focused melee | 1. / · 2. · 3. | The recorded parry bonuses are 4x for knives, 3x for polearms, and 2.5x for bucklers. |
| Keep movement speed | 1. / · 2. Classes with -5% reduction · 3. Classes with -20% reduction | The material records 0% movement reduction for knives and , -5% for many common classes, and -20% for two-handed axes, two-handed clubs, and most tower shields. |
If the winner changes by situation
- Remove any candidate that cannot be crafted or repaired at the current stage.
- Separate maximum damage from practical uptime.
- Retest with the actual target rather than a training assumption.