Scoring and data methodology

What the numbers mean—and what they do not

DFO Gang uses public character inputs from Neople and a separate calculation pipeline maintained by this project. This page describes the model at a level that lets players compare results responsibly without pretending the score is an official damage value.

Methodology reviewed July 10, 2026.

1. From API response to comparable inputs

For a character lookup, DFO Gang requests the public profile, status, equipment, avatar, creature, skill style, and buff-enhancement data needed by the relevant model. Item descriptions and set effects are parsed into a shared stat block. This includes multiplicative damage effects, elemental values, reinforcement, cooldown effects, object damage, buff enhancement, and supported conditional equipment behavior.

When a weapon or set has mutually exclusive modes, the calculator evaluates the valid combinations and retains the strongest result for the selected scoring mode. Missing or unsupported effects are not silently treated as proven; they are a source of model uncertainty and may require a mapping update.

2. Dealer score

The dealer pipeline first produces an equipment-adjusted damage value. At a high level, its major multiplicative groups are:

damage value Ă— attack amplification Ă— parsed damage effects Ă— reinforcement Ă— elemental multiplier Ă— buff enhancement Ă— Oath effects Ă— mist effects

The strongest applicable element is used for the elemental multiplier. If supported class skill coefficients are available, the result is combined with that skill-based coefficient. Otherwise the calculator uses its legacy cooldown-and-object-damage model. The displayed number is intended for relative comparison inside DFO Gang, especially within the same class and calculation version.

3. Buffer score

Support classes are not ranked with the dealer formula. The buffer pipeline builds the applicable stat and attack contribution from combat gear, buff-swap gear, main buff, third awakening, aura, and supported passive effects. Class-specific behavior—such as Enchantress favoritism—is handled separately where the model supports it.

relative party stat contribution Ă— relative party attack contribution Ă— score scale

The resulting Buff Score is a comparison aid for modeled party contribution. Party composition, skill timing, dungeon rules, and player execution remain outside a single profile-based number.

4. Rankings

Rankings are built from stored public-character snapshots. Fame and DFO Gang Score are separate sort modes: fame reflects the game's progression measure, while score reflects this site's model. A high value in one column does not guarantee the same position in the other.

Each ranking report shows the character count, page, and most recent available update date. Characters can move or disappear as their public data changes, eligibility changes, or the calculation version is corrected.

5. Class statistics

The current statistics dataset starts with characters at or above 61,757 fame, then aggregates choices by advanced class. Reports expose their character count and update time and may include equipment-set share, Oaths, Oath Crystals, title or weapon-avatar levels, VP selections, enhancement selections, and average skill levels.

“Top 10% threshold” is the score boundary calculated from the eligible class population; it is not a claim that every table on the page uses only that top tenth. Percentages should be read with the displayed sample size, particularly for small classes.

6. Known limitations

  • The official API is a snapshot of public data and can lag behind an in-game change.
  • Training-room settings, encounter mechanics, party buffs, rotation quality, and latency are not fully represented.
  • Conditional item text can require interpretation; newly released or changed items may need parser updates.
  • Cross-class score comparisons are less reliable than comparisons within one class because skill kits differ.
  • A formula or mapping correction can change historical relative positions without a player changing gear.

Use the score to form a testable comparison, then inspect the equipment details and assumptions. It should support judgment, not replace it.