ParseCard

How Awards Work

A deep dive into how ParseCard analyzes your fights and picks the six awards that appear on every card.

The Award Engine

Every ParseCard is built around six awards. When you generate a card, the award engine fetches the fight data from Warcraftlogs and runs it through six specialized detectors. Each detector analyzes a different aspect of the fight and produces a ranked list of candidate awards. The engine then selects the best six from across all detectors, following a set of rules that ensure variety and fairness.

The Six Detectors

Each detector focuses on a specific category of performance data from the combat log:

Damage Detector

Analyzes DPS output, damage breakdowns, and relative performance within the group. Identifies top damage dealers, players who significantly outperformed their role, and those who fell behind. Evaluates both raw numbers and context like fight duration and group composition.

Healing Detector

Evaluates healing output, overhealing, and healing efficiency. Recognizes clutch healers who kept the group alive during dangerous moments, efficient healers who minimized waste, and highlights cases where healing was notably low relative to the damage the group was taking.

Deaths Detector

Tracks who died, when they died, and what killed them. Distinguishes between unavoidable deaths from raid-wide damage and avoidable deaths from standing in mechanics. Awards range from praise for surviving the entire fight to roasts for being the first player down within seconds of the pull.

Mechanics Detector

Checks for avoidable damage taken, interrupt counts, dispels, and specific boss mechanic handling. Players who dodge everything earn praise. Players who eat every fire patch, miss every interrupt, or trigger avoidable raid damage earn a roast. This detector understands per-encounter mechanics for current raid content.

General Detector

Covers cross-cutting performance metrics that do not fit neatly into the other categories. Things like overall activity percentage, time spent alive, buff uptime, and comparative performance within role. This detector catches the players who were consistently strong or consistently absent across all dimensions.

Special Detector

Handles unique or rare situations that the other detectors miss. This includes unusual events like solo-surviving a wipe for an extended period, dealing damage during phases where most players cannot, or achieving notable firsts and lasts in the fight timeline.

Praise and Roasts

Awards are categorized as either praise or roast. Praise awards recognize genuinely good play. Roast awards call out mistakes or poor performance in a playful way. The balance is deliberate:

  • Approximately 60% praise and 40% roasts on every card.
  • A maximum of one roast per player per card. No one gets piled on.
  • Even in a bad fight, there are always at least one or two praise awards recognizing whoever did the best with a bad situation.

Roasts only ever target in-game performance. They reference deaths, damage taken, low output, and mechanical failures. They never reference the player personally, their skill level overall, or anything outside the specific fight being analyzed.

Tone Settings

When generating a card, you can choose from three tone levels that control how award text is written:

  • Friendly — Praise is warm and encouraging. Roasts are gentle, framed as light observations rather than callouts.
  • Bold — Praise is confident and direct. Roasts have more edge and are clearly calling out the mistake.
  • Brutal — Praise is matter-of-fact. Roasts are blunt and pull no punches. Best for guilds that enjoy harsh banter.

An important detail: the tone only changes the wording of each award. It does not change which awards are selected. Running the same fight at Friendly and Brutal will produce the same six awards with the same recipients, just with different text. This means the card is always an accurate reflection of the fight data, regardless of tone.

How Awards Are Selected

With six detectors each producing multiple candidates, the engine needs to narrow down to exactly six awards. The selection process follows these priorities:

  1. Priority scoring — Each candidate award has a priority based on how notable the underlying performance was. Bigger outliers score higher.
  2. Category balance — The engine enforces the 60/40 praise-to-roast ratio, ensuring every card has a mix of both.
  3. Player diversity — The engine avoids stacking multiple awards on the same player when possible, and never gives more than one roast to a single player.
  4. Detector diversity — Awards from different detectors are preferred over clustering from a single detector, giving the card a well-rounded view of the fight.

Deterministic Results

Award selection is deterministic. The engine uses a seed derived from the report code and fight ID, which means generating a card for the same fight will always produce the exact same awards. You cannot reroll awards by regenerating. This is intentional: the card is a faithful snapshot of what happened in the fight, not a random selection.

Over 40 Award Types

The award catalog contains over 40 distinct award types, each with unique icons, titles, and descriptions. The catalog is managed through a database, which means new awards can be added and existing awards can be refined without requiring code changes. Awards cover everything from top-DPS recognition to comedic callouts for dying to the same mechanic twice in one fight.

Want to see how awards contribute to long-term rankings? Check out the Leaderboards guide.