How the Analysis Works
These recommendations are built from a large dataset of Guild Raid replay videos collected from creators across multiple raid seasons.
Each replay is reviewed and cataloged with:
- Boss
- Difficulty Tier
- Damage Dealt
- Five-Hero Team Composition
- Machine of War
- Upload Date
The objective is not simply to identify the single highest damage replay, but to determine which complete teams consistently perform best in today's Guild Raid environment.
Step 1: Measuring Replay Quality
Rankings use validated replays from L4, L5, M1, M2 and M3, the stages that loop during a season. L1–L3 are cleared once per season and do not contribute to rankings or summary statistics.
Each replay's damage is divided by the median damage recorded for the same boss and stage. This compares performance within a stage before combining results across stages.
Step 2: Weighting the Current Meta
More recent replays receive greater influence, with their weight halving every 90 days relative to the latest qualifying replay date. For each exact team, replay scores are averaged within each stage, then combined using stage weights of L4 1.25, L5 1.33, M1 1.42, M2 1.50 and M3 1.75.
A coverage factor also accounts for how many of the boss's observed looping stages the team has demonstrated. It is the square root of the team's observed stage-weight share. Maps remain combined.
Step 3: Identifying the Best Teams
Exact teams are scored consistently, then each archetype is represented by its highest-scoring qualifying team.
An exact team consists of:
- The five heroes
- The selected Machine of War
Machine of War is treated as part of the team itself because changing the Machine of War frequently changes both the strategy and overall performance.
Step 4: Community Confidence
Replay support has diminishing returns and stops increasing at ten replays.
The support multiplier is min(replay count / 10, 1)^0.42. The final score is 100 × stage-normalized, recency-weighted performance × coverage × support. Replays beyond ten still contribute to performance and damage statistics; they do not increase the support multiplier.
An exact team of five distinct heroes and one Machine of War needs at least 2 qualifying L4–M3 replays to be considered for the final selection.
Step 5: Selecting the Final Recommendations
For each boss, the highest-ranked qualifying archetype becomes the META recommendation, while the strongest qualifying alternative archetype becomes the ALTERNATE recommendation.
Within each recommendation, a single exact team is selected consisting of five heroes and one Machine of War.
Magnus is the sole exception to the different-archetype rule: Dante is the Meta Custodes lineup, and Atlacoya is the selected Alternate Custodes lineup. The Atlacoya selection is an editorial choice; it does not change the scoring formula used for the other bosses.
Replay Examples
Each team has up to ten exact-lineup replay examples. L4–M3 examples appear first, ordered by damage from highest to lowest. If fewer than ten are available, the remaining places are filled with L1–L3 examples, also ordered by damage.
Every replay displayed matches the recommended:
- Five heroes
- Machine of War
L1–L3 examples are included only to help study positioning, deployment order, movement, and execution. They never affect eligibility, rankings, replay counts, average damage or maximum damage. Displayed damage statistics are raw damage across all qualifying L4–M3 replays for the exact team, rather than weighted scores or just the ten examples shown.
Team recommendations are reviewed for each guide update. New matching replays can refresh the evidence for an existing lineup without automatically changing the recommended teams.