Guild War Top 10 Defend Teams

• 10 complete teams, 35 distinct heroes. Each hero appears at most twice, and no pair of teams shares more than two heroes. Teams with shared heroes are alternatives; Machines of War may repeat.
• Evidence across wars. Candidates needed at least 89 attempts, 10 wars and five known defending guilds, with no more than 20% of attempts from one war or 50% from one guild. This was the highest attempt threshold that allowed ten teams under the overlap limits.
• Ordered by Failure Rate. Highest unrounded rate first.
• Complete battle records. Both clears and failures count. Cleanup attacks and incomplete or ambiguous records are excluded from performance metrics.

Source: TacticusXYZ. Captures end October 6, 2026; the report is dated October 18, 2026, with community names and hero order updated October 19, 2026.

Edition

October 2026

Seasons

25–27

Battlefield tiers

4–5

Recorded wars

162

Eligible non-cleanup attacks

69,521

Exact selected-defense attempts

2,275

The Season 25–27 / BF4–5 sample contains 89,116 recorded attacks before exclusions. The attack page uses the same seasonal and Battlefield scope.

Failure Rate

Attempts leaving at least one defender alive / eligible attempts against the exact five heroes and selected Machine of War. Higher favors the defense.

Meds Up Failure

Failures / eligible attempts recorded with Medicae Up, against that exact defense. Each Battlefield row uses only its own Meds Up subset.

Battlefield share

Eligible attempts in a Battlefield / all eligible attempts against that exact defense. The headline shows the most common tier; lower tier number breaks ties.

Meds Up

At least one recorded Defender Health (Medicae) buff. Missing buff information remains unknown and does not count as Meds Up.

Clear Rate

Clears / eligible counter attempts, including failures. A clear defeats all five defenders, with or without attacker deaths.

Meds Up Clear Rate

Clears / eligible counter attempts recorded with Medicae Up. The denominator is that counter's Meds Up attempts.

Perfect share of Attempts

Clears with all five attackers surviving / all eligible counter attempts, including failures.

Perfect share of Clears

Clears with all five attackers surviving / all clears by that counter.

Machine of War usage

Recorded occurrences of a Machine / all appearances of the exact five defender heroes. Includes cleanup and incomplete attack outcomes, plus unknown Machines in the denominator.

Counts and missing data

Each rate shows its numerator / denominator. Attempts are recorded attacks, not unique deployments or players. — means no observations for that metric; 0.0% is a measured zero.

• Defense performance: Main metrics and Battlefield results require all five named defenders and their selected, most common Machine of War. All eligible opponents count, whether or not they contain a listed attack core.
• Observed Defenses: Each row is an exact five-hero lineup plus its recorded Machine of War, keeping at least four of the listed defenders. Hold Rate is attempts leaving at least one defender alive / attempts for that exact combination. Apply one swap at a time; headline rates and counters do not carry over.
• Evidence minimum: Observed Defenses shows up to 10 combinations with at least 10 attempts, sorted by highest Hold Rate. Limited sample means fewer than 20 attempts. These exploratory variants use a lower evidence threshold than the ranked defenses above. Missing support is labelled MoW not recorded.
• Counter matching: Every named attack core hero must be present; remaining heroes and both Machines of War are unrestricted. The observed defense must share at least three of the target team's five heroes. GSC, Blood Angels and Orks require all five named heroes; their four-of-five flex cohorts are excluded here.
• Counter order and support: All ten attack cores remain visible for every defense, sorted by total clear count descending, then attempts, wars and canonical core ID. Supported means at least 30 eligible attempts across 10 wars; smaller samples are Low support, and zero attempts means No data. Support labels describe the overall matchup, not each Meds Up subset.
• One attack, one observation: Each attack counts once within a matchup, even if several hero trios match. Related defenses can share observations, so their counter totals are not additive.
• Eligible outcomes: Performance requires five distinct identified heroes on each side, known death states, and the source's initial-convention marker (an em dash with no numeric debuff). A numeric debuff of at least one is Cleanup; numeric zero or ambiguous markers are excluded. The marker does not verify full starting health or the first attack against a deployment.
• Machines of War: Only options with at least 5% unrounded usage are shown, ordered by occurrence count. Missing or ambiguous Machines remain in the denominator. Recorded support does not establish that its abilities were activated, and the most common option is not selected by its failure rate.
• Battlefield composition: BF4–5 attempts sum to each team's total. Share describes where this team's observations occurred, not its prevalence among all defenses in a tier. Meds Up failure rates use the saved attempts for each Battlefield.
• Interpreting results: These are observed uploaded logs, not all Guild War play. Tier mix, opponent choice, power, caps, maps, season and repeated wars are not adjusted for. Broader counter rates do not directly estimate the exact five-hero-plus-Machine matchup, and Meds Up comparisons do not measure Medicae's causal effect.