Build for the hit you cannot answer.

An offline defense readiness check for Evony farming keeps: audit the troops actually home, reject active-only assumptions and budget recovery before training more T1.

A farmer’s defense must work without its owner. If the plan requires moving cavalry, changing gear and healing between hits, the plan has employed you. Unfortunately, you are asleep.

Kat’s verdictKeep level and troop counts cannot establish a safe defense. Before another large training bill, audit the unattended setup and the recovery you can afford. This is damage limitation, not permission to let a bubble expire.

Write down the keep that will actually be there

Record troops by type and tier, including monster cavalry and gathering marches when they return. Record the appointed wall main and assistant, active skills, exact gear refines, in-city buffs, participating subordinate cities and their debuffs. Add Hospital free space, recovery research, Source of Life and the resources and speedups available for repair. Keep account screenshots private.

Delete imaginary protection

Exclude manual ghosting, equipment swaps, fresh temporary buffs and healing between hits unless they really remain available while unattended. Official rules exclude assistant equipment attributes; expensive assistant gear is not extra wall protection. Confirm skill compatibility and conditions in the current game interface.

A T1 wall is a system, not a troop order

Specialist guides describe large T1 mounted defenses supported by flat refines. They also warn about siege and exposed high-tier troops. Historical player discussions often assume active ghosting. None establishes a universal safe T1 count for an offline farmer. A favorable ranged report does not prove survival against siege, a stronger rally or repeated hits. More troops can enlarge the repair bill before they solve the weakness.

Budget the limiting resource, not the total

Training and recovery consume different mixes of food, lumber, stone and ore. Add them only for a quick headline. The real recovery limit is the resource that covers the fewest comparable report-sized repairs.

Recovery coverage = the lowest of food available ÷ food needed, lumber available ÷ lumber needed, stone available ÷ stone needed, and ore available ÷ ore needed

Worked example: stone says no

One comparable hit required 220M food, 90M lumber, 140M stone and 60M ore.

The reserve contains 500M food, 260M lumber, 240M stone and 180M ore.
Coverage ratios: food 2.27, lumber 2.89, stone 1.71 and ore 3.00.
Stone limits the reserve to 1.71 full hit-sized repairs. A 2-hit plan needs every ratio to reach at least 2.00. The combined total looks generous at 1,180M available against 510M per hit, but 2 stone bills need 280M and only 240M is available: a 40M shortfall.

Make the training gate explicit

If coverage is below the planned hit count, improve the limiting reserve, recovery capacity or exposed troop plan before mass training. If the wall pair conflicts or the refines do not support the chosen low-tier type, fix the setup first. If the evidence required ghosting, equipment changes or healing between hits, label it active defense, not offline defense.

The total looked comfortable. Stone read the footnotes.

Protect recovery before the training queue

Once recovery coverage passes the chosen hit count, do not treat the whole surplus as training money. Ring-fence the recovery reserve first. Copy the resource price for one unchanged reference batch from the current training screen; another account’s cost is not evidence.

Spendable resource = maximum of 0 and resource available minus planned hits × recovery needed per hit. Safe batch count = the lowest spendable resource ÷ current displayed cost for one reference batch, ignoring resources with a zero batch cost.

Worked example: the queue can wait

Using the same reserve, protect one report-sized recovery before training anything.

Spendable amounts are 280M food, 170M lumber, 100M stone and 120M ore.
Suppose the current screen prices one unchanged reference batch at 20M food, 10M lumber, 8M stone and 4M ore. These are illustrative inputs, not universal Evony costs.
The batch multipliers are 14, 17, 12.50 and 30. Stone again sets the limit.
The safe cap is 12 whole reference batches. Keep speedups in a separate budget. Fractions do not train troops; they only decorate the spreadsheet.

If the plan protects two comparable recoveries, stone needs 280M but only 240M is held. Spendable stone is 0, so the safe training cap is 0 until the reserve or plan changes.

The training queue was ready. The recovery cupboard had filed an objection.

Use reports without buying an expensive experiment

Start with existing anonymized reports. Compare troop losses by tier, attacker type and strength, buffs, reinforcements and whether healing occurred. Battlefield reports can help compare combat setups, but official rules give main-city troops unlimited Hospital capacity and free resource healing there. That recovery result does not transfer to ordinary map defense. Source of Life revival is an SvS rule and excludes subordinate-city troops; check the live event terms and cost.

The spending decision

Fix missing wall assignments and incompatible skills first. Then fund relevant refines and recovery capacity within a fixed budget. Expand the troop build only when comparable evidence supports the intended job. If the only good report required your intervention, label that setup active defense. The wall has not learned to use your phone.

Sources: Official assistant and battlefield rules; Official SvS revival restriction; NERFPLZ: historical T1 wall analysis; Evony Guru: defense configuration; Player discussion: T1 defense assumptions.

Factual review: 2026-10-06. Official rules support the assistant and recovery distinctions. Specialist advice and historical player discussion inform the T1 cautions, not a verified survival threshold. The audit, recovery reserve and batch cap are editorial decision tools; no controlled offline test, universal training price or win probability is claimed.