The tree is not finished. Good.

The cost cliff is real, but the totals are specialist data

The Theria Military Academy guide reports 327,464 Tactic Scrolls and 54,616,833,000 Gold to complete the MA5 research layer. It reports 112,367 Scrolls and 7,790,053,145 Gold for MA6. Those are guide totals, not a current official price table. Confirm each node in the client before committing.

On those totals, full MA5 costs 2.91 times the Scrolls and 7.01 times the Gold of full MA6. MA5 alone also costs 106,902 more Scrolls than the reported 220,562 total for MA1 through MA4. Completion is doing a great deal of work here, mostly to your inventory.

MA6 changes the menu, not just the percentage

Evony's official version 4.50.0 announcement introduced Melee Formation and Long Range Formation in the Military Academy. Theria reports that the four troop-type formation lines open at MA1, while Melee and Long Range open at MA6. That makes the K40-to-K41 decision a branch comparison, not a race from 99% to 100%.

Later progression adds another currency wall. The official Tactical Compendium note, updated December 23, 2025, says Charge and Sweep Formation unlock at MA11 with Keep 46, and require Compendiums only after certain technology levels. Do not drain every Scroll now on the theory that later research uses a different item entirely. Some later nodes begin without Compendiums; the exact prerequisites still matter.

Choose the next node by the role it changes

Make one sentence describe the account's next job: mounted rally setter, ranged battlefield march, siege rally, reinforcing march, or in-city defence. Then compare only nodes that improve that job under the wording shown in the current client. A formation percentage in an unused role is not diversification. It is storage with a progress bar.

• Attack role: favour the troop type, attack condition and march you actually send; do not count defending-only wording.
• Defence role: favour in-city conditions, relevant troop layers and useful enemy debuffs; do not buy attack-only totals for the wall.
• Rally or reinforcement role: verify that the wording applies to leading, joining or reinforcing before counting it. Those are different jobs.

For debuff nodes, use the opponent-dependent method in the debuff-cap guide. For rally-setter growth, compare the research gain with the measured bottleneck in the Rally Capacity guide. Research power is not a combat condition. It is a receipt.

Build a packet before pressing Multiple Research

The official version 5.04.0 notes say Multiple Research can raise a technology to a chosen level, but the batch can finish only with speedups or Gems; waiting and Alliance Help are disabled. The same update added branch-completion percentages and a power-gain display. Those conveniences make a large click easier, not wiser.

Before a batch, record the exact node and target level, every prerequisite, Scroll and Gold cost, timer, applicable combat condition, and the current displayed buff or debuff. After the research, capture the new display and check an appropriate report. If the value does not appear where expected, stop the branch before repeating the mistake at greater efficiency.

Use three stopping points

• Stop at the prerequisite: when the next MA5 level serves no current role and MA6 is reachable.
• Stop at the measured gain: when the relevant displayed buff, debuff or report changes less than the next node costs justify.
• Stop before the batch: when the condition wording is unclear or Multiple Research would require Gems or speedups you did not budget.

If MA6 is months away, selective MA5 research can still be correct. If MA6 is available, leaving an MA5 branch unfinished can also be correct. The decision is not “MA5 or MA6.” It is the next useful node versus the next useful node, with prerequisites and the account's actual role attached.

A completed tree is pleasing. A useful account is occasionally untidy.

— Kat