Holding a coalition of old enemies

1. Bread first

Every action you take is paid in annona, and an army on the Rhône or the Loire eats even when it does nothing; the Goths, once they have joined, eat twice as much again. The prefect Ferreolus is your only large source of grain: use him almost every fortnight. A bot that does little but march and call on Ferreolus finishes with a median of 19, and Orléans is plundered in 31 % of its games, because the auxiliaries go home when the bread runs out and the Goths come late or not at all.

2. The Goths: when, not whether

Without the Visigoths there is no relief of Orléans and no battle worth the name. They join at commitment 3, and then need a fortnight to march up from Toulouse. The emperor's embassy gives one point (Jordanes' version: Theoderic agrees at once); Avitus gives two, but only once Attila is in Belgica (Sidonius' version: the Goths wait at home until he fetches them); gifts to Toulouse give one, at twice the price. Call them too early and they eat your grain on the Loire while Attila is still at Metz; too late and Orléans waits. The bot's rhythm (embassy, Avitus as soon as he can go, the army on the Loire by early May) brings the relief in the first fortnight of the siege.

3. Orléans: Jordanes or Gregory

The walls start weak, and the outcome follows the sources. With earthworks thrown up in time (Jordanes), the city is relieved before the walls are in danger in 99 % of the bot's games. Without them, it is Gregory's story: relieved as the walls shook in 64 %. The earthworks cost two annona and an action; they are worth a point in the accounting and they protect you from the worst case, an Alan king who opens a gate. Keep Sangiban's commitment above 1, or build the earthworks, and the gate stays shut.

4. Do not fight near Metz

You can offer battle before Attila reaches the Loire. Orléans is never besieged, and the battle is fought "not far from Metz", as Hydatius has it. But Attila's host is still whole, not worn by the siege, and your coalition has had less time to gather: the allied line breaks in 31 % of these games, against 18 % after the relief, and the median falls to 21. Hydatius wrote in Galicia, and knew Gaul by report; the game's Metz is the riskier field too.

5. The field and the hill

After the relief, follow Attila and choose the ground: you fight on the Catalaunian fields, and your attempt on the hill succeeds in 75 % of games. Press him hard on the road to Troyes and you fight at Mauriacus with fewer stragglers in his host, but you take the hill only 62 % of the time. Send Thorismud up the hill (Jordanes 201), and put Sangiban in the middle, "so that a crowd of loyal men should enclose the man of whose mind they were less sure" (197). Theoderic dies in about 56 % of battles, by his own horsemen or by Andagis' spear, as Jordanes leaves it open.

6. The wagon fort: three motives

The game's central decision is the one the sources disagree on.

• Thorismud home, for the balance (Jordanes 216): the historical course; median 24.
• The trick for the spoils (the Copenhagen continuation, Gregory): the best median, 27. But the Goths and the Franks learn of it in about 55–60 % of games, and then your coalition at the end is worth little.
• Besiege him (Jordanes 213): Attila goes home weaker, but the whole Gothic army stays in the north and your standing at court suffers; median 18, the worst of the three. Rome feared the Goths more than the Huns, and the game agrees with Jordanes' Aetius.

How clearly you beat Attila in the battle decides how large a host he takes home, and so what Italy suffers in 452.

7. Prosper's reproach: hold the Alps

Prosper blamed Aetius for leaving the passes of the Julian Alps open in 452. In the late summer of 451 you can garrison them. With the garrisons, the bot loses a median of one Italian city instead of all three, Aquileia, Milan and Pavia; the median rises to 25. Garrisons and the trick together give the bot's best result, a median of 29. The price is standing at court: the actions you spend on the Alps are not spent on the emperor.

8. 454

In every source the emperor Valentinian killed Aetius with his own hand in 454. In the game your standing at court lowers the odds, and whether you press for the marriage of your son Gaudentius to the emperor's daughter decides how much. Staying away from the palace, the bot survives in about 30 % of games; pressing the marriage, in 18 %, but a marriage that survives is worth three more points. Marcellinus wrote that with Aetius "fell the Hesperian realm"; in most games it still does.

9. Reading the accounting

The accounting separates what the field decided (Orléans, the battle, Italy) from what the palace decided (your standing, the coalition, the Goths, the spoils, your life). The historical course, as far as the sources agree on it, scores in the low twenties: Orléans saved at the last moment, Attila shut in and let go, three Italian cities lost, and the patrician killed. The strongest bot play reaches a median of about thirty. Neither number is a verdict on Aetius: the game scores choices the sources describe, and three different chroniclers gave three different reasons for the most important one.