Romans: how they do against each civ
The chance of Romans winning against each rival at equal Elo, in real ranked games, with a likely range and how many games are behind it. Tap a rival to see the matchup by map and by game length.
In 1v1
| Rival | Romans win | Games |
|---|---|---|
| Saracens | 53% (51 to 56) | 626 |
| Japanese | 52% (51 to 54) | 727 |
| Koreans | 52% (50 to 55) · even | 230 |
| Maya | 52% (50 to 55) · even | 187 |
| Dravidians | 51% (49 to 53) · even | 325 |
| Magyars | 51% (48 to 54) · even | 374 |
| Bengalis | 51% (49 to 53) · even | 516 |
| Ethiopians | 51% (48 to 53) · even | 321 |
| Spanish | 51% (49 to 52) · even | 738 |
| Mongols | 51% (49 to 52) · even | 420 |
| Shu | 51% (48 to 53) · even | 141 |
| Wu | 51% (48 to 53) · even | 397 |
| Britons | 50% (47 to 53) · even | 322 |
| Byzantines | 50% (48 to 51) · even | 551 |
| Vikings | 50% (48 to 52) · even | 552 |
| Burgundians | 50% (47 to 52) · even | 319 |
| Malians | 49% (46 to 53) · even | 606 |
| Celts | 49% (47 to 52) · even | 167 |
| Franks | 49% (46 to 52) · even | 200 |
| Vietnamese | 49% (48 to 50) · even | 1,618 |
| Hindustanis | 49% (46 to 52) · even | 460 |
| Muisca | 49% (46 to 52) · even | 159 |
| Bulgarians | 49% (47 to 51) · even | 912 |
| Burmese | 49% (46 to 52) · even | 467 |
| Persians | 49% (47 to 51) · even | 188 |
| Goths | 49% (46 to 51) · even | 329 |
| Slavs | 49% (47 to 51) · even | 683 |
| Khitans | 49% (46 to 51) · even | 137 |
| Poles | 49% (46 to 51) · even | 278 |
| Mapuche | 48% (46 to 51) · even | 146 |
| Cumans | 48% (47 to 50) · even | 502 |
| Aztecs | 48% (46 to 50) | 1,417 |
| Khmer | 48% (46 to 50) · even | 273 |
| Sicilians | 48% (46 to 50) | 376 |
| Berbers | 48% (46 to 50) · even | 1,176 |
| Bohemians | 48% (46 to 50) | 811 |
| Jurchens | 48% (44 to 51) · even | 222 |
| Portuguese | 48% (46 to 50) | 753 |
| Armenians | 47% (44 to 50) · even | 373 |
| Wei | 47% (44 to 50) · even | 187 |
| Tatars | 47% (44 to 50) | 374 |
| Turks | 47% (45 to 49) | 366 |
| Malay | 47% (45 to 49) | 303 |
| Teutons | 47% (45 to 49) | 430 |
| Georgians | 47% (45 to 48) | 322 |
| Gurjaras | 47% (44 to 49) | 428 |
| Huns | 47% (44 to 49) | 452 |
| Tupi | 46% (44 to 49) | 163 |
| Italians | 46% (44 to 49) | 635 |
| Inca | 46% (43 to 49) | 450 |
| Chinese | 46% (43 to 48) | 473 |
| Lithuanians | 44% (41 to 46) | 491 |
In team games
The chance of a team with Romans instead of an average civ, at equal Elo. Team games count appearances of the civ (each player), not games.
| Map | Win | Appearances |
|---|---|---|
| All maps | 48% (47 to 49) | 27,812 |
| Arabia | 48% (47 to 49) | 2,227 |
| Arena | 48% (47 to 49) | 2,409 |
| Black Forest | 48% (47 to 49) | 1,616 |
| Nomad | 48% (47 to 49) | 2,050 |
How it's calculated
- Ranked random-map games from the current patch, without mirrors or games under 5 in-game minutes.
- First the players' Elo difference is removed, so what's left comes from the civs: chances are at equal Elo.
- A model adds up the strength of each civ and each matchup by map, Elo band and length; whatever has few games stays close to the overall number, so a cell with 4 games doesn't make things up.
- The range is where 90% of the likely values fall. It's even if it touches 50%, and for game length something is only claimed if the shortest and longest bands don't touch.
- Length is in minutes of the game's clock.
Source: aoe2-guide, games from 2026-06-01 to 2026-07-03, patch Update 185872. 328,369 1v1 games and 162,421 team games.