Most players build one "best" march and take it to every fight. The game quietly punishes them for it, because Rise of Kingdoms is not one combat game — it is three, and the same pair of commanders can be elite in one and a liability in another. That isn't a slogan. We measured it, across hundreds of thousands of battles, and this guide shows you the numbers.
Here is the shape of the three games, and then the part almost nobody can show you: why the third game refuses to have a tier list at all — and what to do instead.
What are the best pairs for rally, garrison and open field?
- Open field: Qin Shi Huang + Yi Seong-Gye, at the top of the record's ladder.
- Garrison: Gorgo + Heraclius, the best-evidenced wall.
- Rally: it depends on the wall. No rally pair is best until you name the garrison you are hitting.
What are the best open field pairs?
In the open field, both sides chose to fight, opponents mix freely, and over thousands of engagements the matchup noise averages out. That's why an open-field ranking is a legitimate object: when we test our pairing model against a year of empirical battle data, the agreement is strong and stable (rank correlation +0.73 on high-confidence pairs (measured after the 2026-09 kit merges), corroborated by two independent data sources that agree with each other at +0.97).
The empirical top of the open-field ladder, from the record rather than opinion: Qin Shi Huang + Yi Seong-Gye sits above Qin Shi Huang + Zhuge Liang (the data is explicit about that order, whatever your alliance chat says), with Sun Tzu Prime + Liu Che and Sun Tzu Prime + Bai Qi in the S-anchor band. All archer- and infantry-led skill engines — open field is where skill-cycling pairs earn their reputation.
Your open-field decision: invest your field marches by the ladder (which commanders are worth building at all is its own guide: Which Commanders Should You Actually Invest In? Ranked by Role, Season, and What You Spend), and match troop type to your museum and kingdom buffs (the civilization half of that is Best Civilization in Rise of Kingdoms — And Why The Answer Changes Twice) — the ladder assumes you aren't fighting at a troop-type disadvantage. The formation you march in is a separate choice: The Complete Formation Guide: Which One, Why, and the Inscriptions That Make It Worth Having.
What is the best garrison pair?
Garrison looks like it should be the rankable one. A defender doesn't choose its attackers, so every wall faces roughly the same incoming mix, and a wall's quality ought to be a stable property of the wall.
We tested that and it isn't true — or at least, we can't yet show it is. Our garrison scoring model has now been measured against the record three separate times, on three independently selected samples. On the first two it did not predict: its ranking ran slightly against the record on our deepest read of the record, a twenty-pair set (rank correlation −0.28), and showed no relationship on a smaller sample selected by a different rule in a different season (−0.07). After the September 2026 kit merges the garrison cell now agrees with the record, +0.74 on 51 pairs — but on a different sample, so the table below still rests on the battle record, not the model. We also tried three different mechanical explanations for why one ordering of a pair outperforms the reverse, and all three were refuted by the game's own rules before they reached the model.
So we are not going to rank walls for you by a predicted score. Here is what we can stand behind instead: what the record actually shows people garrisoning with, and how deep that evidence runs. (How the captain and the reinforcements behind a wall work is its own guide: How Garrisons & Reinforcements Actually Work in Rise of Kingdoms.)
| Garrison pair | Governors | Garrison battles | Garrison share | Trade |
|---|---|---|---|---|
| Gorgo + Heraclius | 1,280 | 20,932 | 81% | 72% |
| Gorgo + Hector | 676 | 20,174 | 93% | 89% |
| David IV + Hector | 638 | 37,056 | 68% | 96% |
| Choe Yeong + Heraclius | 327 | 9,397 | 89% | 77% |
| Gorgo + Charles Martel | 239 | 8,982 | 89% | 90% |
| Heraclius + Yi Sun-Sin | 174 | 5,070 | 80% | 75% |
How to read this table. Governors is how many distinct players ran that pair — it is the depth of the evidence, and it is the column to trust. Garrison share is how much of that pair's combat record is defending rather than attacking; a high share means it is a genuine wall and not a field pair that occasionally sat in a city. Trade is kill points taken divided by kill points lost.
And one warning that matters more than the table. Trade looks like a quality score and it is not safe to use as one here. In our full twenty-pair record the highest trade figures sit on the shallowest rows — pairs with two, four, or twelve governors post 140%, 146%, 187%, while the 1,280-governor standard posts 72%. That is what thin samples and soft matchups look like, not what elite walls look like. Rank by depth, read trade as context, and do not chase the big percentage on a pair nobody runs.
Walls trade under 100% by design. Their job is buying time, not profit.
Your garrison decision: build Gorgo + Heraclius. It is the volume-proven standard — 1,280 governors and 20,932 measured garrison battles, more evidence than any other wall in the record by a factor of two. If you already run Gorgo and want a second wall, Gorgo + Hector is the next best-evidenced at 676 governors and a 93% garrison share, the purest defensive record in the set.
A correction we owe you — and its own correction. An earlier version of this guide recommended Attila + David IV as "the record's answer for a whale account." That was our model talking, not the record, and the recommendation was rightly withdrawn. We then went further and called it a field pair, off a smaller sample that read only 20% of its battles as garrison. The deep record has settled that second claim the other way, and it really is a garrison pair: 64.5% of its 94,645 recorded battles were garrison defences (835 governors, our deepest read of the record, 2026-08-23). (A nine-pair sample of 979 rows read 20% — a different, smaller sample.) The lesson stands and sharpens: the original sin was recommending off a model score, and the fix nearly overcorrected off a smaller sample. The model-scored table stays withdrawn; on the deepest read we hold, the pair belongs in the garrison conversation.
What are the best rally pairs?
There is no single answer until you name the wall. Here's the finding that separates this guide from every tier-list site, and we can prove each sentence:
How much of rally damage is skill damage?
Less than you think. Across every measured rally pair, skill damage is only ~10% of total output (it's ~23% in open field). The other 90% is normal-channel damage — basic attacks, counters, combos, and smite, which the game classes as normal damage. That's why the community's #1 pair, Scipio Aemilianus + Ivar, is built entirely of smite amplification: it lives in the channel that actually carries rallies.
Do the hardest-hitting pairs trade best?
No — the hardest hitters trade worst. Measured across high-confidence rally pairs, the more damage a pair deals per second, the worse it tends to trade: damage-per-second correlates with trade efficiency at −0.48. Subutai+Achilles produces a monstrous 44,800 damage/s and trades at 129%. Shapur+Ashurbanipal produces 7,300/s — and trades at 165%, which is why 1,420 governors have run it across 219,069 recorded battles. If you've been picking rally pairs by damage screenshots, you've been optimizing the wrong number.
Does any trait predict rally trade?
No single trait of the pair does. We tested three, honestly: raw damage (it tracks trade the wrong way, −0.65), survivability (its sign flips between datasets — glass-cannon Attila+Takeda tops one sample at 320% trade; zero-debuff tank pair Pakal+Charles Martel tops another at 214%), and debuff loading (it looked predictive on a small sample, +0.60, then showed no relationship the moment coverage got honest, −0.09). A quantity whose best predictors flip sign isn't a property of the attacker at all.
Why does the wall matter more than the pair?
Because rally trade is a property of the matchup — and that we validated. For a fixed attacker, rank the opponents it faced by our defender-strength score and the measured trade falls right down the line: the stronger the defender, the worse the trade, a median rank correlation of −0.53 across six of six independent test cells, with zero tuned constants. Meanwhile the walls themselves impose a uniform tax: both a skill rally and a normal rally gained almost identical trade (+93 and +84 points) when hitting off-meta garrisons instead of the meta walls. The wall in front of you moves your result more than the pair behind you.
How should you lead a rally?
The expert rally doctrine — the one the data permits — has three lines:
- Never blind-rally. Scout the garrison pair first. Against an Attila-anchored defence, silence plans are dead (silence-immune) and swarming feeds its AoE counters — bring your best single heavy rally or pick another target. Against Gorgo-anchored walls, note her own kit is active-skill reliant: a Silence-carrying lead (Scipio Aemilianus) attacks the wall's engine, not just its health bar. Both of those are kit facts, readable off the commanders' own skill text — they do not depend on any ranking.
- The strongest account leads. Only the rally leader's technology, VIP, and equipment apply to the entire rallied army (expert-verified). A perfect pair under a mid account is a mid rally; settle who leads before you argue about who pairs.
- Rank conditionally or not at all. Shapur+Ashurbanipal is the trade king into hard walls. Scipio+Ivar is the smite-channel answer where anti-smite is absent. Attila+Subutai is the accessible volume pick that does both jobs adequately. "Which is best" has no answer until you name the wall — and anyone who gives you one anyway is reading a tier list, not the record.
What should you change before your next fight?
Open field: invest by the ladder above. Garrison: build the best-evidenced wall, not the best-scored one — and scout before you rally into it. Rally: stop asking "what's the best rally pair" and start asking "what's the best rally pair against that garrison" — and put your biggest account in the lead slot tonight, because that one change costs nothing and moves every rally you join.
What this guide can't know
A static page can't see your roster or the wall in front of you, and the right pair depends on both — that is the job of our matchup scoring: your roster, scored against the actual wall in front of you, with expected trade.
What the advisor does instead: you tell it about your account; screenshot reading is live on paid plans.
Free: Your first five reads are full ones — the whole answer, chart and all. After that, your questions get the short read: the key fact, straight, and the right next question to ask. Companion $19/month or $190/year · Emperor $69/month or $690/year · two months free on annual