AnswerDeep diveClash of Clans
What is the best TH18 army in Clash of Clans?
Short answer
A Thrower core with three Healers, three Ruin Witches and a Totem and Revive spell kit. It 3-stars 89% of simulated max bases, about 75% on the real ladder, roughly 5 to 10 points above the best real builds. Since the October balance changes, the best air army scores higher (98% against 91%).
The army
| Slot | Take | Why |
|---|---|---|
| Troops (352) | 9 Thrower, 3 Healer, 3 Ruin Witch, 3 Super Wall Breaker, 1 Apprentice Warden, 1 Ice Golem, 1 Druid, 1 Headhunter, 1 Balloon, 2 Goblin | Six-tile range, cheap healing, free bodies from rubble. The Apprentice Warden’s aura covers troops the Warden’s ring misses. |
| Spells (11 + 4) | 5 Totem, 4 Revive, 1 Earthquake, 1 Poison | Brew 4 Revive, 2 Totem and the Poison; request 3 Totem and 1 Earthquake. Poison the Guardian or defending heroes once they engage (+2 to 3 points against Logger and Longshot bases). |
| Siege Machine | Troop Launcher | 89%, against 84% for Siege Barracks or Sky Wagon and 80% for Log Launcher or Wall Wrecker. |
| Clan Castle (55) | 1 Root Rider, 1 Ice Golem, 2 Balloon, 1 Rocket Balloon, 1 Wall Breaker | 3 Ice Golem and 2 Balloon is a tie and easier to request. |
| Barbarian King | Spiky Ball + Snake Bracelet, Frosty | The best King pair with ground armies in the equipment test. |
| Archer Queen | Magic Mirror + Action Figure, Electro Owl or Unicorn | The two pets tie. |
| Grand Warden | Life Gem + Heroic Torch, Sneezy | Life Gem is the most valuable item in the game here. |
| Dragon Duke | Fire Heart + Revenge Deck, Phoenix | Revenge Deck scores 1 to 3 points above Electro Fangs in every test. |
Air is ahead since October. On 600 fresh bases with the October changes in, the best TH18 air army 3-stars 98% against 91% for this one, and holds up far better on fully maxed bases (80% against 45%). If your air troops are leveled, start there.
How to play it. Pick one side and drop everything along it in a single line: Ice Golem, then the Troop Launcher, Throwers and heroes, then Wall Breakers, with Ruin Witches and Healers behind. Put a Totem wherever the most fire is landing and a Revive on each hero as it falls. The average attack ends at 103 seconds, so the timer is rarely the problem (1% of attacks).
The October changes. Supercell announced them on 17 Oct and they went live in the 20 Oct maintenance break: Fire Heart lost its damage bonus, about 8% of its bonus HP and some healing, and Ruin Knights dropped to 3,800 HP, 330 damage per second, slower movement and 18 housing. That is almost exactly the mild guess tested here before the numbers were out (3,800 HP, 328 damage per second), on 600 fresh bases:
- The army as it stood drops from 91% to 89.5% (mild) or 87.5% (harsh).
- Revenge Deck in place of Electro Fangs wins most of that back (92.5% mild, 90% harsh). Keep Fire Heart: the best Duke pair without it scores 1 to 2 points lower.
- Three Ruin Witches is still the right count. Two or fewer score 2 to 5 points lower.
- The buffed Hog Riders, Miners, P.E.K.K.A and Meteor Golem still cost 2 to 6 points when swapped in for three Throwers, even with a generous 15% buff. The live buffs are smaller, 5 to 11%.
Supercell’s own war data backs the Witch core: TH18 attacks with Ruin Witches 3-star about 76% of the time, the highest of any troop that sees meaningful use.
Why it wins
Ranking troops by hitpoints and damage per housing space gives the wrong answer at TH18. Testing 48 troops each as a whole army, that paper ranking doesn’t predict the result: the correlation is slightly negative. The paper top five (Hog Rider, Balloon, Barbarian, Super Barbarian, P.E.K.K.A) score 10 to 40%; the Thrower is 31st on paper and third in the test.
| Lever | Measured effect | What it does to the list |
|---|---|---|
| Heroes | They deal 59% of the army’s damage; losing one costs 29 to 51 points. | Revives, Totems, Healers and tanks all exist to keep four heroes alive. |
| Range | Melee to 6.5 tiles: 20% to 72% at equal stats. | Throwers as the core troop. |
| Unit size | 1-housing to 40-housing units: about double the 3-star rate at equal stats. | No swarms; the two Goblins are cleanup only. |
| Healing | The first Healer returns almost 6 Throwers’ worth of hitpoints; most of the healing comes from the first four. | Three Healers, not eight. |
| Bodies from rubble | Three Ruin Witches add 11 points over none. | Each raises up to ten 4,000-hitpoint Ruin Knights from destroyed buildings. |
| Walls | The Troop Launcher beats the Log Launcher and Wall Wrecker by 9 points. | Troop Launcher, three Super Wall Breakers, one Earthquake. |
| Uptime | Defenses land 18% of their paper damage. | Totems, to waste the 15 seconds each defense gets. |
Swapping three Throwers for the same housing of any of the other 48 troops made the army worse or left it unchanged.
Range beats size, and size beats swarms
Moving the same unit from melee to 6.5 tiles takes it from 20% to 72%. A ranged unit hits buildings over walls, is in reach of fewer defenses at once and wastes less time walking. That is why the Thrower leads the ladder with mediocre stats per housing.
| Attack range (tiles) | Ground (3-star rate (%)) | Air (3-star rate (%)) |
|---|---|---|
| 0.5 | 20.0 | 15.0 |
| 2 | 31.7 | 22.5 |
| 3.5 | 45.8 | 37.5 |
| 5 | 52.5 | 51.7 |
| 6.5 | 71.7 | 79.2 |
| 8 | 86.7 | 89.2 |
| 10 | 96.7 | 96.7 |
Splash punishes small units hard. With identical hitpoints and damage per housing space, an army of 40-housing units 3-stars about twice as often as an army of 1-housing units. Spring Traps add to it: each one deletes any ground troop of 18 housing or less.
| Housing space per unit | Ground, melee (3-star rate (%)) | Ground, 5-tile range (3-star rate (%)) | Air, 3-tile range (3-star rate (%)) |
|---|---|---|---|
| 1 | 11.7 | 30.8 | 7.5 |
| 2 | 12.5 | 33.3 | 12.5 |
| 5 | 16.7 | 41.7 | 16.7 |
| 10 | 20.0 | 52.5 | 29.2 |
| 20 | 21.7 | 60.0 | 34.2 |
| 40 | 23.3 | 62.5 | 41.7 |
Three Healers, not eight
Behind Throwers, the first Healer restores about 15,500 hitpoints per battle, almost six Throwers’ worth of health for less housing than one Thrower. Healing on one target stops stacking at five Healers’ worth (about 410 HP per second), and Air Defenses pick Healers off: of 8 Healers, 5.7 are alive at 60 seconds and 3.4 at 90.
| Healers in the army | Average of 7 troops (3-star rate (%)) | Throwers (3-star rate (%)) |
|---|---|---|
| 0 | 40.7 | 41.7 |
| 1 | 47.3 | 59.2 |
| 2 | 47.0 | 63.3 |
| 3 | 46.9 | 58.3 |
| 4 | 48.4 | 60.8 |
| 5 | 49.3 | 61.7 |
| 6 | 51.5 | 70.0 |
| 7 | 48.2 | 55.8 |
| 8 | 47.1 | 60.8 |
| 10 | 47.3 | 50.8 |
| 12 | 46.1 | 60.8 |
The real ladder agrees: builds with 8 to 10 Healers score lower (31% for 15 Throwers + 8 Healers).
Why defenses do less than their card says
A max TH18 base delivers about 18% of its paper damage: 9,200 DPS against ground would be 1.65 million over three minutes, but against the best army it lands 305,000. The average defense is destroyed after 50 seconds and has a target in range for only 15 of them. The Monolith and the multi-target Inferno look modest on paper and do two to three times the damage of the Air Defenses.
Spells, heroes and equipment
Spells lift the best army from 44% to 88%, and Revive does most of it: about 4 points per housing space, against 1.5 for Totem and Freeze. Heroes deal 59% of the damage, so a spell that returns a dead hero at 80% health buys more than any buff. Removing one of the four Revives costs 8 points; removing the sixth Totem costs under 2, so Totems are the place to trade for a Poison or Freeze.
The Warden is the hero you cannot drop: every line-up without him is at 50% or below. Removing one hero from the best four costs 51 points for the Warden, 36 for the Queen, 34 for the Duke and 29 for the King.
Every equipment pair of every hero was tested again on 23 October with the in-game Life Gem rule, on 15 armies: see best hero equipment at TH18. The kit in the army table matches the leaders there for ground armies: Spiky Ball + Snake Bracelet on the King, Magic Mirror + Action Figure on the Queen, Life Gem + Heroic Torch on the Warden and Fire Heart + Revenge Deck on the Duke.
What equipment is worth. Raw hitpoints and damage on the item card don’t predict value; what pays is how much more of the base the whole army takes. Life Gem adds about 19 three-star points over an empty slot, the most of any item, almost all of it through the army: 6.8% more of the base for everyone else and 0.8% for the Warden.
Pets. Frosty on the King, Electro Owl or Unicorn on the Queen (a tie), Sneezy on the Warden, Phoenix on the Duke: the best one-to-one assignment out of all 11,880. The wrong pet costs roughly 5 to 10 points.
If a hero is upgrading. Bring the Minion Prince on Metal Pants + Meteor Staff with Greedy Raven: on 600 fresh bases the army 3-stars 87.2% with the King out, 87.9% with the Queen out and 82.5% with the Duke out, against 90.6% with all four. The Royal Champion is 5 to 7 points behind him every time. With the Warden out, nothing replaces him: the Prince leaves the army at 74.5%. Every case, for this army and the air army, is on the best TH18 army when a hero is upgrading.
If your execution is sloppy
Use the same army. Bad play costs every army about 19 to 50 points between flawless and very sloppy, and the optimized armies stay 15 to 31 points ahead of the best real build the whole way down.
| Quality of play | Best ground (3-star rate (%)) | Best air (3-star rate (%)) | Top real Thrower build (3-star rate (%)) | Top real Dragon build (3-star rate (%)) | 15 Thrower + 8 Healer (3-star rate (%)) |
|---|---|---|---|---|---|
| Flawless | 96.7 | 99.7 | 80.3 | 72.4 | 59.3 |
| Tidy | 94.3 | 98.8 | 79.7 | 76.1 | 47.3 |
| Typical | 90.6 | 98.1 | 73.2 | 67.2 | 29.8 |
| Sloppy | 81.6 | 88.5 | 58.0 | 51.8 | 14.8 |
| Very sloppy | 72.7 | 80.6 | 50.5 | 44.4 | 9.6 |
Three things matter more than the list when you’re not a precise player:
- Don’t copy the simplest army. Fifteen Throwers and eight Healers looks forgiving and is the least forgiving army tested: 59% in perfect hands, 10% in very sloppy ones, 31% on the real ladder. It has no tanks, no wall breakers and no second source of bodies.
- Deploy in one line and skip the funnel. For the typical player, dropping everything across one side scores 89%. Sending two heroes to the flanks first scores 60%, because the setup burns half a minute and the heroes fight alone.
- Watch the hero bars. A Revive on the first hero to fall is worth more than any other single action.
| Opponent bases | Best ground | Best air | Top real Thrower build | Top real Dragon build |
|---|---|---|---|---|
| 15% weaker | 98.5% | 99.7% | 89% | 87% |
| As calibrated | 91% | 98% | 73% | 67% |
| 15% stronger | 77% | 89% | 53% | 47% |
| Every base fully maxed | 45% | 80% | 18% | 11% |
The base you hit still matters: a 15% swing in defensive strength moves a real build 15 to 20 points but the air army only 2 to 9. Against the hardest bases the air army holds up far better, 80% against 45%.
Version and platform notes
The data is current to the late-September 2026 Thrower nerf and the October 2026 changes, live since 20 Oct (above). Re-run on the live numbers, the level-of-play chart and the opponent-bases table move by a point or less. Totem Spell, Sneezy, Ruin Witch, Angry Spell and Sky Wagon are on Supercell’s watch list for the next balance pass. Everything here applies the same on iOS and Android.
How we counted
We built a battle simulator from the game’s own data tables (every building, trap, troop, spell, hero, equipment item and pet at max TH18 level) and replayed 31 real armies on generated bases. The real results come from the TrophyCoach ranked-battle tracker: 57 exact TH18 builds over a 30-day window to 5 to 6 October 2026, 52 of them used.
Three settings were tuned until the simulator matched the measured results: defensive damage ×1.3, enemy base strength spread evenly from 50% to 100% of that, and a fairly sloppy player (slow deploys, spells off target, late hero abilities). It then ranks real armies in nearly the same order the ladder does: rank correlation 0.74, average miss 9 points, overall bias −3 points. Those are the re-run figures of 23 October, with the in-game Life Gem rule (below); before that fix, with the same three settings, they were 0.78, 9 points and −1.
| Item | Series | Simulated 3-star rate (%) | Measured 3-star rate (%) | Weight |
|---|---|---|---|---|
| Dragons, build 1 | Air | 62.5 | 60.5 | 2,066 |
| Dragons, build 2 | Air | 43.3 | 55.9 | 2,030 |
| Dragons, build 3 | Air | 59.2 | 61.0 | 1,089 |
| Dragons, build 4 | Air | 48.3 | 60.0 | 919 |
| Thrower, build 1 | Ground | 58.3 | 70.5 | 237 |
| 15 Thrower + 8 Healer | Ground | 20.0 | 31.2 | 218 |
| Thrower, Healer, Ruin Witch | Ground | 61.7 | 73.1 | 193 |
| Thrower, build 4 | Ground | 81.7 | 71.3 | 192 |
| Super Bowler, build 1 | Ground | 47.5 | 65.5 | 722 |
| Super Bowler, build 2 | Ground | 50.0 | 63.3 | 360 |
| Super Bowler, build 4 | Ground | 64.2 | 65.8 | 222 |
| Electro Dragons, build 1 (zap) | Air | 32.5 | 33.9 | 555 |
| Electro Dragons, build 4 | Air | 47.5 | 33.1 | 493 |
| Dragon Riders, build 3 | Air | 52.5 | 59.3 | 81 |
| Super Yeti, build 3 | Ground | 58.3 | 61.9 | 105 |
| Super Yeti, build 4 | Ground | 41.7 | 42.5 | 94 |
| Yeti smash, build 1 | Ground | 79.2 | 70.1 | 164 |
| Yeti smash, build 2 | Ground | 35.8 | 56.6 | 113 |
| Yeti smash, build 3 | Ground | 58.3 | 60.0 | 105 |
| Yeti smash, build 4 | Ground | 60.8 | 56.5 | 85 |
| Root Rider, build 2 | Ground | 38.3 | 48.5 | 66 |
| Root Rider, build 3 | Ground | 41.7 | 41.3 | 63 |
| Root Rider, build 4 | Ground | 50.8 | 36.7 | 60 |
| Super Archer + Ruin Witch | Ground | 72.5 | 66.5 | 215 |
| Thrower + 4 Ruin Witch | Ground | 74.2 | 44.8 | 67 |
| Ruin Witch mass | Ground | 71.7 | 61.3 | 212 |
| Witches, build 2 | Ground | 59.2 | 65.2 | 69 |
| Meteor Golem, build 2 | Ground | 37.5 | 38.1 | 84 |
| Meteor Golem, build 3 | Ground | 20.8 | 27.6 | 76 |
| Super Dragons, build 2 | Air | 50.0 | 55.1 | 107 |
| Super Dragons, build 4 | Air | 22.5 | 34.8 | 69 |
Each army result in this report averages 100 to 600 simulated attacks on max TH18 bases. The ranges under each chart are the uncertainty from that sample size.
Limits
- Everything maxed. Troops, spells, heroes, equipment and pets are at max TH18 level. Lower levels shift the ranking.
- Generated bases, not real ones. The test bases are procedurally built max TH18 layouts with the real building, wall and trap counts. Real anti-meta base design isn’t captured.
- A simple attacker. The simulated player deploys in a line or a ball, drops spells by rule and never does hero charges. Skill armies such as Duke Charge Hydra (61 to 82% in real data) score 13 to 37% here, so they are excluded, not disproved. Timing-heavy spells (Angry, Haste, Ice Block, Overgrowth) are probably underrated for the same reason.
- Not modelled: Clone and Recall spells. A few sub-unit and pet effects are estimates (Super Barbarian rage, Rocket Balloon boost, Frosty’s slow).
- Two results overruled. The air search first picked Electro Dragon spam at 81%, which the ladder ranks last (27 to 41%); generated bases pack buildings closer than real anti-chain layouts, so Electro Dragons were excluded. Ruin Witches are capped at four because the simulator overrates armies with more by 10 to 20 points.
- Ties are ties. With 150 to 300 simulated attacks per army, differences under about 5 points are noise. The top dozen candidates are all within that band.
- One correction so far. The first version let a troop keep the Life Gem bonus for the rest of the battle after touching the Warden’s ring; in the game it ends one second after the troop leaves. Fixing it made Life Gem + Heroic Torch the best Warden pair (88%) and added the Apprentice Warden to the army. On 23 October every chart and table on this page, the hero and spell figures in the text and the calibration were re-run with the fix and the live October numbers. The hero substitutes were re-run with the fix on 24 October. Not re-run: the October estimate tests (18 Oct) and the pet assignment, which still use the older rule.
Related questions
How many Healers should a TH18 army use?
Three. The first three or four Healers do most of the healing, and three and five score the same within noise (89% against 86%). From eight on, the healing stops growing.
What is the best siege machine at TH18?
The Troop Launcher. It scores 89%, against 84% for the Siege Barracks or Sky Wagon and 80% for the Log Launcher or Wall Wrecker.
Which hero can you least afford to lose at TH18?
The Grand Warden. Every line-up without him scores 50% or less, and dropping him costs 51 points, more than any other hero.
Is the air army better than the ground army at TH18?
Since the October changes, yes. On 600 fresh bases the best air army 3-stars 98% against 91% for this ground army, and 80% against 45% when every base is fully maxed.
What should you bring when a hero is upgrading?
The Minion Prince, for any hero: 87 to 88% with the King or Queen out and 82.5% with the Duke out, against 90.6% with all four. Nothing replaces the Warden: with the Prince in his place the army falls to 74.5%.
Sources
- Army meta, Town Hall 18 (30-day window)
- Balance Changes Incoming: October 2026
- Ruin Knight
- Town Hall 18 Crash Lands update notes
- July balance update
- August update notes
- The Anime Fury update notes
- Static game data (buildings, troops, spells, heroes)
- Decoded client tables 18.400.21 and wiki snapshot
- TH18 meta report 2026: what's working, what's failing
This content is not affiliated with, endorsed, sponsored, or specifically approved by Supercell and Supercell is not responsible for it. For more information see Supercell's Fan Content Policy: www.supercell.com/fan-content-policy.
Spotted something wrong or out of date? Report a mistake. Fixes usually go out within a few days.