Pick your three lead heroes (rally or garrison). The page shows the four joiners that add the most, or an easier-to-fill lineup within 0.5%.
Same lead heroes, same best joiners. Your stats don't change the answer. Only the battle type (PvP or Bear), the troop types sent, and the joiners you have do.
Rally power = Stats × Buffs × Skills
1. Stats. Your battle report % from gear, research, charms and heroes. They work in two pairs: (Attack × Lethality) and (Defense × Health). Inside each pair a point helps most on the lower stat, so raise the lower stat of each pair first. Joiners can't fix this.
2. Buffs (other multipliers). Widgets, pets, city buffs and titles (King, Marshal and so on). Same-stat buffs add up, then multiply your stats. Enemy-down buffs shrink their stat. In a rally, only the leader's widgets count.
3. Skills. The lead heroes' skills plus each joiner's first skill. Same-stat bonuses add up in one bucket, and the buckets multiply. This page only covers this part.
Kills scale with: your stats ÷ enemy stats × skill boost
└ fixed by you ────────┘ └ joiners only change this
A joiner's worth = skill boost with them ÷ skill boost without them
(your stats cancel out)
PvP counts both sides:
Damage you deal = (Attack × Lethality) ÷ (their Defense × Health) × your offense buckets
Damage you take = (their Attack × Lethality) ÷ (Defense × Health) ÷ your defense buckets
So two players with the same three lead heroes want the same joiners, whatever their stats. Each joiner brings its own multiplier, and crowding decides how much it really adds.
Heroes newer than this are left out everywhere, lead heroes included.
How much each skill group is boosted: first by your three lead heroes alone, then after the best four joiners are added. Only groups that end up with something in them are listed. The bottom row multiplies every group together: your combined boost from the skill groups joiners can share. Lead chance skills other than Petra's, and widgets, are not counted.
| Skill group | Lead team skills in it | Lead team total | With best 4 joiners |
|---|
Every expedition skill the three lead heroes bring, plus each hero's widget. Skills marked "not counted" still help the rally (except construction skills), but they are chance-based, limited to some troop types, or in an unpublished group, so they don't change which joiners are best. Widgets are not counted in the joiner math yet; a rally widget does nothing when you defend, a defender widget does nothing in your own rally, and solo marches get no widget bonus at all.
| Hero | Skill | What it does | Skill group |
|---|
your damage dealt damage you take
Which heroes can join your rallies? Every hero starts on. Within each group, heroes rated higher in Likely wanted come first. Tap one to turn it off (for example Petra, or a hero nobody in your alliance has built). Everything below recounts with only the heroes left on.
* Take only one Petra. A second Petra only adds three more 50% checks: her chance to trigger goes from 87.5% to 98.4%, about +5.5 points on average, so the slot is better spent elsewhere.
Best guess: the game doesn't publish skill codes. The skill groups here are our best inference from each skill's wording and other released information.
Each pick is counted on top of the picks above it. "Overlaps" names who already has a skill in that group.
Good enough beats perfect: when another lineup is at most 0.5% higher, the best four shows the one most alliances can fill (in order: fewer heroes from the two newest generations, not counting staples like Ava or Hero Roulette heroes like Diego; then fewer different heroes; then the higher value; exact ties go to heroes rated higher in Likely wanted, then to older heroes) and lists the higher one below as "harder to fill". Use whoever is easiest to get into the rally.
A general list for saved joiner formations. It uses the newest generation on your server (picked at the top). It checks every PvP lead team built from the three newest generations up to that one, plus Amadeus, Petra and Ava (common leads, once released on your server), and counts how often each ability is in the best four this page actually shows, using the same tie-breakers. Crowding is included: each lead team gets its own best four. Heroes you turn off in your joiner pool are left out here too. Always PvP with 3 troop types. High = in the best four for 50% or more of these lead teams, Medium = 20–49%, Low = under 20%, Never = not picked for any of them.
| What they bring | Joiners with it | Likely wanted |
|---|
What each joiner adds as the first and only joiner. These numbers change after each pick, because every pick crowds its group. For backups, turn off the heroes you can't get in Shape your joiner pool.
| What they bring | Joiners with it | Really adds |
|---|
Not every copy of a skill counts the same. On paper (its nominal value), every 25% copy adds +25%. Its real gain is relative: it is measured against the group total before it, so each new copy is a smaller step up from a bigger total.
| Copy in the group | 25% skill (Chenko, Gordon…) | 20% skill (Fahd, Eric…) | ||
|---|---|---|---|---|
| Group total | Real gain | Group total | Real gain | |
| 0 (empty) | 0% | — | 0% | — |
| 1 | 25% | +25.0% | 20% | +20.0% |
| 2 | 50% | +20.0% | 40% | +16.7% |
| 3 | 75% | +16.7% | 60% | +14.3% |
| 4 | 100% | +14.3% | 80% | +12.5% |
| 5 | 125% | +12.5% | 100% | +11.1% |
| 6 | 150% | +11.1% | 120% | +10.0% |
| 7 | 175% | +10.0% | 140% | +9.1% |
Up to 3 copies can come from the lead team, then up to 4 from joiners, so 7 is the most a group can hold. Copy 1 (highlighted) is a skill going into an empty group. A fresh 20% (+20%) ties a 25% that would be copy 2, and beats any 25% that would be copy 3 or later. Within one group, every 25% copy still adds the same 25 points: with 100 damage, copy 1 takes it to 125 and copy 2 to 150. The percentage shrinks because each copy is measured against a bigger total. The same copy in a different group would multiply instead (125 × 1.25 = 156.25), so crowding costs about 6 damage per 100 here, and more with each copy.
Bonuses to the same stat (attack, lethality, defense, health, and so on) add together inside one bucket. The buckets then multiply together for the final result. So each extra copy in a bucket is a smaller step up, and a joiner only adds its full card value when its bucket is empty.
| Row | In the group | Group total | This step adds | Joiners so far |
|---|
Row 0 is the lead team (rally or garrison): every skill its three heroes have in this group. Rows 1–4 are joiners added one at a time; only 4 joiner skills count. Effect is how much that row really raises the result. For debuffs, "enemy left" shows the same thing from their side. A −25% card (Ava) leaves them 80%, the same as +25% for you, not 75%. A −20% card (Fahd, Eric) leaves 83%, the same as +20%.
No. Kills come from your stats divided by theirs, so a cut to their side and a boost to yours move the result by the same amount.
You 1,500% vs them 4,000% Start 1,500 ÷ 4,000 = 0.375 You ×1.25 1,875 ÷ 4,000 = 0.469 Them ÷1.25 1,500 ÷ 3,200 = 0.469 same result
So the size of anyone's stats never decides between a buff and a debuff. Two things do: the size of the skill (Fahd and Eric are 20%, so they divide by 1.2), and how full its group already is.
Takeaway: always pick the joiner with the largest relative increase, whether it buffs you or debuffs them.
Left out on purpose: Vivian, because her lethality skill works out near 20% while Chenko and Yeonwoo give 25% in the same group. Thrud and Alcar, because their skills only reach some troop types, so their value depends on formation. Amadeus, because Chenko and Yeonwoo give the same 25% lethality and he is usually kept as a lead hero.
Petra is included because a rally always sends at least two troop types. Her skill makes one 50% check per troop type; if any check succeeds, the +50% applies. With 3 types she triggers 87.5% of the time (about 43.75% on average), with 2 types 75% (about 37.5%). Only one Petra is counted, and she is dropped if the lead team already has her. The troop types picker at the top only changes this Petra number.
Chance-based skills: outside of Petra, chance-based skills are traditionally not wanted for joiners. Chance and turn-based lead skills (for example "40% chance: damage dealt +50%") are listed but not counted, because nobody has published their skill group.
Charles (enemy lethality down) is assumed to be his own group, separate from Eric's enemy attack down, until a battle test shows otherwise.
Assumed: every joiner's first skill is level 5 (the hero is 4-star or better), so it brings the full % on its card.
Assumed: in PvP, your damage dealt and damage taken count the same (both sides attack every round). Bear is one-sided, so damage output is what matters.
Sources, most trusted first: Frakinator (frakinator.streamlit.app), Sloth / KingshotOptimizer, Daryl / kingshotguides.com; then kingshotjoinercalculator.netlify.app, kingshotcalculator.com and kingshotdata.com for skill wording. Gen 9 heroes are not in yet.