TODO: This is a work in progress; edits/feedback appreciated!

While AI players will often spend their extra gold on putting 40 province defense in every province, humans generally invest their gold into troops and infrastructure. As a result, it is commonly valuable to build raiding forces designed to beat 1-6 province defense with the minimal possible cost. One common way to do this is with thugs or mages scripted to buff elite troops or summon elementals, but it's also possible to do this cheaply with troops or summons.

One important thing to note is that what works depends significantly on the nation you're fighting, because province defense contains a national commander, and these vary widely in quality. While MA Marignon gets the unarmored human Friar (dangerous only via banishing undead), EA Vanheim gets a Vanherse, and Shinuyama gets a Bakemono General, both of which are quite capable compared with the rest of the PD combined. In many raids, the PD routs without the commander entering melee; but the nature of the PD commander can be very significant for raiders designed to snipe the commander.

In this guide, we will discuss several affordable options for beating 6 PD of a common province defense type reliably with summons and independent recruits. All of them can be countered cheaply by an opponent who anticipates what you're going to send, often as cheaply as raising PD to 10, adding a handful of national units, or bringing a single mage, perhaps with 1-2 gems. But those light defenses require the opponent to commit gold (for PD) or mobile troops to defend the province, and would lose to a stronger army of your own.

So while you may successfully use one of these exact compositions in a multiplayer game, or you can also supplement them with more elite resources designed to beat more determined resistance. (For example, some Iron Pigs, which are very efficient at dispatching PD, might help a Frozen Heart caster target an enemy thug, by killing all the other targets).

This guide is intended as source of data about about what troops/summons are particularly economical for fighting indies, as a source of inspiration for how you can assemble ultra cheap raiding parties.

For this guide, we will focus on an average PD type: Reliably 6PD in a province owned by ea-arcoscephale reliably with 20% or fewer losses. If you're expecting 1PD or are OK with a decent chance of failure, you can send less! The test was done using the Land Arena map, which has infantry types common in the middle age.

• 1 Myrmidon Champion PD commander. This an armored, skilled human is about average for national PD commanders.
• 6 Militia
• 6 Archer
• 3 Heavy Infantry

Note that some PD types, like Lion Tribe, are considerably weaker, whereas others, with Heavy Cavalry, Barbarian, or Crossbowman units, counter some of these compositions.

Note that while this small province defense force only costs 21 to create, it would cost 220 to recruit that force as a mobile army. However, independents are weak troops, and province defense forces uses poor tactics that mean the units engage at different times, and one can reliably beat PD with substantially cheaper squads.

To be displayed on this page, one needs to be able to succeed reliably with at most 7 gems of summons or 150 gold (assuming you spend another 30-60 gold on a commander).

Some items are sufficient to allow a mundane commander to defeat province defense.

• 3: Any commander with an Amulet of the Dead placed in a corner on Cast Spells. National scouts high HP, armor, and/or a shield like the Avvite Scout or Van Scout are ideal for this role, since stealthy is invaluable for raiders, and the main risk against low PD is your raider being killed by an arrow.
• 8: Any commander with a Bottle of Living Water.
• 10: A Sleeper is a rare commander summon without magic paths who can solo our province defense challenge. This option is more budget than it might seem, since you don't need to pay for a separate commander, and sleepers that are not raiding are useful for their primary job (the best summoned army leader).

These are parties that can be led with a cheap independent Commander or Barbarian Chief. We include here independent troops one can purchase with gold.

• 120-150 : 4-5 independent Heavy Cavalry, in the back, on Hold+Attack.
• 120-150 : 12-15 Heavy Infantry (even the weak EA spear ones!) in back on Hold+Attack. Expect to lose a couple.
• 150-195 : 12-15 Barbarian in front on Attack. Expect to lose a couple.
• TODO: Bone Tribe Beast Hunter.
• 4: 3 Iron Pig in the front.
• 5 1 Manticore on hold+attack rear is fairly reliable; there's some risk if he doesn't land on the commander. 2 Manticore win readily even without sniping the commander.
• 6: 2 Troll from Contact Trolls. Cheaper with Troll King's Court.
• 6 2 Hill Giant. Requires 2 mage turns.
• 6: 4 Great Eagle on Hold+Attack Rear.
• 7: 6 Kydnid in back on Hold+Attack.
• 8: 2 Cave Crab placed in the back. Their main problem is low attack skill; they do dramatically better with 1 experience star.
• 8: Half of an Animal Horde cast from a forest or swamp.
• 10 10 Lamia (8 with the Lamia Queen discount). Critical mass is important; with 8 one loses ~3.

Only economical with full Dragon Master discounts:

• 6: 4 Ice Drake. (4 units for 6). Inconsistent due to only 50% chance to fire breath weapon, so can take losses.
• 3: 3 Cave Drake. (4 units for 4). Low attack skill and high encumbrance mean fewer will fatigue out and die.
• 4 7 Wyvern. (5 units for 3). Hold + Attack closest, rear, and archers are all viable scripts.

TODO: Other Drakes are likely economical with full Dragon Master discounts as well.

Nations that picked Undead Leadership on their bless can lead these parties with a 60 independent Priest. Otherwise, usually the cheapest option is 3 for a Mound King.

• 4 8 Longdead Horseman from Pale Riders; both Attack and Attack Rear are options.
• 6 20-24 generic mixed longdead from Reanimation. Expect significant losses, but perfect morale means it's quite reliable. Late game, one can use a Mound Fiend to freespawn 20 longdead every 3 turns.
• 6 2 Ghost from Summon Ghosts. At slightly higher research, a Wraith Lord can freespawn one ghost a turn, which is cheaper than directly summoning them after about 15 turns.
• 8 (3 late game): 2 Wight. A Bane with a Wight bodyguard is preferable to a Mound King leading 2 Wight; before Legion of Wights, both options cost 11 total.
• 8 2 Ziz on hold+attack rear are fairly reliable. Their low protection means one can lose one to retreating infantry.
• 10 12 Bone Fiend. Requires two mage turns. Most losses are from archer fire.
• 15 (7 late game): 1 Demon Knight at front on Attack.
• 20 (15 late game): 2 Storm Demon at front on Fire.
• 33: 3 Vampire units. While this consumes 3 mage turns, the vampires are immortal, and thus can be used again repeatedly, making this potentially a budget option. Hold + Attack closest is preferred; the fight can go very long, as vampires get repelled a lot.
• The Hidden In Terrain spells are efficient, in the sense that the ~40-50 units summoned are probably fair to value at about 1 gem each, and 7 units (say, 2 Unfrozen Warrior and 5 Unfrozen in cold 2 beats our PD challenge easily without losses. But they're an extraordinarily expensive bundle.

Many undead and demons have perfect morale, and will keep fighting even after heavy casualties. However, if the army HP rout happens, any mindless undead will dissolve.

Nations that picked Magic Leadership on their bless can lead these parties with a 60 independent Priest. The Sleeper (10) is the cheapest generic summon with magic leadership, which is not budget at all. But some national commanders have magic leaders that don't cost 2 commander pointers: The Batab, Hybrid Commander, Naga Chief, Beast Trainer, Attendant of the Oracles, Shuten-doji, Illithid Lord, and Houri.

Otherwise, you'll be bringing a mage. The examples below assume you're bringing the indy priest; a mage substantially increases the cost of what you're bringing, and also dramatically changes your options. Even a 1 casting Protection or a 1 casting Body Ethereal and Luck is enough to empower 1-2 squares of unremarkable troops to beat our PD challenge.

Construct magic beings often have the never heals tag, and may need to retreat to a lab to heal before a second raid.

• 12: 2 Dark Vines, 2 1-unit squads at the front is best to avoid infantry slipping past your slow, tanky plants.
• 5: 6 Fire Ant. Stealthy!
• 6: 2 Green Lion on Fire Closest.
• 7: 5 Fire Snake.
• 8: 2 Summer Lion.
• 7: 5 Mechanical Men.
• 8 4 Gargoyle.
• 8 4 Living Statue. As few as 2 could win, if not using an indy priest as the commander (infantry consistently run past 2 and stab the priest).

Many national summons are more economical than the above options for raiding PD. This is especially true for sacreds if you have a strong bless. Options that have been tested are collected here.

• 2: 5 Magma Child [Agartha]
• 3: or freespawn: 5 Longdead Rephaite from Reanimation [Hinnom; Ashdod; Gath].
• 5: 1 unblessed Morrigan [Fomoria]
• 15: 3 unblessed Rakshasa Warrior [Lanka; Bander Log; Patala]
• 4: 4 Karasu Tengu on Hold+Attack Archers [Yomi; Shinuyama; Jomon]

The following are options which are not efficient raiders, in the sense that they either outright lose with 8+ gems worth of the summon, require 3+ mage turns to mass, or win with major losses (Mainly happens to units with perfect morale).

Flyer results are highly RNG dependent, depending whether the PD commander is near the archers.

• 6: Two commanders each with Bow of War lose.
• 10 5 Serpent Fiend take 5 mage turns, and can lose even when landing on an isolated commander.
• 15 5 Spine Devil take 5 mage turns. Likely win due to 30 morale, but 2 losses is typical.
• 16: 2 Frost Fiend on Hold+Fire can win, but it's unreliable even in Cold 3.
• 20: 4 Fiend of Darkness can win if they snipe the commander, but can also be wiped. Could be economical with Forces of Darkness discount.
• 21: 3 Devil can win if they snipe the commander, but can also be wiped and often taken attrition. Could be economical with Infernal Forces discount, as extra imps really help.
• 12: 12 Fiery Imp often rout to archer fire; more requires 3 mage turns.
• 8 40 Wolf from Pack of Wolves (or Call of the Wild) lose.
• 8 10 Killer Mantis win with ~30% losses; 7 lose badly.
• 6 2 Amphiptere lose; more requires 3 mage turns.
• 8 10 Horned Serpent win with ~30% losses; 7 lose badly.
• 10 2 Kithaironic Lion lose.
• 10 25 Sea Dog might lose, or suffer >50% casualties if victorious. Requires 5 mage turns.
• 7 5 Bog Beast lose.
• 9 8 Leogryph lose.
• 8 10 Woodland Spirit win with 20% losses. Also homesickness is bad for raiders.
• 9 3 Lumber Construct lose, despite requiring 3 mage turns.
• 6 3 Forest Giant win, but require 3 mage turns. 2 lose badly.
• 7 4 Swamp Drake (28 without Dragon Master discount!) can lose; poison routs the infantry too slowly.
• 9: 12 Manikin lose. (This spell scales particularly well with path).
• 10: 5 Mandragora lose; 7 win with 2 losses. (This spell scales particularly well with path).
• 6 (with Ivy Crown discount): 6 Vine Ogre consistently suffer losses, and can HP rout and all die, even with multiple squads to reduce bunching. More requires 3 mage turns.
• 5: 25 Vine Man wins with 30% losses. Requires 5 mage turns with Ivy Crown discount.
• 10 1 Tartarian Monstra unit loses often, even starting with no afflictions. Likely can win if it gets a lucky rout from its fear.
• 12 A solo Bane Lord with his default gear loses.
• 7 1 Behemoth loses.
• 8 ~30 Ghoul from Around Hunger. Comes with a Ghoul commander, but also can lose.
• 10 20 Longdead Archer lose badly.
• 8 5 Cave Grub lose badly.
• 8 8 Clockwork Soldiers fatigue out before reaching the archers.
• 8 8 Clockwork Horrors fatigue out after routing the PD but before killing the commander. Most of them die to excessive fatigue during the battle cleanup phase.
• 8 4 Ogre can lose, or win with 1-2 losses.
• 8 3 Fall Bear lose badly any season.
• 15 1 Crusher loses eventually to commander dying. He rarely hits past shields, and damage is too low to kill through a shield.
• 4: 24 Corpse Construct is a tossup. Requires 4 mage turns, though within budget on gems.
• 8 2 Spring Hawk lose badly any season.
• 10 5 Draconian can lose or suffer major attrition even when landing on isolated commander.
• 9 5 Gryphon suffer 40% losses even if they perfectly land on the commander.
• 10 Call of the Winds can lose without killing anything.
• 8 5 Winter Wolf suffer attrition even in winter.
• 8 3 Yeti lose badly.
• 15 15 Cave Cow lose badly. It's interesting – they are undisciplined, and it doesn't seem possible to get them to target the infantry with their ranged attack, so they just fire ineffectually at the archers while being chopped up.
• 8 4 Gelatinous Cube lose ~2; archers do too much damage.
• 10 10 Clayman win with ~20% losses.
• 6 3 Scorpion Beast. Can win with 2; unreliable below 4.
• 10 10 Terracotta Soldier win with ~50% losses.
• 200 : Even 20 Deer Tribe Warrior is likely to lose.
• 200 : A War Elephant often loses.

National options:

• 7 9 unblessed Barghest from Summon Barghests win with ~50% losses. [Fomoria]
• 8 20 Black Dog lose badly [Fomoria].

Please be encouraged to add additional options to the guide! Test using the Arena map with defender Dominion present (affects Morale). With the Debug mod, you can Wish for national summons of other nations. Generally, an option should be documented with the minimum number of troops required to win reliably (Wins 3 times in a row with ⇐15% losses all 3 times). If unreliable or finicky, record that. Document the formation if you had a materially worse result with other formations you tried.

Calculate gem costs for summons using the minimum paths for the spell.

TODO: Recheck all the gem costs for this algorithm.