Kings of War 4.1 Army Efficiency Rankings After Fracture

Fracture is the first real update point for Kings of War Fourth Edition, so I am going to call this “4.1” for the purposes of the data. That label is not official but is just useful shorthand. The launch version of 4th Edition gave us one set of army lists, unit costs, and early assumptions. Fracture gives us the first major chance to ask how much those assumptions changed.

When I first ranked Fourth Edition armies by unit cost-effectiveness, Empire of Dust and Undead looked excellent. That made sense in the simulator. They had efficient profiles, strong defensive baselines, and plenty of units that performed well in direct combat. Since then, Empire of Dust has generally been recognized as one of the weaker 4E armies on the table, while Ratkin — a faction my model had near the bottom — has looked like one of the strongest competitive armies. That is the useful tension in this article.

This is not a full army tier list. It is a look at which armies have the most cost-effective units in a controlled combat model. That tells us something real, but it does not tell us everything that matters in Kings of War.

Small sample and data warning

This is an early 4.1 pass. I expect some of these numbers to move. New-edition data is messy. Army lists get updated. Profiles get renamed. Categories change. Special rules are easy to misread or miscode. Some errors only show up after people start building lists, asking rules questions, and catching edge cases.

That is especially true right after a book like Fracture. The data set I used has 956 4.1 profile rows and includes Elo, Elo per point, Elo difference, and combined cost effectiveness. I cleaned obvious label issues for the analysis, but there are still a few flags worth keeping in mind.  So treat this as a working model, not a final verdict. I reserve the right to be wrong once the next batch of corrections punches the spreadsheet in the face.

What the model is measuring

The simulator starts by testing units in combat. Each unit profile is run through simulated fights against other unit profiles (a total of 1.25 million simulated combats). Those fights use the unit’s basic combat stats and the special rules I have coded into the model. The output is an Elo-style rating. A higher Elo means the unit performs better in the simulated combat environment.

That gives us raw combat strength, but raw strength is not the same as efficiency. A 260-point horde should beat a 90-point regiment. The useful question is whether the unit performs better than its points cost suggests. That is where cost effectiveness comes in.

For each unit, I calculate two main efficiency measures:

1. Elo per point
   This rewards units that produce a lot of combat value for their cost. Cheap units that perform even slightly above expectation can score very well here.
2. Elo difference
   This asks whether a unit’s actual Elo is higher or lower than we would expect for its points cost and unit type. It helps prevent the model from only loving the cheapest profiles.

Both numbers are standardized across the data set so they can be compared on the same scale. The final cost-effectiveness score is a weighted blend:

Combined Cost Effectiveness = 60% Elo per point + 40% Elo difference

That weighting is a choice, not a law of nature. I use it because I want the model to reward efficient units, but not turn every cheap speed bump into a superstar.

How the army score is built

Fourth Edition army construction is based around categories, so the army score uses those categories too. For each faction, the script selects:

The best profile in each category gets a 1.5x weight. The second profile, where applicable, gets a 1.0x weight. That creates a “best-in-slot” score. It is meant to answer this question: If I were trying to build around the most mathematically efficient pieces in this army, how strong is the package?

An army can score well because it has one extreme outlier. An army can score poorly because its strengths come from command orders, movement, scenario pressure, redundancy, auras, spells, shooting angles, or other things the simulator does not fully capture. Ratkin are the clearest warning label here.

The 4.1 best-in-slot rankings

The top of the table changed a lot. Empire of Dust and Undead were first and second in the 4.0 version of the model. In this 4.1 pass, they fall to seventh and sixth. Goblins move into first, followed by Kingdoms of Men, Xirkaali, and Ogres.

Empire of Dust has the largest best-in-slot drop at -3.360. Undead drops -2.890. Ogres has the largest gain at +1.305, with Kingdoms of Men, Xirkaali, and Abyssal Dwarfs also making clear gains. The practical read is that the early undead efficiency package was pulled back. The top of the model is now more compressed, and the leaders are not all leading for the same reason.

Cost effectiveness is not the same as army strength

This is the most important part of the article. The CSV also includes actual simulated Elo, not just cost effectiveness. When we rank armies by average Elo, the picture changes.

That table is why I am being careful with the language. Goblins rank first in best-in-slot cost effectiveness, but only nineteenth in average Elo. Kingdoms of Men rank second in cost effectiveness, but eighteenth in average Elo. Xirkaali rank third in cost effectiveness, but sixteenth in average Elo. That does not mean the cost-effectiveness score is useless. It means it is measuring something different.

Cheap units can be extremely efficient without being individually powerful. Expensive units can be very strong while looking less efficient per point. A faction can have a few efficient list-building pieces but still struggle to convert that into a complete tournament army. A faction can also have mediocre raw unit scores and still win games through pressure, scenario play, redundancy, and matchup control.

Ratkin are the obvious example. They remain seventeenth in the best-in-slot table and twentieth in non-character average Elo. On the table, though, Ratkin have been one of the armies players actually need to solve. 

What changed for the Fracture armies

Fracture substantially updates four armies: Ogres, Forces of the Abyss, Forces of Nature, and Basileans. The data points in different directions for each one.

Those are four different redesign stories.

Ogres gained the clearest raw strength

Ogres are the cleanest winner in this pass. Their average Elo rises from 1587.3 to 1636.8. Their non-character average rises from 1648.3 to 1704.3. That puts Ogres first in average Elo and first in non-character average Elo in the 4.1 CSV. Their best-in-slot score also jumps from 9.840 to 11.145, the largest gain in the model.

The Red Goblin Blaster is doing a lot of work here. It has a 4.03 cost-effectiveness score, which is the highest individual score in the current data set. The Goblin Blaster is right behind it at 3.97, which also explains why Goblins top the best-in-slot table.

For Ogres, the player takeaway is pretty simple: the army looks stronger in both raw Elo and cost efficiency. The caution is that some of the best-in-slot gain is concentrated in Support. If that profile gets corrected, restricted, or simply proves less useful on real tables than it does in the model, the army score could shift.

Forces of the Abyss look healthier, but different

Forces of the Abyss are interesting because the average roster improved more than the best-in-slot score. Their average Elo rises by 39.2. Their non-character average rises by 58.6, the largest non-character gain among the four Fracture armies. That points toward a healthier body of the army. The best-in-slot score only rises by 0.260, though, and the army slips from twelfth to thirteenth because other factions improved faster.

The unit-level changes help explain why. Succubi Regiment gains 118 Elo, the largest matched-profile gain in this pass. Lower Abyssals and Abyssal Guard also look more relevant. At the same time, Tortured Souls and the Well of Souls take major hits. Tortured Souls Regiment drops 285 Elo, Tortured Souls Troop drops 244, and the Well of Souls drops 221.

That looks like an internal power transfer. The army is stronger overall, but some of the old efficiency anchors are no longer carrying the same weight. For players, that probably means Forces of the Abyss deserve a fresh list-building pass rather than a simple update to old lists. The good pieces may still be there, but the jobs have moved around.

Forces of Nature gained options and lost average strength

Forces of Nature is the awkward one. The army adds ten non-character options, more than any other faction in this pass. At the same time, average Elo drops by 90.6 and non-character average Elo drops by 68.2. That sounds bad, but the best-in-slot score rises from 8.170 to 8.885. The rank also improves one place, from thirteenth to twelfth.

That combination tells us something useful. Forces of Nature may have gained a wider toolbox, but the average tool in the box is weaker. The best package improved, while the overall roster became easier to build badly.

The selected package also changes a lot. Tribal Warriors and Forest Shamblers become the top Core choices. Earth Elementals move into the Specialist package. Greater Earth Elemental and Guardian Brutes become leading Support options.

For players, this points toward a roster-construction test. Forces of Nature may reward players who identify the efficient spine early. It may punish players who assume the expanded roster means every new option is equally safe.

Basileans improved below the headline

Basileans look almost flat if we only check the best-in-slot table. Their score rises from 6.260 to 6.435. Their rank stays sixteenth. Their all-profile average Elo actually falls by 5.2. The non-character number is much better. Basileans gain 49.4 non-character Elo, rising from 1522.8 to 1572.2. That suggests the ordinary list-building pieces improved, while the all-unit average is being pulled around by the broader character pool.

Men-at-arms, spear-armed Men-at-arms, and Ogre Palace Guard all look better in the efficiency model. The Support category still seems to be the problem. In the best-in-slot package, Elohi Troop and Phoenix both carry negative cost-effectiveness scores, which means even the best available Support choices reduce the total. For Basilean players, I would not read the sixteenth-place ranking as “nothing changed.” The rank is flat, but the army underneath it looks healthier.

The biggest exact-profile Elo movers

These are matched profiles with the same faction, unit, type, and size in both versions. Renamed or rebuilt units are excluded, so this is a conservative table.

Biggest gains

Biggest declines

The declines are larger than the gains. That fits the broader 4.1 pattern. This update looks more like a correction of several strong 4.0 pieces than a broad power increase across the game.

What I got wrong the first time

The honest part of this project is that the model missed some important things. Empire of Dust looked strong in the launch model. In actual play, the faction has generally been treated as one of the weaker 4E armies. Ratkin looked weak in the launch model. In actual play, Ratkin have been one of the armies players are most worried about. That is a useful reminder of what direct-combat simulation can and cannot see.

The model sees stats, costs, and coded special rules. It can tell us whether a profile punches above its points cost in a controlled fight. It can find units that are probably undercosted. It can show whether an army has efficient building blocks.

It is much worse at measuring the things that decide many Kings of War games:

• scenario pressure
• Unit Strength that survives into turns five and six
• threat projection
• board control
• chaff quality
• spell and aura value
• matchup spread
• deployment flexibility
• unlock pressure
• how easy the army is to pilot
• how hard the army is to play against

That is why Ratkin can look poor in the model and strong in events. A high-pressure army with good redundancy and scenario play can win games without winning a clean unit-efficiency contest. It is also why Empire of Dust can look good in a simulator and still feel underpowered on the table. If the army’s efficient pieces do not translate into board control, scoring, or favorable matchups, the raw math does not carry the faction.

Player takeaway

Edition 4.1 does not give us a clean new tier list. It gives us a better set of questions.

Goblins lead the best-in-slot efficiency model, but their average Elo is low. Ogres look strong in both average Elo and cost efficiency. Forces of the Abyss appear healthier, though some familiar pieces were cut down hard. Forces of Nature may have gained a broader toolbox but a lower average floor. Basileans probably improved more than their rank suggests.

The bigger lesson is that unit efficiency is only one part of army strength. The model can tell us where the cheap power is. It can tell us which units look undercosted in direct combat. It can show how Fracture moved the math.

The table will tell us the rest. If Ratkin keep winning, if Ogres start converting, or if Empire of Dust remains weak despite improved cleanup, those results will teach us what the simulator still cannot see.