Building New Shadow War: Armageddon Armies: A Statistical Approach

Shadow War: Armageddon has created some serious buzz. Enough that the limited release of the boxed set constituted a genuine crisis for some folks, and enough that Games Workshop appears to be scrambling to provide additional support.

One of the things they’ve done is release new force lists for armies outside Space Marine Scouts, Imperial Guard Veterans and Orks.

This is awesome.

But can we take it further?

The forces GW chose to look at seem pretty clearly chosen primarily as “Light infantry well supported by plastic kits”. Which means there are some glaring holes – Kroot infantry stalking (and snacking) their way through ruins. Eldar Rangers, which is an omission so obvious it’s painful. Eldar Corsairs, because surely wealthy up-hive traitor nobles would hire exotic bodyguards. Sisters of Battle (rumored to be coming). Dark Eldar Kabalites instead of Wyches. Horus Heresy-era forces because Shadow War: Terra seems like a blast…

You get the idea.

The nice part is since SW:A lifted most of the units from Warhammer 40K codexes, there’s the possibility of nice, clean conversions between them. But they’re very clearly on different point scales, so I started wondering if there’s an easy translation between the games. Someone suggested “40K cost X 10” as a good rule of thumb, and while that’s okay it’s not amazing.

So, as always, I turned to the data itself.

I compared the cost of SW:A to their nearest 40K analog to see if there was a nice regression fit that would get you close. The good news? There is.

Now admittedly, getting the two things to line up involves some fuzzy math. I ignore (for now) that 40K units tend to come with more equipment. Some unit costs are assumed (for example, a Veteran IG Sergeant is assumed to cost 10 points). And I only use the forces I have access to – which means primarily no Tyranids or Genestealer Cult units.

Lets just look at the raw relationship first, to see if this is worth pursuing:

There’s clearly something there – while there’s definitely some noise, it looks like there’s a strong linear relationship between the two scales. One thing I added to make the model a little more complex was an indicator of whether the unit came from the book or the web (e.g. Space Marines vs. Dark Eldar) in case the possibly rushed online content was a little less reliable. As it turns out, that’s true. Web units got a value of 1, while book units got a value of 0. When you run a linear model, you get the following formula to convert a 40K unit to an SW:A unit:

SWA = 8.88 + (8.74*Codex Points) + (-21.19*Supplement)

Graphically, the fit is pretty decent

A perfectly predictive model would have all those dots sitting on the red line. Right now, for the most part, they’re pretty close. On average the model prediction is off by about 15 points, sometimes high and sometimes low. Compared to the “multiply by 10” rule-of-thumb, which is off by an average of 28.7 points. So a pretty significant improvement.

The Web Units:

 

Digging a little deeper, lets look at that last term in the formula: “(-21.19*Supplement)”. That means supplement models are, in general, undercosted by about 21 points compared to their book counterparts. Breaking down how well the model fits, they are also less clean translations – when you break down how much the model was off by between book units and web units, the average “miss” is 8.87 points vs. 18.5 points respectively. There’s almost twice the error in the conversion. The same holds true for the rule-of-thumb version but it’s actually worse, with about triple the error.

Now, this could be from any one of a number things. It may indeed be that the web units were rushed, and a little less thought was put into them. It could be that the more you deviate from the relatively clean rules of the Space Marines, IG and Orks into exotic things that the designers felt that the points weren’t reflective. We’ll see in time.

New Units:

What about the new units – the newbie troops and special weapons types that don’t exist as separate entries in the codex. By and large, the specialists are just a generic trooper + 10 points, consistently enough that I didn’t bother modeling it. The junior troopers are a little more difficult, so I ended up modeling those, and got a base cost of about 0.88 times that of the standard trooper.

Using This In Practice:

So how do we actually use this? Lets say we wanted to create a new faction, the Sisters of Battle (I chose this before they were released…).

We need a commander, a standard trooper unit, a junior trooper and a specialist.

From the 40K codex, a Battle Sister is 12 points, and a Veteran Sister Superior is 22 (12 for the Sister Superior + 10 for the Upgrade). Feeding that into the model (and ignoring the undercosting of non-book units):

SW:A Battle Sister = 8.88 + (8.74 * 12) = 113.76

SW:A VSS = 8.88 + (8.74*22) = 201.16

SW:A Neophyte Battle Sister = 113.76 * 0.88 = 100.11

SW:A Battle Sister Specialist = 123.76

“But Variance Hammer, no one is going to let me play my codex with two digit decimal point costs…”

Excellent point hypothetical reader!

If we round points cost to the nearest 5, so 115 and 205 respectively, the model fit isn’t actually all that much worse (going from being off by an average of ~ 15 points to an average of 15.5 points) for a significant gain in sanity. So lets round that to:

SW:A Battle Sister: 115

SW:A VSS: 200

SW:A Novitiate Battle Sister: 100

SW:A Battle Sister Specialist: 125

The costs are pretty in line with what the nearest analog army in the book, the Space Marine Scouts cost, the Sisters being marginally more expensive in 40K as well. Hilariously, in the middle of writing this post GW released the points costs!

SW:A Battle Sister: 115

SW:A VSS: 200

SW:A Novitiate Battle Sister: 100

SW:A Battle Sister Specialist: 125

How’d we do? Well…the Specialist and Novitiate costs are right on if you adjust the Battle Sister score – the Specialist is + 10 points, the Novitiate is 0.88 X Points almost exactly. The Battle Sister herself is a touch expensive at 115 vs. 90, and the VSS is the same, at 200 vs. 175.

Herein lies a twist – if I had used the full regression model, which also suggests subtracting 21.19 points for any model not in the main book, we would have gotten 95 points for a Battle Sister and 180 points for a VSS, which are very close to what actually came out. This seems to confirm the idea that the non-book units are actively having their points costs adjusted downward. Curious.

Equipment:

Some of you may have noticed that I’m entirely ignoring equipment right now. There’s a reason for that – the points costing for equipment is a mess. I built a list of equipment with clear SW:A to 40K analogs (mostly weapons, some equipment) and did a bit of adjusting to some of them. Most of this was when a 40K unit had a weapon and the upgrade is replacing, whereas in SW:A it’s purchasing the weapon outright. In these cases, I subtracted the cost of the base weapon in SW:A from the cost of the upgraded weapon. So for example, it costs a Wych Syren 15 points to replace her splinter pistol with a blast pistol. Wyches in SW:A don’t have splinter pistols, so I subtracted the 20 points it would cost to have a splinter pistol from the 50 points it would cost to have a blast pistol to arrive at a cost of 30 points. I’ve dropped the Harlequin weapons from this analysis because it’s a codex I don’t have electronically and I’m at a Starbucks, and because of the large amount of equipment in the game, I’m not starved for data.

First, the raw relationship:

Like I said – all over the place. And yes, there are even some negative values in there – after doing the adjusting I mentioned above, several Grey Knight weapons are negatively valued because the Nemesis Force Sword is more expensive than several other options.

Trying to model this, we get:

SW:A Cost = 27.42 + (5.00*Codex Cost) + (-17.83 * Supplement)

Again, some evidence of undercosting in the supplements, but I wouldn’t make much of it because…well, this is the fit:

That’s bad. Really quite bad. No one should be proud of that fit. Gear in SW:A is weird. Part of it is that it’s a much greater departure from 40K – there’s a lot more stats and granularity on the weapons.

So my advice? Use the formula (the full formula) for unit costs, and then for equipment costs, try to find the nearest existing analog in another army. Because there’s really no evidence that equipment can be ported over directly without massive errors.

Happy homebrewing everyone!

Enjoy what you read? Enjoyed that it was ad free? Both of those things are courtesy of our generous Patreon supporters. If you’d like more quantitatively driven thoughts on 40K and miniatures wargaming, and a hand in deciding what we cover, please consider joining them.