"40k Metrics"

[Something Wycked]

Again, I'm sorry sorry sorry, this is pure Mathhammer. If you don't like Mathhammer, don't read on. :)

 

I've seen a couple posts about this here on the B&C, and I thought it at least deserved a nod in the Tactica forum.

 

This BoLS post talks about using the average damage output for an army in four categories to gauge its potential. The four metrics are Marine Equivalents (MEQ) killed in shooting and close combat, Rhinos destroyed, and Land Raiders killed, all with a multiplier of 5 for the minimum turns played.

 

As a base example, the author dissected the four undefeated armies at the NOVA tourney:

 

Tony Kopach (Space Wolves)

 

DMS: 19.28

DMCC: 28.29

DRPG: 59.18

DLRPG: 17.04

 

Andrew “Stelek” Sutton (Space Wolves)

 

DMS: 25.05

DMCC: 24.82

DRPG: 64.80

DLRPG: 19.27

 

Justin “Dashofpepper” Hildebrandt (Orks)

 

DMS: 18.90

DMCC: 45.75

DRPG: 80.00

DLRPG: 31.04

 

Mark Ferrik (Blood Angels)

 

DMS: 22.21

DMCC: 15.15

DRPG: 82.35

DLRPG: 33.74

I was curious. I wanted to know how my emerging Grey Knights list would measure up!

 

Dead MEQ, Shooting: 18.074

Dead MEQ, Close Combat : 18.222 (note 1)

Dead Rhinos: 23.705 + 42.395 = 66.1 (note 2)

Dead Land Raiders: 5.925 + 26.565 = 32.49 (note 3)

 

1: The CC number might be low, it looked as though the program I was using wasn't doing the rerolls properly. I assumed the Champion would get 3 hits in- might be a bit over the top. But eh, its mathhammer.

 

2: This metric is supposed to be shooting stats only, except for armies that focus heavily on melee to get the job done. For me, the first number is all the

squads shooting (minus the DK's HIs) and the second number is the DK's assaulting. Also, the metric is supposed to take into account Penetrating hits only; I added in the "Weapon Destroyed" and "Immobilized" results from the Stormbolter glancing hits, halved. That is, if a Rhino has been made into a wreck due to being weaponless and immobilized, that counts to me. I didn't use the full number of Glancing hits (obviously,) just added their chances together and then divided by two to form a psuedo-penetrating hit for these numbers.

 

2: This metric is supposed to be shooting stats only, except for armies that focus heavily on melee to get the job done. For me, the first number is all the squads shooting (minus the DK's HIs) and the second number is the DK's assaulting.

 

I'm rather impressed by this- I think the list I intend to be competitive really has a chance with the amount of damage output it can bring to the table.

 

Thoughts? Also, don't be afraid to do the math for your list to see the killy potential, I think it would be interesting to compare everyone's results.

[appiah4]

I measured my list as well.. My results were:

 

DMS 25,2

DMCC 27,1

DRPG 63,7

DLRPG 18,3

 

Not sure how it compares..

[Laughingdagger]

DMS: 48.72

DMCC: 21.92

DRPG: 42.93 (/3 = 14.31)

DLRPG: 26.64 (/3 = 8.88)

 

 

For DRPG and DLRPG are you guys dividing the penetrating hits by 3? Since it takes roughly on average 3 to wreck one or LR? (Or so the original author of the article stated)

[Something Wycked]

I measured my list as well.. My results were:

 

Not sure how it compares..

Very very comparable to the NOVA lists! :cuss Your list, by this metric, has a lot in common with those very competitive lists.

 

For DRPG and DLRPG are you guys dividing the penetrating hits by 3? Since it takes roughly on average 3 to wreck one or LR? (Or so the original author of the article stated)

The way he wrote that is actually pretty confusing- he's not dividing the scores by 3. He's limiting each CC unit to 3 penetrating hits per turn, if they would ordinarily score more- so each CC unit is limited to a score of 15 for a game (3 penetrating hits per turn x5 = 15).

 

Edit: I miscalculated the MEQ shooting and CC scores- I multiplied them by 5 because of the vehicle formula, but the MEQ scores are supposed to be per-turn.

[GreatEscape_13]

I'm generally all for application of statistics to questions of what units function better than others. But this sort of total grading seems very curious to me.

 

Okay, so I've got these four numbers. And they all are reasonably useful tools for telling how an army does one phase (shooting, cc, transport-busting, heavy armor breaking). But how do we make legitimate comparisons across the four numbers. They cannot be easily combined into a single score, without some very subjective structuring of the compound statistic (are the four weighted equally, what about the early-turn need for transport popping declining as they disappear, etc.).

 

So the previous two posts reported their lists as (respectively):

 

DMS: 48.72

DMCC: 21.92

DRPG: 42.93 (/3 = 14.31)

DLRPG: 26.64 (/3 = 8.88)

 

DMS 25,2

DMCC 27,1

DRPG 63,7

DLRPG 18,3

 

Okay, so now what? Do we compare the lists to each other? How do we view these two? Well, one is better at popping transports while one is better at killing marines. Wow. Did we need a computer program to do that? Or could we have just looked at the lists and said "ah, lots of AP 2 in this list, vs. ooh, lots of AC spam in this list" and known which was better?

 

One comparison that can be made is that we see how those lists rate against the same categories of four armies that won at one particular tourney. So what do we learn from that comparison? We would need to at least consider composition of opponents for an appropriate measure of comparison. What if tourney-winner space wolves army #1 earned his scores when facing all MEQs all day, while tourney-winner space wolves #2 faced nothing but cheap infantry hordes. And the BOTH went undefeated due to such important factors as: tactics, set-up, first turn, and the random ravages of dice. Making such comparisons numerically to the armies on this metric loses that portion of the story, which is at least as important I might think.

 

Anyhow, if this metric is going to be desirable for comparisons across different army lists, then we better find ways to ensure that we understand what comparisons can be made with it and what cannot be judged with it.

 

Cheers.

[DevianID]

The issue with this metric is that the numbers are misleading.

 

IE, consider a unit of 15 lootas. Their anti-rhino numbers would be 1.11 per turn roughly, and 5.56 for a game. That said, the Loota unit can never kill more than 1 rhino in a turn, and they do not have a 100% chance to destroy 1 rhino. A squad of rhinos, sure, but those dont exist.

 

Also, for killing MEQ with shooting, some weapons work better when the unit has no cover, some weapons dont care about cover. Blast weapons can kill more meq than anything else, but only if bunched. Thus the MEQ numbers are not consistant between armies and terrain.

 

I am not saying the metrics have no place, BUT many comparisons of the numbers are provenly false, because there is an unknown factor in the equasion concerning things like whether to include cover, or how many models will be hit by a blast/template, or if you total expected values knowning that expected values for units like lootas will provide incorrect information.

[Something Wycked]

Great points to be raised. I should have brought up my own misgivings about the numbers in the first post. These numbers absolutely do not take into consideration the "fuzzy" areas of playing the game on a board- the nebulous parts of generalship that can't be quantified.

 

They're a good baseline, and nothing more; they give an insight into the killing potential of the list given the best of conditions- perfect range, no cover, your units always getting the charge, etc.

 

IE, consider a unit of 15 lootas. Their anti-rhino numbers would be 1.11 per turn roughly, and 5.56 for a game. That said, the Loota unit can never kill more than 1 rhino in a turn, and they do not have a 100% chance to destroy 1 rhino.

That's why the metric is based on average penetrating hits and not statistical kills :lol:

 

BUT many comparisons of the numbers are provenly false, because there is an unknown factor in the equasion concerning things like whether to include cover, or how many models will be hit by a blast/template, or if you total expected values knowning that expected values for units like lootas will provide incorrect information.

I wouldn't go so far as to say "provenly false" because this is a baseline; no cover is taken into account to see what kind of numbers your army can put out under optimal conditions. The blast/template question is resolved by static figures just for the sake of comparison (the blog has those numbers listed.)

 

Anyhow, if this metric is going to be desirable for comparisons across different army lists, then we better find ways to ensure that we understand what comparisons can be made with it and what cannot be judged with it.

Exactly the right direction to go with it. The only comparison this style of mathhammer gives is optimal killing potential, if everything is going your way- your opponent isn't using cover, your units are at full strength, the targets are at your optimal ranges, and you get the charge off.

 

Just a baseline :rolleyes:

[Paulochromis]

Should also point out that the original author suggests that inter-codex comparisons aren't all that useful, but intra-codex comparisons can be.

Use the metrics as a way of seeing how your list measures up to "respected" lists from your own codex.

 

Cheers, Paul.

[ilnar]

Probably also worth considering unit composition.

Consider a tactical squad, plasma gun and LL

The PG is clearly an anti MEQ gun and the LC is an AT weapon.

Clearly the unit cannot fire both.

But a DA 5 man tac squad with PG and a 5 man dev squad with LC can fire both. At almost the same points cost.

So it becomes not only a matter of the theoretical damage possible per turn, but how much of that fire power can be utilized in a given turn at a given target.

[Laughingdagger]

A 10 man tac squad with PG and LC CAN fire both, combat squads my friend.

[GreatEscape_13]

The only comparison this style of mathhammer gives is optimal killing potential, if everything is going your way- your opponent isn't using cover, your units are at full strength, the targets are at your optimal ranges, and you get the charge off.

 

I immediately thought "Well, instead we should calculate effectiveness at minimal killing potential rather than optimal"... then realized that every army would score 0's (opponent in cover, unit at minimum strength, target out of range for shooting and charging). Took me a second to recognize the folly, heh.

 

Baseline, as you suggest Wycked, is probably the right way to think. These optimal numbers trend down the more complications that are added in. And we might see other builds/combos more effective under certain conditions--which is the huge difficulty of the raw scores that started this debate.

 

Anyhow, interesting to consider at least! Bring forth the mathhammer (in small, digestable doses)!

 

Cheers.

[Ming]

The original blog author came up with these metrics to evaluate in a general sense one list from another, in part choosing these 4 to keep it simple, and as an alternative to just counting numbers of vehicles or HWs, or CCWs, etc. He did find a qualitative relationship in a tournament between top lists and bottom lists.

 

The overall thesis is that if you have a base list and are looking to improve it, strive for changes that increase your metrics. Further, he suggested comparing a individual units metrics vs its point cost as a tool for deciding where you might get the most bang for the buck. It is a nice management tool and a relatively good way to compare one unit or one army list to another, or quantify your potential weaknesses to see where you might look for improvements.

 

I went so far myself as to create a spreadsheet where common unit model types are on separate spreadsheet lines, to facilitate list changes and having a means to model list change improvements. Have not had the time to rigorously apply it however.

 

One could, if so inclined, model improvements not to the SM equivalent, but to your likely opponents, just to aid as a tool in further list building / evaluating, for example dead raiders per turn (AV10) or dead tyranid gaunts per turn, or FNP BAs.

 

Tools are tools.

[ilnar]

A 10 man tac squad with PG and LC CAN fire both, combat squads my friend.

 

Missing the point a little, but yes

[Laughingdagger]

Rereading the instructions on the BolS article gave me some insight as to what was wrong with my calculations, and I redid them and came up with this.

 

DMS: 29.29

DMCC: 23.93

DRPG: 107.2

DLRPG: 57.87

 

This set makes more sense, and was simpler to calculate the way the author intended.

 

Edit: Whoops, that was for my 1850 list.

[Something Wycked]

Very respectable numbers for only 1850, and the original numbers are on 2000 point armies. What sort of list are you playing?

[Laughingdagger]

C:SM with Vulkan, a libby, 3 tac squads in rhinos with melta, combi-melta and missile launcher, 3 dreads in pods with heavy flamer, a 5 man hammernator squad, and 2 vindicators with siege shields. As yet undecided if my tac squads need some CC deterrent or meltabombs since I've only faced eldar, guard, and space wolf opponents so far. Wrecked by Logan Beefnar and all SW's having amazing CC, don't even need CC for eldar and guard, so I'm unsure. Where I'd spare points from to fund it is also a thought.

 

 

Edit: How did I miss that I have 4 elite choices? God I feel dumb. Anyway back to the drawing board.