A call for AVRE data

[Guest]

Welcome, Guest. Please login or register.
Did you miss your activation email?

1 Hour 1 Day 1 Week 1 Month Forever

Login with username, password and session length

[Mysthalin]

Because I will have a modicum of free time to spare after exams are over and I will not be doing my favorite subject (Econometrics) any more during next semester I've decided to go ahead and call for the EiRR community to give me something to do, and at the same time potentially put an end to the discussion of whether or not the AVRE is worth it through a quantitative field-test.

What do I mean by that?

Simple. I will look at replays that you, the community, will provide to me via attachments and take note of the kills your AVREs will garner through their time on the battlefield - and then translate these kills into resources lost by the enemy.

After I have collected this data, I will simply run a simple statistical analysis of what the average factor of destruction by the AVRE is and how near or far this number is from the cost of the AVRE itself. If I'm nice enough, I'll slap on some econometric analysis of it as well (such as the factors which define an AVRE's usefulness in a game).

What do I want?

I want you to submit replays of you lot using AVREs. I don't care if the AVREs are 0-kill fuck-ups, or if they are 50 kill gods. I don't care if you've won the games handily or if you've lost miserably. I do not care whether the AVRE was a realistic contributor to the game. I do not care if the game is entertaining - if you have a replay of you using the AVRE, I want you to submit it.

The more of you submit your replays, and the more replays I get - the more accurate my results will be. Use them in any company build with any doctrinal abilities you may.

Have at it folks, and please submit your replays to the AVRE Data thread:
http://forums.europeinruins.com/index.php?topic=25901.msg453432;topicseen#new

[TheWindCriesMary]

Thanks for doing this Mysthalin, though I doubt all of the guys in the other thread swearing up and down that the AVRE is fine will be submitting any replays.

On another note, will you also be evaluating opportunity cost?

[Mysthalin]

What do you mean by me evaluating opportunity cost, specifically?

If you mean what could have been had if another unit that costs as much or nearly as much as the AVRE could have done with the same resources - no. I am operating under the assumption that, on average, the other units in EiRR are balanced according to price. Hence, the resource cost of the unit will by definition be its opportunity cost.

Violating the above assumption would require me to engage in analyzing a ludicrous amount of data (watching replays of other units being used). Then I'd also have the problem of having to slap on arbitrary values of OP/UP onto units, etc.

Something similar to that can be done far more efficiently in a completely different study I have in mind that deals entirely with stats and requires no replays - but I can't be arsed doing that one.

[TheWindCriesMary]

You can actually evaluate the former kind of opportunity cost as well by comparing the AVREs to the staghounds who are also in all of the replays you've got. You don't even need to compare them  1:2. Try it 1:1 (and imperfect comparison I know, but valuable)-- I think it'd be really interesting to see how the considerably cheaper (and non doctrine unlock) unit compares.

-Wind

[Mysthalin]

You do realize that this would increase the effective workload by a factor of 3-4 in just taking note of what the staghounds are killing. Is a lot simpler to do with the AVRE which fires only ever so often.

Then I'd also have to keep tabs on who is using the staghound to make sure the analysis holds any real statistical value. If it's two different players using the two different units and I don't account for this, I introduce both inconsistency and bias into the data through several ommited variables (cross-section differences in doctrine choice and skill level of user) which are ridiculously stochastic and arbitrary if attempted to define. Plus, not everyone who will be submitting the replays will use staghounds.

These issues would not really factor in if I had a large enough sample - things would be "correct on average" through the law of large numbers - but I'm not in any way shape or form going to document the data of a thousand+ replays.

[TheWindCriesMary]

That's why I just asked if you would be evaluating opportunity cost (eg. not having a staghound is the opportunity cost of buying an avre) instead of saying "Mysthalin evaluate opportunity cost in regards to staghounds": it would be a lot more work.

By no means would I hold it against you to not want to add on this extra layer of work. It would be interesting though if you did have time for it.

[tank130]

[Mysthalin]

Look, if this goes well and I'll still have time to spare afterwards I might do a generalistic regression analysis of how a unit's price is defined by 5 variables:

1. Effective survivability
2. Effective firepower (might be expanded into two - brute firepower and penetration)
3. Effective mobility
4. Dummy variable defining cloak
5. Sight range

It won't be definitive proof of what's OP and what's not, but it will be fun and should point out the major outliers of which units are ridiculous and which aren't (or show that pricing is done essentially arbitrarily).

It'd be way far away though, as it'd be the most boring number crunching ever.

[TheWindCriesMary]

That would be really interesting for sure. Again though I definitely can appreciate how much work it is so, from the way I see it, it's a "awesome if you do it", "wouldn't blame you for a second if you don't" type thing.

I'm really interested to see how the AVRE analysis goes though for sure.

[Mysthalin]

One good thing about this other potential study is that it would require 0 looking at replays, and I could re-run the analysis to associate the popcap-to-effectiveness formula as well in basically a heartbeat.

A bad thing is that other than me, the late gamesguy and maybe hicks with some guidance nobody would understand the output of the regressions lol so I'd have to spoon-feed you all the most interesting findings.

[Tymathee]

[tank130]

AVRE usage in action
3v3 on D-Day – Reinforcements

Allies
Bolt
Wind
Andy

Axis
Ray
Tank
Shoganot

4:00 – First AVRE deployed by Wind – Code name W1
5:10 - W1 destroys 2 man Tank Buster squad in house
7:08 – attacked by Marder & Pak, 2 hits each = ~15% health
          -Destroys another 2 man PG squad in house
7:42 – W1 destroyed by precision strike
7:57 – Second AVRE deployed by Wind – Code name W2
8:35 – First AVRE deployed by Bolt – Code name B1
9:07 – W2 attacked by Marder & Pak. 2 Shots each
9:17 – W2 destroys Vet 1, full health pak crew
9:20 – W2 down to 50% health, leaves to repair
9:21 -  B1 fires at 2 moving PG squads. One squad destroyed, the other down to 5% Health
           - Attacked by Marder – 1 shot
12:13 – B1 hit by Pak
12:16 – B1 destroys Pak gun & crew. B1 = ~30% Health
13:44 – W2 attacked by Marder, 2 shots = ~ 60% Health
15:03 – W2 fires at house containing 3 man, vet2 Tank Buster squad. Squad gets out, building destroyed
15:15 – B1 Hulls down & Repairs
15:27 – W2 attacked by 3 man, vet2 and 2 man Tank Buster squad = ~50% Health
16:38 - W2 attacked by 3 man, vet2 and 2 man Tank Buster squad = ~10% Health
16:50 – W2 runs into middle of city, no support, into Marder = W2 Destroyed
18:04 – B1 attacked by 3 man Tank buster squad
18:08 – B1 destroys moving vet2-3 man, and 2 man Tank buster squads
18:13 – B1 Attacked by Marder. Two hits = ~ 50% Health
20:16 – B1 Attacked by Marder again
20:21 – B1 fires at moving 3 man tank buster squad. Kills 1 man but destroys Marder
21:30 – B1 Hulls down to repair (2nd repair)
24:40 – B1 attacked by Pak (2 hits) and Jadg (1 hit)
24:48 – B1 Decrews Vet 2 Pak
24:49 – B1 hit by Goliath & Jadg (1 hit)
24:57 – B1 hit by Jadg (1 hit) and Panzer faust
25:03 – B1 destroyed by Jadg
27:12 – Second AVRE deployed by Bolt – Code Name B2
27:19 – Third AVRE deployed by Wind – Code name W3
28:34 – B2 hit by Pak (2 hits)
28:39 – B2 Decrews Pak
           -B2 attacked by Jadg (1 hit) = ~ 60% Health
28:54 – W3 fires at moving, vanilla PG’s. = Misses
30:41 – W3 fires at moving Tank Busters, misses but knocks squad down to 10%
31:15 – B2 destroys 5 man volks squad in church
31:16 – B2 destroyed by jadg
32:38 – W3 destroys stolen 17 pdr – crew & gun
34:16 – W3, Middle of city, unsupported – Destroyed by Panther & Jadg



Analysis: You are doing it wrong.

Replay attached

[Smokaz]

?


Tankedit: funny as hell..... but ya, unrelated pics removed.

[Mysthalin]

Well done tank, though I'll also be counting in damage to tanks (rough approximation of their percentage health loss translated into their resource cost) and taking into account units that retreated due to the AVRE's shots.

[EIRRMod]

Well done tank, though I'll also be counting in damage to tanks (rough approximation of their percentage health loss translated into their resource cost) and taking into account units that retreated due to the AVRE's shots.

Oh so sort of an aggregate damage (in resources) they caused.

Interesting
Cant wait to see the details!