Greedy Goblin

Thursday, June 4, 2015

MoA pilot analysis, 2015 Feb and March

I wrote how the desperate act of the Evil Imperium to demoralize MoA was nothing but little annoyance to those who make statistics. Well, I make statistics and I was annoyed. If I exclude it by being obvious troll, I could exclude a bunch of CFC kills too. And some idiot MoA losing freighters. And... it's a slippery slope. However I can't just leave it there either. But I've found the solution. Instead of looking for the alliance, I look for its pilots. This way the Revenant ruins the killboard of only one "MoA pilot", the rest are unaffected. To test this, I ran the new statistics on the old February+March data. April+May will come soon.

This is the performance of the MoA pilots against the Evil Imperium. Kills and losses are de-inflated, so if a pilot got 10% damage on a 300M AFK Ishtar, he gets 30M booked. Only damage done to and losses from the Evil Imperium are listed. The table below lists the top pilots of MoA, sorted by damage done.

# Pilot Name M to Imp M by Imp Ratio Top kills Top losses
1 Erebus SilentKill 10993 2049 84% 2485M, 1531M, 734M, 285M, 271M, 238M, 221M, 218M, 100M, 94M,
2 rsantos 7903 1478 84% 761M, 309M, 277M, 255M, 253M, 169M, 153M, 97M, 97M, 93M,
3 Guillejejeje XDD 7132 1283 85% 621M, 398M, 382M, 361M, 341M, 95M, 93M, 93M, 88M, 86M,
4 Samael Curtis 6603 1658 80% 701M, 440M, 261M, 233M, 233M, 84M, 81M, 79M, 78M, 76M,
5 gib5on 6510 1820 78% 1425M, 606M, 426M, 184M, 178M, 197M, 195M, 189M, 170M, 169M,
6 Maurice Bourdon 6366 725 90% 2549M, 725M, 368M, 190M, 134M, 180M, 163M, 85M, 51M, 38M,
7 TIRANOSSAUROS 6261 849 88% 620M, 455M, 281M, 281M, 264M, 215M, 172M, 114M, 105M, 99M,
8 Atreon Sek'Hyren 5595 1067 84% 969M, 488M, 390M, 346M, 226M, 311M, 223M, 196M, 94M, 75M,
9 Eridon Hermetz 5512 2984 65% 823M, 547M, 358M, 315M, 300M, 2602M, 207M, 132M, 43M,
10 Hemmo Paskiainen 5343 0 100% 637M, 432M, 361M, 325M, 260M,
11 DuCkie101 5230 1085 83% 1046M, 389M, 287M, 241M, 190M, 383M, 297M, 104M, 97M, 85M,
12 Kauker 5216 1419 79% 570M, 490M, 359M, 277M, 218M, 457M, 376M, 263M, 196M, 120M,
13 Kinis Deren 4760 557 90% 642M, 200M, 171M, 158M, 138M, 183M, 163M, 37M, 34M, 33M,
14 Neok1337 4712 1497 76% 298M, 283M, 159M, 152M, 146M, 840M, 295M, 122M, 99M, 42M,
15 PiXEL UA 4154 6073 41% 558M, 377M, 359M, 235M, 232M, 2095M, 1164M, 809M, 641M, 635M,
16 Vizant 4131 1459 74% 461M, 455M, 252M, 163M, 158M, 267M, 264M, 152M, 96M, 95M,
17 Ransu Asanari 4100 356 92% 918M, 345M, 338M, 190M, 124M, 124M, 68M, 40M, 38M, 20M,
18 Gen Eve 4076 301 93% 489M, 217M, 214M, 192M, 126M, 34M, 34M, 33M, 33M, 33M,
19 Elgwapo 4044 681 86% 286M, 277M, 244M, 222M, 159M, 120M, 116M, 103M, 101M, 76M,
20 Death Trapp 4030 228 95% 847M, 615M, 237M, 150M, 130M, 48M, 35M, 30M, 27M, 22M,
21 xXCojakXx 3933 575 87% 508M, 166M, 135M, 118M, 83M, 153M, 84M, 72M, 57M, 39M,
22 Tocan Netrow 3877 1840 68% 1045M, 377M, 333M, 277M, 213M, 828M, 380M, 261M, 227M, 75M,
23 Dave911 3866 256 94% 601M, 231M, 166M, 143M, 135M, 44M, 41M, 39M, 35M, 30M,
24 Zefirnii Koks 3806 1654 70% 320M, 242M, 217M, 190M, 175M, 268M, 229M, 158M, 145M, 93M,
25 MASSADEATH 3737 1526 71% 310M, 270M, 142M, 122M, 105M, 963M, 110M, 75M, 63M, 54M,
26 Jizzmaster Mckenzie 3674 4799 43% 1156M, 354M, 169M, 116M, 100M, 806M, 787M, 777M, 554M, 290M,
27 h4x0r84 3565 1061 77% 253M, 245M, 99M, 95M, 84M, 203M, 96M, 63M, 63M, 63M,
28 Sir Rexor 3503 613 85% 315M, 235M, 162M, 93M, 85M, 89M, 77M, 77M, 76M, 66M,
29 bonkerss 3367 462 88% 519M, 412M, 108M, 93M, 82M, 219M, 45M, 34M, 33M, 33M,
30 Furiae Kaisang 3338 1028 76% 277M, 226M, 189M, 176M, 173M, 330M, 79M, 78M, 74M, 67M,
31 Hugo Stiglitz01 3301 557 86% 259M, 258M, 160M, 118M, 95M, 35M, 34M, 32M, 32M, 30M,
32 Lupus Aurelius 3138 31 99% 606M, 279M, 197M, 171M, 167M, 31M,
33 Johnny Tengu 3115 1376 69% 526M, 447M, 373M, 342M, 208M, 356M, 217M, 208M, 191M, 95M,
34 fihs shark 3061 628 83% 588M, 553M, 249M, 115M, 106M, 137M, 84M, 83M, 40M, 36M,
35 Komiser Kemal 2979 436 87% 237M, 163M, 162M, 122M, 112M, 336M, 39M, 31M, 30M,
36 Renwall Stoneheart 2875 285 91% 182M, 143M, 142M, 113M, 109M, 247M, 37M,
37 Torch Toomb 2749 335 89% 927M, 183M, 154M, 120M, 113M, 54M, 53M, 49M, 43M, 35M,
38 Rogueweapon 2717 246 92% 167M, 166M, 114M, 109M, 97M, 59M, 57M, 33M, 31M, 28M,
39 VTyx Soul 2629 242 92% 152M, 149M, 132M, 88M, 81M, 189M, 41M, 13M,
40 antip Soikutsu 2615 1003 72% 351M, 254M, 237M, 150M, 111M, 236M, 210M, 203M, 73M, 71M,
41 Reginald Shadowsmurf 2527 62 98% 575M, 510M, 326M, 258M, 197M, 33M, 29M,
42 Elana Su 2508 1006 71% 227M, 160M, 88M, 69M, 69M, 161M, 91M, 88M, 85M, 83M,
43 Fragwit 2291 195 92% 475M, 185M, 177M, 171M, 146M, 59M, 59M, 46M, 31M,
44 Delnar Katuivo 2282 264 90% 242M, 234M, 121M, 118M, 115M, 101M, 49M, 47M, 35M, 33M,
45 Corey Edward 2278 952 71% 194M, 113M, 108M, 108M, 107M, 232M, 190M, 108M, 107M, 92M,
46 Allucination 2232 157 93% 244M, 224M, 204M, 156M, 146M, 60M, 38M, 32M, 27M,
47 El comepiedras Jastaspista 2210 645 77% 352M, 280M, 215M, 147M, 84M, 137M, 135M, 128M, 91M, 80M,
48 Jaedes 2160 777 74% 249M, 113M, 92M, 85M, 84M, 233M, 88M, 74M, 69M, 65M,
49 Percival79 2108 364 85% 285M, 201M, 145M, 133M, 114M, 65M, 61M, 58M, 41M, 40M,
50 Balalaika 3struna 2073 570 78% 640M, 194M, 194M, 103M, 95M, 179M, 151M, 122M, 92M, 24M,
51 Sergei Veydin 2042 531 79% 254M, 243M, 214M, 106M, 95M, 68M, 66M, 35M, 33M, 33M,
52 Kellen Galvin 1988 72 96% 175M, 113M, 90M, 84M, 79M, 22M, 21M, 21M,
53 Dieter Rams 1965 2906 40% 326M, 128M, 115M, 112M, 94M, 1678M, 821M, 165M, 44M, 35M,
54 159Pinky 1948 1 100% 378M, 245M, 147M, 124M, 93M,
55 Safour 1919 861 69% 288M, 184M, 118M, 110M, 100M, 244M, 217M, 45M, 39M, 39M,
56 Nevil Kincade 1881 85 96% 443M, 239M, 183M, 151M, 129M, 47M, 32M,
57 FieryCorn 1846 562 77% 155M, 93M, 69M, 64M, 54M, 79M, 37M, 35M, 35M, 32M,
58 stunkerd II 1830 789 70% 533M, 273M, 150M, 128M, 124M, 251M, 176M, 104M, 67M, 66M,
59 Cognac 1801 540 77% 463M, 135M, 110M, 76M, 67M, 230M, 219M, 48M, 33M, 11M,
60 Leaf Blossom 1782 72 96% 547M, 87M, 81M, 75M, 72M, 45M, 12M, 10M,
61 Kayi Brixius 1737 432 80% 236M, 205M, 195M, 124M, 70M, 217M, 39M, 39M, 39M, 29M,
62 xBRANEx 1701 634 73% 171M, 121M, 92M, 89M, 82M, 219M, 186M, 128M, 42M, 38M,
63 Nicholass Charante 1700 432 80% 426M, 312M, 223M, 189M, 111M, 114M, 63M, 59M, 55M, 52M,
64 Paul Kevin-Short 1668 961 63% 108M, 63M, 59M, 58M, 47M, 220M, 213M, 213M, 78M, 38M,
65 46khz Walang 1647 586 74% 287M, 138M, 136M, 115M, 88M, 219M, 78M, 43M, 39M, 38M,
66 Seiver Gennarr 1632 0 100% 318M, 239M, 237M, 230M, 177M,
67 forgottenone Udan 1535 678 69% 214M, 95M, 73M, 72M, 66M, 202M, 87M, 37M, 36M, 33M,
68 Jokelerie 1518 170 90% 172M, 106M, 91M, 85M, 80M, 65M, 39M, 38M, 18M, 10M,
69 ooma 1485 137 92% 277M, 199M, 138M, 130M, 96M, 137M,
70 Gaizka Alud 1466 372 80% 128M, 111M, 95M, 81M, 80M, 186M, 75M, 39M, 34M, 20M,
71 Dhamon Audanie 1421 338 81% 123M, 95M, 73M, 65M, 49M, 47M, 45M, 45M, 35M, 32M,
72 JAY WRIGHT 1417 142 91% 67M, 60M, 59M, 55M, 53M, 38M, 31M, 31M, 18M, 13M,
73 Undertovv 1414 0 100% 322M, 258M, 114M, 104M, 96M,
74 dMbN3wB 1386 301 82% 221M, 142M, 116M, 57M, 50M, 37M, 33M, 32M, 31M, 30M,
75 Major Snafu 1350 483 74% 347M, 142M, 137M, 116M, 102M, 229M, 39M, 38M, 36M, 36M,
76 leothegraywolf 1295 346 79% 185M, 176M, 130M, 93M, 83M, 90M, 73M, 36M, 32M, 31M,
77 DanDare420 1295 76 94% 185M, 167M, 85M, 73M, 59M, 29M, 26M, 11M, 10M,
78 Luhter Thurston 1278 1248 51% 159M, 90M, 56M, 54M, 50M, 213M, 197M, 101M, 93M, 77M,
79 Gumby Taron 1250 545 70% 159M, 115M, 107M, 75M, 70M, 343M, 31M, 30M, 29M, 28M,
80 Nodire Hermetz 1226 135 90% 150M, 136M, 125M, 109M, 93M, 40M, 37M, 34M, 24M,
81 Andrus Harnold 1215 384 76% 380M, 343M, 79M, 36M, 35M, 69M, 27M, 17M, 17M, 17M,
82 August Hayek 1213 52 96% 161M, 155M, 148M, 113M, 64M, 39M, 12M,
83 Akballah Kassan 1205 265 82% 107M, 67M, 64M, 62M, 50M, 45M, 35M, 31M, 29M, 27M,
84 Whisperen 1195 133 90% 215M, 166M, 107M, 85M, 74M, 73M, 61M,
85 Angloital 1167 88 93% 197M, 76M, 66M, 64M, 50M, 36M, 36M, 15M,
86 Athalwolf 1149 306 79% 81M, 81M, 60M, 58M, 56M, 131M, 45M, 38M, 38M, 29M,
87 Castiel Visage 1138 544 68% 286M, 238M, 179M, 119M, 110M, 468M, 27M, 20M, 18M, 11M,
88 Runcible Spo0n 1126 228 83% 256M, 98M, 88M, 85M, 79M, 63M, 35M, 35M, 33M, 31M,
89 Veronica Nalelmir 1103 89 93% 330M, 99M, 75M, 73M, 69M, 45M, 31M,
90 Ixtisy 1083 136 89% 260M, 111M, 89M, 67M, 48M, 47M, 39M, 38M, 12M,
91 Trounced 1067 86 93% 287M, 211M, 156M, 104M, 38M, 57M, 30M,
92 Jon Eriker 1063 465 70% 422M, 244M, 63M, 38M, 30M, 212M, 193M, 35M, 14M, 11M,
93 Satek Oreye 1063 260 80% 667M, 47M, 46M, 45M, 29M, 104M, 52M, 33M, 28M, 17M,
94 Eos Canon 1062 90 92% 230M, 184M, 81M, 80M, 71M, 47M, 17M, 17M,
95 Ed Hubble 1054 136 89% 84M, 64M, 64M, 58M, 55M, 42M, 40M, 34M, 20M,
96 Madam Reason 1041 66 94% 651M, 63M, 51M, 49M, 47M, 34M, 33M,
97 The BIind 1039 0 100% 634M, 217M, 65M, 64M, 31M,
98 Daegara Odenson 1027 997 51% 150M, 111M, 101M, 54M, 53M, 379M, 208M, 190M, 91M, 91M,
99 lockghar 1019 255 80% 113M, 89M, 73M, 71M, 68M, 183M, 42M, 30M,
100 Rastlor 1005 1362 42% 894M, 65M, 41M, 753M, 525M, 38M, 38M,
101 Everyone else 81203 59360 58%
102 Total 352843 131250 73%

It's hard to be not amazed. Out of 100 pilots only 4 had bad and 2 neutral killboard ratios. The majority of the MoA pilots have very green killboard. Having 90%+ de-inflated killboard ratio in nullsec is probably the hardest thing to achieve and dozens of MoA pilots did that! This is the first time I use the term "Elite PvP", without irony. As a bonus, even the "everyone else" group is in the green, so I didn't cherry-pick a few good pilots from a sea of bad, but a few excellent ones from the sea of goods.

This statistics is also a great example for what I wrote about members and power. MoA has 1282 members, but the top 100 (8% of the members) caused 77% of the damage to the minions of Evil. So MoA can "failcascade", losing hundreds of missioners without losing any combat performance.

PS: totally unrelated giga-moron. Why would anyone fit this thing?


Arrendis said...

Because it's an incursion ship, and that's apparently how that incursion group fits their Machariels. He can obviously afford it - and probably was able to afford to replace the parts that didn't drop.

Do I think it's stupidly excessive? Yup. But then, I'm not above splitting the take from an incursion site with the max number of people involved. He and his group are maximizing their individual isk/hr, like good little Randians.

Anonymous said...

You took the top pilots, of course the ratios will be higher! That doesn't mean that the typical behaviour. Worse still the top pilots aren't even the best because they tend to be pilots who have just got lucky high value kills. Take the first player, he's got 4b from 2 pods he killed, likely in fights where other people were scoring real kills while he was friglolling. He just happens to target faster than them so gets fist shot at the pod.

I also have to question your methods. The second player shows a 761m kill, yet 30% of that kill was a NPC. Why is that not dewhored, since that's also inflating their kills values by damage they did not do. For example, if you show up as a billion isk battleship is in hull and you one-shot the rest of it, is that really deserving of the full kill value as points?

Orson Brawl said...

That's still over 300 hours of incursion running to pay for that fit assuming a generous 150m/hr in income. He'd be better off going for less bling and investing the difference. Assuming his goal is making ISK and not making the most expensive incursion boat he can get.

Gevlon said...

NPC contribution is removed from player kills, since it's reasonable to assume that the ratter wouldn't die to the rat without the player arriving, so the player is 100% responsible for the kill.

Anonymous said...

"NPC contribution is removed from player kills, since it's reasonable to assume that the ratter wouldn't die to the rat without the player arriving, so the player is 100% responsible for the kill."
In the same way though a ship is unlikely to die without a tackler stopping him getting away, so why do they not get 100% of the kill? Seems to me that giving them full credit when they are just finishing off an already damaged ship, often tackled by the NPCs too which stops them running in the first place is just a way to mean that killing a ratter makes someone look like a better PvPer than someone who kills an actual PvPer.

Gevlon said...

If multiple players are on the kill, they all get credited. The kill must be distributed among the participants to avoid inflation, there is no "fair" method to do it, all distributions are arbitrary.

There is a simple reason to treat NPCs differently than players: ratters massacre NPCs in insane number. While ratting losses happen, in 99.9% of such "battles", the player wins. So when the ratter dies, it's reasonable to attribute this to the attacking PvP-er, and not the rats, even when their contribution is obvious (the ratter wouldn't even been there if the rats don't show up). Also, the rats often have extreme damage done, since they were shooting an active tanked ratter for hours before the PvP-er shown up.

Anonymous said...

small gang podpolice in larger fights (those in <30% TiDi that light up in the hourly metric of the map statistics). try to kill all the pods loot carrier wrecks for the chance of officer/deadspace modules. With some luck you have some shiny kills and some billions worth in you cargo.
I can't say that it is elite pvp. I did it because I didn't know that spheres and scrams are different effects .. so I thought my interceptor was fully warpscramble immune, otherwise I wouldn't have dared to go to a large fight. I got some good loot out of it and have enough ISK for my humble EVE needs.

anon357 said...

> MoA has 1282 members, but the top 100 (8% of the members) caused 77% of the damage to the minions of Evil. So MoA can "failcascade", losing hundreds of missioners without losing any combat performance.

There's some flaw to this statement, though. The most active members are boosted by having those less active members along. Notice how even among these "top 100" members, many of their top kills are contributions to "large fleet" kills / mostly blops drops. If, say, the pilots in spot 101-200 were to leave the alliance and go highsec missioning, these top 100 pilots would lose a lot of effectiveness by not being able to take on as large targets, as big fleets, etc, simply because they don't have as much backup and as many members available.

@Anon, Gevlon is right in that there is no fair distribution. By some right, all kills of a fleet should be attributed to the FC because they're making the call to be at that place at that time. Another could say, whoever lands initial tackle deserves the entire kill, because the ship would probably have gotten away otherwise. Another approach is to let damage dealt determine kill contribution, as Goblin does. None of these methods is wrong, they all have their merits, and this one has a nice way of at least letting a high ranking correlate at least moderately with activity levels.

There are of course other methods as well - and I'm sure he's open to suggestions if you suggest another distribution method and argue decently for why it's a good metric. But this one fulfills the task set before it acceptably, while also being easy to process and show, and algorithmically simple.

Unknown said...

I'd really exclude Pixel UA from that stats. He pilots expensive and untested fits a lot. To be honest he is one of the few pilots who deserves to be called "elite pvp" and "balls of steel" while others being mostly gatecampers/stealthbombers/frigllolers/f1 monkeys.

Anonymous said...

So clicking through the pilots you had chosens' killboards I realized why they are so effecient. All the top kills for each pilot were all ships that had numerous bombers involved. Of course their killboard will be green, they hot drop countless ratting/hauling victims fitting with expensive hulls/cargo. If they lose something it's not a big loss seeing as that they are in bombers.

My question is: What exactly is so elite pvp about right clicking a blackops - jumping through - then pressing F1 ?

Anonymous said...

I don't get it. Why not just have NPC damage the equivalent of "nobody". So if Player A did 50%, player B did 20% and an NPC did 30% on a 1b kill they get the following distribution:

A: 500m
B: 200m

As you are doing it currently, a PvPer fighting a PvPer shows as worse at PvP than someone killing a ratter. Surely you must be able to tell that's broken? Technically speaking, pod kill values should actually be added to the associated ship kill where available and the damage for both ship and pod should be used to distribute the total value of both kills too. Otherwise you are giving more credit to players who sit there shooting pods rather than actually fighting. That's why a lot of decent PvP groups ignore pod kills when looking at rewarding members.

"Also, the rats often have extreme damage done, since they were shooting an active tanked ratter for hours before the PvP-er shown up."
That's not how damage works, otherwise rats would be getting 99% of damage in nearly all cases. The damage distribution shows is what damage they did that contributed to the actual whole kill.

Arrendis said...

If multiple players are on the kill, they all get credited. The kill must be distributed among the participants to avoid inflation, there is no "fair" method to do it, all distributions are arbitrary.

Arbitrary is fine, so long as you're consistent. If everyone can be assured the data they're looking at has all had the same processes applied, then they can be sure that that within the data set, all relationships are reasonably accurate. There are a lot of different ways you could parse a kill: highest damage, giving tackle a specific weight, etc. They all have their downsides.

Highest damage inflicted ignores contributions in non-damaging ways: tackle, target painting, possibly logi keeping the dps ships alive, etc.

Weighting for tackle/support only works when the tackle or support appear as such - if a huginn paints and webs a target, he's definitely making it much easier for everyone else to apply damage. But if he also fires some missiles into its hull, he's going to show up as 'shot missiles at it' on the killmail. Any tackle that applies damage will show up as damage application. A sabre who bubbles a target that never even tries to go into warp (because he knows he's bubbled) won't show up, just like Logistics ships tend to never show up at all, despite (especially in small gang work) often being the only reason anything lives long enough to get kills.

Gevlon has his methodology, and questioning the method is perfectly reasonable. Questioning or even outright disputing his conclusions is also totally reasonable. But he's consistent in how he filters the data. The results he presents are reliable within the context of their relationship with one another. Will there be edge cases and outliers? Of course. There always are. In the bigger picture, they amount to statistical noise, nothing more.

Anonymous said...

Here here! The guy that killed the 700 mil Isthar!

A ratter sits in a plex tanking it soaking 90K damage doing just fine, I show up and do 10K damage to kill him. why should I be rewarded 10%.

Rats switch aggro to me most of the times anyway!

Warden of the North said...

FYI Burn Amarr was recently announced and scheduled.