Monday, 28 January 2013

Patch 8.3 / January tank winrates and OP Rating

I ran a report on my local, large NoobMeter.com database looking at what results people have had in 8.3 and found it quite interesting.

How is this report different from the many others on other sites? Well, I also included Player winrate (WR) for players which own a particular tank in the data, and also calculate a rating called "OP rating" which compares Tank WR with Player WR.

As you can see from the results, the new tanks are generally first played by the most skilled players. If one blindly evaluates new tanks using their WR, that can lead to incorrect conclusions. For example, just because tank X has 52% WR while other tanks in that category have 50% doesn't mean it is OP - perhaps it is simply played by players who on average have 54% WR, and is thus in fact under-performing!

This is why this report provides a more accurate picture.

The report was run on the EU server data between 17th of January to 27th of January, but the conclusions should carry over to other servers.

 Explanation of the columns:
  • Tank WR - victories / battles in percent for this particular tank 
  • Player WR - victories / battles in percent for players playing this tank 
  • OP Rating - Tank WR / Player WR - the higher the more OP

Note that I don't really have all the battles/players represented (I'm missing some low-battle players) - this is why the Tank WR for low-tier tanks are larger than Player WR for these tanks. The sample set isn't 100% representative of the server population, it has the NoobMeter.com users slightly over-represented, and they are generally above average players. I plan to get a more representative set for the next report.

The t10 tanks all have poor OP ratings, I think that is because they are owned by the best players (good Player WR ratings), who can crush the low-tier random games against less experienced players, but then when playing t10 tanks they meet other good players - whether in CW, TC or even randoms, and thus no longer win that much.



Level 10 MT

#TankNationBattlesTank WRPlayer WROP rating
1Bat Chatillon 25 tFR54,35651.6%54.5%0.95
2T62ARU28,52551.0%54.9%0.93
3E-50 Ausf. MDE20,21850.3%54.6%0.92
4M48A1US10,38151.0%55.5%0.92
5121CH96848.5%55.9%0.87
6FV4202UK1,90047.3%55.1%0.86
The poor showing for 121 and FV4202 could be due to low sample sizes, but it seems that M48A1 has been nerfed too much. It can also be seen that it is no longer played much by the players.

Level 10 HT

#TankNationBattlesTank WRPlayer WROP rating
1T57 HeavyUS9,31353.6%55.4%0.97
2T110E5US41,22350.7%54.0%0.94
3E-100DE50,86549.0%52.7%0.93
4MausDE12,43649.1%53.2%0.92
5IS-7RU71,55548.7%52.8%0.92
6IS-4RU23,58349.1%53.5%0.92
7AMX 50BFR10,79750.6%55.2%0.92
8113CH55548.8%53.7%0.91
9FV215bUK3,27749.3%54.8%0.90
The T57 Heavy seems to overperform, the T110E5 is still doing good. 113 and FV205b have small sample sizes.

Level 10 TD

#TankNationBattlesTank WRPlayer WROP rating
1AMX 50 Foch 155FR9,57353.7%55.3%0.97
2T110E3US7,14251.5%53.9%0.96
3Object 268RU15,50451.9%55.0%0.94
4T110E4US18,03749.7%54.2%0.92
5JagdPz E-100DE21,20148.6%53.1%0.91
6Object 263RU1,50350.5%55.4%0.91
AMX 50 Foch 155 is ridiculously OP, the JagdPz E-100 clearly struggles, while the o263 has a small sample size.

Level 9 MT

#TankNationBattlesTank WRPlayer WROP rating
1WZ-120CH1,42154.3%52.4%1.03
2T54E1US17,93253.6%54.2%0.99
3T-54RU61,67950.7%53.1%0.95
4Centurion Mk.7UK8,52152.0%54.6%0.95
5E-50DE44,33950.0%52.9%0.95
6M46 PattonUS22,49650.4%53.4%0.94
7Lorraine 40 tFR36,79747.8%52.1%0.92
The T54E1 seems quite OP. T-54 is doing, possibly due to the best players playing it with HEAT where it is very much OP. The Patton and especially Lorraine 40t can be seen as nerfed too much. 


Level 9 HT

#TankNationBattlesTank WRPlayer WROP rating
1ST-IRU27,96649.8%51.7%0.96
2Conqueror Mk.1UK26,17251.2%53.4%0.96
3E-75DE116,94349.6%51.8%0.96
4AMX 50 120FR24,30651.0%53.6%0.95
5M103US71,29349.5%52.1%0.95
6IS-8RU129,09548.5%51.4%0.94
7VK 4502 (P) Ausf. BDE27,62648.3%52.3%0.92
8WZ-111CH41646.4%52.3%0.89
VK4502B is underperforming as it has for ages, but no buff is coming. WZ-111 has a small sample size. 


Level 9 TD

#TankNationBattlesTank WRPlayer WROP rating
1AMX 50 FochFR12,78553.2%54.4%0.98
2Object 704RU36,38151.6%53.1%0.97
3T95US17,51651.1%52.8%0.97
4JagdtigerDE51,80749.9%52.5%0.95
5T30US36,42449.4%52.6%0.94
6SU-122-54RU6,20150.7%55.1%0.92
Another Foch who is overperforming. Strangely, the SU-122-54 is not doing very well.


Level 8 LT

#TankNationBattlesTank WRPlayer WROP rating
1WZ-132CH1,44957.0%56.4%1.01
2AMX 13 90FR66,29652.2%53.2%0.98

Level 8 MT

#TankNationBattlesTank WRPlayer WROP rating
1T69US29,65452.6%54.2%0.97
2CenturionUK24,07051.9%53.6%0.97
3T26E4 SuperPershingUS129,62449.4%51.1%0.97
4T-44RU61,53550.2%52.1%0.96
5Type 59CH170,33650.9%52.8%0.96
6Panther IIDE41,10149.4%51.7%0.95
7M26 PershingUS29,04549.3%52.0%0.95
8T-34-2CH1,65148.3%53.5%0.90
T69 and Centurion both performing well. The Pershing is strangely now the least successful of the "old" mediums.  In the premium category, T26E4 Superpershing overperforms the Type59. 


Level 8 HT

#TankNationBattlesTank WRPlayer WROP rating
1110CH3,44156.7%53.7%1.06
2IS-6RU43,64049.7%50.1%0.99
3KV-5RU75,33950.5%51.3%0.98
4T32US75,14751.2%52.4%0.98
5IS-3RU275,64950.1%51.4%0.97
6CaernarvonUK44,70051.5%52.9%0.97
7AMX 50 100FR40,62051.5%53.1%0.97
8T34US254,02450.0%52.0%0.96
9PzKpfw VIB Tiger IIDE166,12249.0%51.0%0.96
10KV-4RU46,49049.3%51.3%0.96
11LöweDE142,57149.3%51.3%0.96
12M6A2E1US92153.5%56.4%0.95
13VK 4502 (P) Ausf. ADE40,78948.6%51.4%0.94
14FCM 50tFR16,65547.8%50.8%0.94
IS-6 and KV-5 performing very well, I think the gold ammo for credits has helped them as they lacked good guns. FCM 50t and VK4502A are both underperforming.


Here we also clearly see that M6A2E1 is owned by the best players - as it was a pre-order special so mostly players from the beta times got it.

Level 8 TD

#TankNationBattlesTank WRPlayer WROP rating
1JagdPanther IIDE28,48450.3%52.1%0.97
2T28 PrototypeUS46,44949.7%51.7%0.96
3FerdinandDE68,12549.6%51.6%0.96
4ISU-152RU53,71150.3%52.4%0.96
58.8 cm Pak 43 JagdTigerDE23,01748.4%50.8%0.95
6T28US16,62249.3%51.8%0.95
7AMX AC Mle. 1948FR19,11750.5%53.5%0.94
8Uralmash-1RU10,06750.3%54.0%0.93
JagdPanther2 performing very well, while the new Soviet TD line is not performing well, as can be seen in other categories. 


Level 8 SPG

#TankNationBattlesTank WRPlayer WROP rating
1T92US12,10350.3%54.5%0.92
2Object 261RU20,82450.4%54.9%0.92
3GW Typ EDE20,02849.5%54.0%0.92
4Bat Chatillon 155FR14,14949.9%54.7%0.91
Very balanced results for the top-tier arties. 


Level 7 LT

#TankNationBattlesTank WRPlayer WROP rating
1WZ-131CH4,58653.0%53.8%0.99
2T71US58,68452.4%53.6%0.98
3AMX 13 75FR47,24849.6%51.1%0.97
4Type-62CH8,97350.6%52.7%0.96

Level 7 MT

#TankNationBattlesTank WRPlayer WROP rating
1T-34-1CH3,43953.9%54.2%1.00
2T-43RU52,09450.1%50.7%0.99
3T20US30,79249.8%51.3%0.97
4Panther M10DE11,94549.2%51.0%0.96
5PzKpfw V PantherDE63,36649.2%51.1%0.96
6KV-13RU15,66748.3%50.5%0.96
7CometUK51,45649.0%51.6%0.95
8VK 3002 (DB)DE18,53448.5%52.3%0.93

Level 7 HT

#TankNationBattlesTank WRPlayer WROP rating
1IS-2CH6,09557.4%54.4%1.06
2T29US142,29851.5%51.2%1.01
3Black PrinceUK61,40952.0%51.8%1.01
4ISRU231,38149.5%50.0%0.99
5PzKpfw VI Tiger (P)DE92,27850.4%50.9%0.99
6KV-3RU54,84349.6%50.5%0.98
7PzKpfw VI TigerDE201,20049.0%50.2%0.98
8AMX M4 (1945)FR28,95949.5%51.1%0.97

Level 7 TD

#TankNationBattlesTank WRPlayer WROP rating
1SU-122-44RU37,46453.0%52.0%1.02
2T25 ATUS22,65750.7%51.1%0.99
3AT 15AUK5,29351.5%52.1%0.99
4SU-100M1RU14,74351.6%52.6%0.98
5SU-152RU52,43549.9%51.1%0.98
6JagdpantherDE111,41149.9%51.2%0.98
7AMX AC Mle.1946FR21,75951.0%52.7%0.97
8T25/2US64,59949.4%51.1%0.97

Level 7 SPG

#TankNationBattlesTank WRPlayer WROP rating
1M40/M43US28,93349.6%52.9%0.94
2Object 212RU39,08549.7%53.0%0.94
3Lorraine155 51FR36,11550.0%53.5%0.93
4GW TigerDE59,50048.5%52.1%0.93

Level 6 LT

#TankNationBattlesTank WRPlayer WROP rating
1AMX 12tFR81,17149.9%50.6%0.99
2T21US51,47250.0%51.4%0.97
359-16CH9,97351.0%52.7%0.97

Level 6 MT

#TankNationBattlesTank WRPlayer WROP rating
1Type 58CH21,51055.2%53.8%1.03
2VK 3601 (H)DE198,11550.9%50.4%1.01
3T-34-85RU114,13450.6%50.5%1.00
4M4A3E8 ShermanUS34,47351.1%51.1%1.00
5CromwellUK152,90251.1%51.3%1.00
6M4A3E2US30,20551.3%51.6%0.99
7Pz IV SchmalturmDE15,69251.6%51.9%0.99
8VK 3001 (P)DE89,51549.1%49.6%0.99
9VK 3001 (H)DE58,90249.5%50.4%0.98
10PzKpfw V-IVDE30949.5%54.2%0.91
11PzKpfw V-IV AlphaDE10.0%52.6%0.00

Level 6 HT

#TankNationBattlesTank WRPlayer WROP rating
1KV-1SRU871,93851.5%50.7%1.02
2T-150RU77,21150.7%50.0%1.02
3M6US92,96050.2%49.9%1.01
4KV-2RU41,88652.5%52.2%1.00
5TOG IIUK5,42051.8%51.6%1.00
6Churchill VIIUK81,97549.7%50.5%0.98
7ARL 44FR56,91449.4%50.8%0.97

Level 6 TD

#TankNationBattlesTank WRPlayer WROP rating
1SU-100RU74,47051.8%51.1%1.01
2M18 HellcatUS165,87052.1%51.6%1.01
3Dicker MaxDE30,03752.0%51.8%1.00
4M36 SluggerUS30,98850.8%50.8%1.00
5JagdPz IVDE98,84149.0%49.5%0.99
6ARL V39FR23,99749.9%51.8%0.96

Level 6 SPG

#TankNationBattlesTank WRPlayer WROP rating
1M12US34,05149.7%51.3%0.97
2Lorraine155 50FR50,76850.7%52.4%0.97
3S-51RU33,98951.0%52.8%0.97
4SU-14RU25,78249.1%51.2%0.96
5GW PantherDE127,05849.4%51.5%0.96

Level 5 LT

#TankNationBattlesTank WRPlayer WROP rating
1ELC AMXFR129,54850.6%50.4%1.01
2CrusaderUK66,04448.9%49.3%0.99
3VK 2801DE31,96352.6%53.6%0.98
4M24 ChaffeeUS15,01952.7%53.9%0.98
5T-50-2RU53,13050.8%52.9%0.96

Level 5 MT

#TankNationBattlesTank WRPlayer WROP rating
1MatildaRU5,36953.7%51.6%1.04
2T-25DE34,42353.2%51.6%1.03
3M4 ShermanUS65,80752.8%51.2%1.03
4Ram-IIUS3,57253.2%52.0%1.02
5PzKpfw IVDE261,31051.3%50.2%1.02
6M4A2E4US1,47254.1%53.1%1.02
7T-34RU66,20249.9%49.4%1.01
8Matilda Black PrinceUK3,17651.6%51.5%1.00
9T-34CH107,76652.6%52.6%1.00
10PzKpfw III/IVDE109,00649.3%49.4%1.00
11M7US62,67149.8%50.2%0.99

Level 5 HT

#TankNationBattlesTank WRPlayer WROP rating
1KV-220 (action)RU887.5%63.2%1.39
2T1 heavyUS78,90251.8%49.7%1.04
3KV-1RU304,56852.0%50.0%1.04
4T14US20,44552.1%50.5%1.03
5ChurchillRU37,33852.6%51.1%1.03
6Churchill IUK86,83250.5%49.8%1.01
7KV-220 Beta-TestRU3262.5%61.8%1.01
8BDR G1BFR46,03150.3%49.8%1.01

Level 5 TD

#TankNationBattlesTank WRPlayer WROP rating
1M10 WolverineUS47,24251.5%50.2%1.02
2S-35 CAFR18,02952.6%51.5%1.02
3StuG IIIDE288,54951.1%50.1%1.02
4T49US83,45951.1%50.3%1.02
5SU-85RU75,49550.1%50.0%1.00

Level 5 SPG

#TankNationBattlesTank WRPlayer WROP rating
1SU-8RU48,70451.0%50.3%1.01
2AMX 13 F3 AMFR49,65950.4%51.1%0.99
3HummelDE145,80749.5%50.3%0.98
4M41US53,38749.6%50.4%0.98

Level 4 LT

#TankNationBattlesTank WRPlayer WROP rating
1ValentineRU1,77253.2%52.0%1.02
2PzKpfw 38 nADE43,36449.8%48.7%1.02
3CovenanterUK45,25450.0%49.2%1.02
4AMX 40FR23,98949.0%48.6%1.01
5M5 StuartUS44,84050.8%50.4%1.01
6M5A1 StuartCH235,34050.9%50.7%1.01
7ValentineUK8,69849.8%49.6%1.00
8T-50RU119,05950.7%50.5%1.00
9A-20RU35,14148.5%48.7%0.99
10VK 1602 LeopardDE36,31149.9%50.3%0.99

Level 4 MT

#TankNationBattlesTank WRPlayer WROP rating
1MatildaUK38,77354.0%50.2%1.08
2T-28RU55,51650.9%48.9%1.04
3PzKpfw IIIDE46,41751.7%49.9%1.04
4A-32RU21354.0%52.8%1.02
5M3 LeeUS39,34249.7%49.0%1.01

Level 4 HT

#TankNationBattlesTank WRPlayer WROP rating
1PzKpfw B2 740 (f)DE2,43557.9%53.7%1.08
2B1FR17,69249.4%49.0%1.01

Level 4 TD

#TankNationBattlesTank WRPlayer WROP rating
1SU-85BRU32,14353.0%50.0%1.06
2T40US25,94553.5%50.5%1.06
3HetzerDE100,75152.5%50.0%1.05
4M8A1US33,03651.3%49.7%1.03
5Somua S-40FR10,77249.7%50.2%0.99

Level 4 SPG

#TankNationBattlesTank WRPlayer WROP rating
1SU-5RU36,12050.9%49.7%1.02
2GrilleDE132,65851.0%50.1%1.02
3105 leFH18B2FR12,97752.3%51.5%1.02
4AMX 105AMFR44,89950.3%50.3%1.00
5M7 PriestUS46,97648.8%49.3%0.99

Level 3 LT

#TankNationBattlesTank WRPlayer WROP rating
1PzKpfw II Ausf. JDE47270.3%55.3%1.27
2T-127RU5,92360.6%53.3%1.14
3PzKpfw 38 (t)DE20,48356.2%50.2%1.12
4T-15DE3,35857.1%52.2%1.09
5PzKpfw II LuchsDE29,52755.1%50.6%1.09
6M22 LocustUS5,42955.8%51.9%1.08
7Cruiser Mk. IVUK13,52453.1%50.0%1.06
8PzKpfw III Ausf. ADE17,68252.4%49.5%1.06
9AMX 38FR6,28951.1%48.6%1.05
10M3 StuartUS12,44750.9%49.4%1.03
11T-46RU21,68950.1%48.7%1.03
12Type 2597 Chi-HaCH171,72251.1%49.8%1.02
13Cruiser Mk. IIUK5,32451.0%50.4%1.01
14BT-7RU18,69048.9%48.8%1.00
15M3 Stuart LLRU9,01249.2%49.5%0.99

Level 3 MT

#TankNationBattlesTank WRPlayer WROP rating
1PzKpfw S35 739 (f)DE9,31763.3%52.7%1.20
2M2 Medium TankUS16,93953.2%49.2%1.08
3D2FR4,26951.2%48.9%1.05
4Vickers Medium Mk. IIIUK6,37151.1%48.9%1.05

Level 3 TD

#TankNationBattlesTank WRPlayer WROP rating
1FCM36 Pak40FR8,86962.5%53.6%1.17
2Marder IIDE23,22254.8%51.3%1.07
3Renault UE 57FR5,86053.4%50.4%1.06
4T82US16,88952.2%50.2%1.04
5SU-76RU18,69650.7%50.0%1.01

Level 3 SPG

#TankNationBattlesTank WRPlayer WROP rating
1SU-26RU42,17655.9%51.9%1.08
2Lorraine39 L AMFR14,58851.6%49.5%1.04
3WespeDE14,07750.6%49.6%1.02
4Sturmpanzer IIDE23,87549.4%48.5%1.02
5M37US18,75749.0%48.6%1.01

Level 2 LT

#TankNationBattlesTank WRPlayer WROP rating
1PzKpfw 38H735 (f)DE10,65659.7%53.3%1.12
2Cruiser Mk. IIIUK25,07356.0%51.3%1.09
3PzKpfw IIDE30,53555.5%50.9%1.09
4T-26RU8,74454.5%50.1%1.09
5PzKpfw 35 (t)DE16,02555.4%50.9%1.09
6M2 Light TankUS11,99154.2%50.3%1.08
7MkVII TetrarchRU9,26957.5%53.8%1.07
8BT-2RU21,31654.1%50.7%1.07
9Cruiser Mk. IUK3,30153.7%50.7%1.06
10Vickers Mk. ECH46,43851.6%49.6%1.04
11Hotchkiss H35DE3,36851.4%49.8%1.03
12D1FR1,07150.9%49.5%1.03
13T2 Light TankUS16,85552.9%52.3%1.01
14T1 E6US69,74150.7%50.7%1.00

Level 2 MT

#TankNationBattlesTank WRPlayer WROP rating
1T2 Medium TankUS6,80853.2%49.5%1.07
2Vickers Medium Mk. IIUK2,93652.3%49.7%1.05

Level 2 TD

#TankNationBattlesTank WRPlayer WROP rating
1Panzerjäger IDE9,87157.1%51.1%1.12
2T18US50,05657.1%51.9%1.10
3AT-1RU5,01153.7%49.9%1.08
4RenaultFT ACFR3,22452.8%50.3%1.05

Level 2 SPG

#TankNationBattlesTank WRPlayer WROP rating
1T57US12,57153.0%49.1%1.08
2Sturmpanzer I BisonDE8,96651.9%49.9%1.04
3SU-18RU4,58249.7%48.9%1.02
4RenaultBSFR3,29549.2%49.1%1.00

Level 1 LT

#TankNationBattlesTank WRPlayer WROP rating
1MS-1RU17,22555.7%50.8%1.10
2LeichttraktorDE50,80255.3%50.5%1.09
3T1 CunninghamUS27,19855.6%51.9%1.07
4Renault NC-31CH54,10552.0%49.9%1.04
5RenaultFTFR6,29450.7%49.6%1.02

Level 1 MT

#TankNationBattlesTank WRPlayer WROP rating
1Medium Mark IUK8,93452.8%50.3%1.05

14 comments:

  1. While this is very interesting, I think it would be even more revealing if you could calculate OP rating on a per-player basis. That is, instead of dividing overall Tank WR by mean Player WR of players owning said tank, divide individual Tank WR by individual Player WR. You could then use a weighted average (based on number of battles in the tank) for the overall OP rating. You could also then stratify by Player WR. It would be interesting if there are certain skill magnifying or compressing tanks that particularly powerful in the hands of the better players while being extremely weak in the hands of worse players (I imagine this is true for many high tier FR tanks, for example).

    ReplyDelete
  2. Alex, thanks for your comment.

    It is a very good suggestion, although I'm not sure it will result in much different results as the Tank WR and Player WR I use is already a weighted average by the battles in the reporting period (that is, I'm not taking the Tank WR and then calculating the mean for that, instead I'm summing all the victories and all the battles for that tank, and only calculating the (victories/battles) as the last step).

    The main difference, if I see it right, is that in your suggested approach Player WR would be weighted by the # of battles players have played in the particular tank being evaluated, as opposed to the total # battles in the reporting period.

    I will see if I can implement it for the next report.

    The point that some tanks are likely disproportionately OP for best players and disproportionately UP for worst players is also very valid, and the suggestion of proving that by stratifying by Player WR is a great idea too.

    ReplyDelete
  3. By the way, assuming I do end up stratifying by Player WR, any suggestions how to easily make this data presentable? As posting 3 different sets of tables for, say, below average, average, and above average players seems quite ineffective (the report as is takes up a lot of space). I guess I could introduce more columns ("OP rating for players w/ WR 52%+")... just thinking maybe there is some more elegant solution.

    ReplyDelete
    Replies
    1. Could you provide the stratified data only when you mouse-over (or click on) one of the tanks? Just to avoid cluttering things up too much.

      Delete
    2. Should be doable.

      Any suggestions on categories for the stratification?

      "Poor player < 46% WR", "Average player 46%..50% WR", "Good player 50%..54% WR", "Excellent player > 54% WR" perhaps?

      Delete
    3. The perfect way to do it would be scatterplots, with player WR on one, and tank WR on the other, but that may be a BIT too much effort :D

      Delete
    4. Scatter plots are no big deal, I already have them in clan pages on www.noobmeter.com so I can implement them fairly easily here. However I'm not certain the will be that easy to read and suitable for understanding which tanks are OP and which ones aren't.

      Delete
  4. Okay, I see what you're saying. I wasn't quite sure how it was calculated from the initial write-up. For the stratification, I think it depends. Personally, I think it would be most interesting for Tier 8+, which definitely limits the number of tanks one needs to consider. Lower tiers (particularly below 5) are pretty chaotic. I'm not sure what the best stratification would be. Maybe just quartiles? Or maybe the tails are the most interesting, so something like the bottom 10%, 11-50%, 51-90%, top 10% (of winrates).

    ReplyDelete
  5. Okay, I ran such a report, dividing by WR < 46% "good", WR 46%..50% "avg", WR > 50% "bad" and then compared the respective "OP ratings" for the good vs bad group. The problem with using the tails is that we run into too small sample sizes more often.

    Some of the top results were due to insufficient sample size, then came stuff like "SU-122-54", Centurion Mk1, IS-4 (!?), 110, M6A2E1, KV-1S, SU-26, T110. So these would be the tanks where the best players outperform the worst players more than usual.

    Very strange results, I expected the French tanks to be at the top of that list.

    At the bottom of the "good / bad OP ratings" list is stuff like Lorraine 155 51, M22 Locust, AMX38 and other mostly low-tier tanks and arty. Perhaps that means these tanks don't influence the game much therefore the best players cannot win the games for others in them and the worst cannot lose the games for others by taking them.

    I'll think if I can make this presentable, but overall the results are strange enough that I'm not sure there is much value to publishing them. I can upload a 170 MB CSV file with data about battles during the last few weeks if anyone wants to play with it. Or the resulting ~260 row file w/ row for each tank.

    ReplyDelete
    Replies
    1. If you do decide to publish the stratified data, how about just splitting into two categories initially; players with above 50% win ratio and below 50%.

      Delete
    2. Yes, that would make the most sense. Actually above 48% and below 48%, as ~48% is the server mean, I think.

      Delete
  6. This comment has been removed by the author.

    ReplyDelete
  7. Awesome data collection and editing!

    Some remarks to two tanks:
    IS-2, my OP rating: 1,25
    110, my OP rating: 1,18

    These two tanks are hillariously overperformant at the moment.


    Now about another tank:
    I switched to the WZ-111 1-4 ... I MY GOD!
    Compared to the 110
    - my average damage went down from 1950 to 1450
    - my damageration dropped from 1,6 to 0,9
    - my average exp dropped from 1050 to 770

    But suprisingly the vitory ration is at 61,3%, so still OP by 1,04 for me. Mybe I get the "everybody wants to kill me because I'm so rare" bonus

    ReplyDelete
  8. The thing that immediately caught my attention is the massive win% inflation that players engage in. If players had "honest" stats, I would expect Tier X win% to be better than overall win% as the tanks are always top tier in every battle. However, looking at the results, it seems all tier X tanks perform 3-4% below overall. Normally I come down hard on the "stats mean things" side of an argument, but this data looks terrible for win%. Have you thought of doing an OP rating using efficiency/WN7/performance rating?

    ReplyDelete