Simple statistical analysis on overpowerness of CVs

[NextToYou]

 

Siema!

 

I started to make a very simple statistic about correlation of victories/losses related to which team has higher tier carrier/carriers in their team.

 

So rules for this data are very simple to understand:

1) Draws are marked as losses to the table (excluded if both teams had highest tier carrier on equal tier, see rule 2)

2) Battles where both team has highest tier carrier on same tier are excluded from the table

3) Battles without carriers are excluded from the table, also games other than 12v12 ships are excluded (too random)

4) Only regular PvP battles - all standard and domination battles are included. Some battles are played solo, some are played in a division of 2-3 players. No "fail platooning" or other tricks will be used, I like to play as well as my skills allow me and as much for fun solo or with my gaming friends (they vary in winrates between 46-62% and might some time play CVs aswell but none are "dedicated" CV players). I am an OK player myself with 55% winrate currently so I don´t influence to the outcome in a team of 12 random players that much.

 

"Tier" means highest tier in the battle, not the highest tier carrier in the match. For example: if my team has a destroyer on tier 6, and a carrier on tier 4, but enemy does not have carrier at all, "Tier" is 6 in the table and "Allied CV higher" is marked as 1.

 

If enemy has a tier 6 and a tier 5 carrier and my team has a tier 5 and 5 carrier, "Enemy CV higher" is marked as 1. And so on. Very simple to understand.

 

I will update every 50 battles from now on and also update the correlation rate to the bottom of the first post as well with every update.

 

I will also take a screenshot from every battle result to my hard drive which fullfils the condition of other team having a higher tier carrier, but I will not include them to this thread. I can host them by request.

 

Date	Tier	Enemy CV higher	Allied CV higher	Result (0 = loss, 1 = victory, draws are losses)
23.7	6	1		0
23.7	6	1		0
23.7	6		1	1
23.7	6		1	1
23.7	7		1	1
23.7	6		1	1
23.7	7		1	1
23.7	7	1		1
23.7	8	1		0
23.7	8	1		0
23.7	10	1		0
23.7	9		1	1
23.7	6		1	1
23.7	6	1		1
23.7	6	1		1
23.7	6		1	1
23.7	6		1	1
23.7	6	1		1
24.7	7		1	1
24.7	6		1	0
24.7	6		1	1
24.7	6		1	1
24.7	7	1		0
24.7	9		1	1
24.7	9	1		1
24.7	6		1	1
24.7	6	1		1
24.7	6	1		0
24.7	6		1	1
24.7	6	1		1
24.7	6	1		1
24.7	6		1	1
24.7	6		1	1
24.7	6		1	1
24.7	6	1		0
25.7	6		1	1
25.7	6	1		0
25.7	7	1		0
25.7	10	1		1
25.7	10	1		0
26.7	9		1	0
26.7	9		1	1
26.7	6		1	1
26.7	8		1	0
26.7	7		1	1
26.7	7	1		0
26.7	7	1		1
26.7	8		1	1
26.7	6		1	1
26.7	6	1		0

 

Current correlation between having higher tier carrier in team and getting a victory is: 0.74 = 74.0%.

 

Last updated on 26.7.2015, 50 battles now that have followed all rules.

 

[Nethraniel]

Thumbs up, for updating every 50 battles. At least that will get some significant numbers in the future.

[RUMIRUMI]

Well atleast you aren't talking about comparisons from wows with wot this time.

[II_Nemesis_II]

Ehm... how does higher tier carrier winning over lower tier one make carriers OP? I just do not get this "statistic".....

[Kirasa]

Alone removing draws from the data invalidates everything you want to prove.

[NextToYou]

czNemesis, on 23 July 2015 - 02:55 PM, said:

Ehm... how does higher tier carrier winning over lower tier one make carriers OP? I just do not get this "statistic".....

 

The team that has lower tier carrier(s) is compensated (usually) having higher tier other class ships (BB, DD or CA). The battles where allied or enemy team has more tier points overall will even out in the long run - these battles will go 50/50 for both sides with high enough sample size.
 

Kirasa, on 23 July 2015 - 02:57 PM, said:

Alone removing draws from the data invalidates everything you want to prove.

 

Draw is a loss for both teams so why those should be included?
 

[II_Nemesis_II]

 

The team that has lower tier carrier(s) is compensated (usually) having higher tier other class ships (BB, DD or CA). The minor differences of one team having more tier points overall will evben out in the long run - these battles will go 50/50 with high enough sample size.

 

Thats just not true at all, how many games with like 3x T9 on one side and 0x T9 on other I had..... This whole statistic is nothing more than MM issue and does nothing to prove that carriers are OP.... completely pointless effort.

[Kirasa]

 

Draw is a loss for both teams so why those should be included?

 

No, it should count as loss to the team with the higher Tier CV, since you want to prove that having the higher Tier CV gives you more wins. Removing draws from the data is an obvious selection bias, and invalidates the statistics.

[NextToYou]

 

No, it should count as loss to the team with the higher Tier CV, since you want to prove that having the higher Tier CV gives you more wins. Removing draws from the data is an obvious selection bias, and invalidates the statistics.

 

OK I can add draws as a loss to the stats (this way I get bigger sample size faster too).

 

I will edit first post.

[delaci76]

Problem is we need more navies and more carrier players then we should have equal tier v equal tier.

 

WG need to change the MM that much is obvious. Otherwise it's unfair on the lower tier player.

 

Part of the reason am not playing carriers atm. (well am free xping them).

[Kirasa]

Problem is we need more navies and more carrier players then we should have equal tier v equal tier.

 

WG need to change the MM that much is obvious. Otherwise it's unfair on the lower tier player.

 

Patch on the 28th will bring changes to the MM.

[delaci76]

 

Patch on the 28th will bring changes to the MM.

 

Fingers x. The carrier inbalance is the most pressing problem I think.

 

Hopefully some different nation premiums. Would be nice to have a British or German...

[NextToYou]

Updated.

[MudMonkey]

Uhm, you guys know that we only can start talking about "statistically significant" numbers if there are at least more than 1000 cases?

[NextToYou]

Yep, that´s why I am updating this every 50 battles (that are not excluded by set rules).

[VonVolks]

good job reporting and people, poor job complaining. Do you own tests and analysis if you dont like how he has done it...

 

I dont really care about the result, but credit for trying to do things "impartially" and with figures...

[St4n]

Ehm... how does higher tier carrier winning over lower tier one make carriers OP? I just do not get this "statistic".....

Because he counts win and loss of the whole game. If the team with the higher tier CV wins significantly more than the other, it can be concluded that it dominates not only the other CV, but the whole game. If other classes could counter the CV advantage of one team than the results would be close to a 50% winrate.

[mtm78]

Because he counts win and loss of the whole game. If the team with the higher tier CV wins significantly more than the other, it can be concluded that it dominates not only the other CV, but the whole game. If other classes could counter the CV advantage of one team than the results would be close to a 50% winrate.

 

Wrong, it can be countered but not with how most people play in random battles. It requires teamwork to counter an enemy high tier CV. To bad this 'data' is skewed because of flawed perception of game mechanics from players making them not play in intended roles a lot. 

[gbgroger]

I salute this worthy analysis!

[22cm]

 

Wrong, it can be countered but not with how most people play in random battles. It requires teamwork to counter an enemy high tier CV. To bad this 'data' is skewed because of flawed perception of game mechanics from players making them not play in intended roles a lot. 

 

I had a game in my Warspite, where the enemy had a tier 8 JP carrier, and we had a tier 8 US BB as his counterpart. We were forced all to clump together for AA cover, so my team did not have much of a game, not to mention fun.

I was taunting the enemy CV player all game trying to get him to concentrate on me, because Warspite is good at dodging torpedoes, so his first 3 attacks were all concentrated on me. He failed to sink me with 3 attacking waves ( i was sunk by a enemy BB). And with all this teamwork of a good team (all clumping together, me drawing the attention of the CV) we could only get a draw.

 

[Puryfie]

Can you guys guess who won the round? crap like this will just drive people away from the game. In what other game can another player, like carrier captains in this, decide who to kill?

[mtm78]

 

I had a game in my Warspite, where the enemy had a tier 8 JP carrier, and we had a tier 8 US BB as his counterpart. We were forced all to clump together for AA cover, so my team did not have much of a game, not to mention fun.

I was taunting the enemy CV player all game trying to get him to concentrate on me, because Warspite is good at dodging torpedoes, so his first 3 attacks were all concentrated on me. He failed to sink me with 3 attacking waves ( i was sunk by a enemy BB). And with all this teamwork of a good team (all clumping together, me drawing the attention of the CV) we could only get a draw.

 

 

Sometimes a draw is a moral victory 

 

The problem right now is that you can't call for balance changes ( other then MM ) at this time based on performance of a few top tier CV players. With higher server population at those tiers we should get more balanced matches. If things don't work out, that's a good time to look at overall balance, but not right now I think.

[Bushpanther]

 

Wrong, it can be countered but not with how most people play in random battles. It requires teamwork to counter an enemy high tier CV. To bad this 'data' is skewed because of flawed perception of game mechanics from players making them not play in intended roles a lot. 

 

Lack of teamwork is pretty much a feature in this game because coordinating actions without voicechat is pretty hard to begin with.

 

So while you call this data bad, it's actually very much a factual representation of how the game currently pans out. In an ideal world teamwork will counter a higher tier CV, but with the inherent difficulty of coordination in this game having a higher tier carrier on one of the teams is very unbalancing. Which is exactly what that 74% winrate in the OP post reflects.

[Blue_Qbe]

 

 

Great analasis. Keep up the good work and maybe we will see a brighter future.

 

[St4n]

 

No, it should count as loss to the team with the higher Tier CV, since you want to prove that having the higher Tier CV gives you more wins. Removing draws from the data is an obvious selection bias, and invalidates the statistics.

You obviously missed or missinterpreted:

 

1) Draws are marked as losses to the table (excluded if both teams had highest tier carrier on equal tier, see rule 2)

They are allready marked as loss. Battles with same tier CVs are excluded not only in a draw but in any case (rule 2) and rightfully so. If both teams have equal tier CVs there is no "higher tier".

Taking those battles into account with both counted as "higher tier" would drag this statistic wrongly close to 50% (one wins the other losses definitelly) and beyond that down (2 losses in case of a draw).

Including that data into this would be observation bias. Simply proven by the fact that I could as well say 2 same tier CVs are both the "lowest tier" and therfore this data should count for them.