392 replies to this topic

[LittleLovelyGuyFromSpace]

sheriffsherman, on 05 February 2015 - 04:38 PM, said:

 

there is no access to those numbers?  I think thats unfortunate, say for instance my platoon WR is 80% and my regular WR is 51%, i believe that would truly be a good measure of how platooning is helping or not helping individual players.  It's possible that there are folks don't like platooning because they weren't rewarded by it.  So the might be 57%WR solo and 53%WR in platoon explaining the low number of platoons they've been on etc...

 

If it was so simple we would do it from the start. Problem is that we do not know what is the difference between the winrate of platoon and solo victories for players

[SkiFletch]

I keep thinking about the data that is released and trying to come up with a way to calculate it.  There's just nothing since there's no data for platoon losses

[sheriffsherman]

face4stas, on 05 February 2015 - 08:40 AM, said:

 

If it was so simple we would do it from the start. Problem is that we do not know what is the difference between the winrate of platoon and solo victories for players

 

that's too bad, i would love to know how many games i've been platooned in, just for the sake of curiosity.  The numbers are telling me almost 1/3 of my wins have come from platoons(32%)

[Nutellanism]

sheriffsherman, on 05 February 2015 - 08:38 AM, said:

 

there is no access to those numbers?  I think thats unfortunate, say for instance my platoon WR is 80% and my regular WR is 51%, i believe that would truly be a good measure of how platooning is helping or not helping individual players.  It's possible that there are folks don't like platooning because they weren't rewarded by it.  So the might be 57%WR solo and 53%WR in platoon explaining the low number of platoons they've been on etc...

 

(Not that I care) That would also affect people like me. I didn't really know what I was doing until 6000 games and never platooned. Now, I would say I improved and I platoon a lot now. That means itll show that I had much more defeats while going solo rather than platooning because of the noob days. Make sense?

[mbrcmf]

Platooning helps. How much? We will eventually know. The platoon mates are against any normalization because there is some pride attached on their ranking which they feel is primarily due to their skills, and not the platoon. Otherwise, who cares if you are in the top 5%...or 10, if those that are using it only to check other player wr%, strengths or weaknesses...

Interesting after entering some data,the distribution looks very narrow at 1 Sigma, wonder if this is truly a normal distribution.....and factors such as platooning, short timers make up the top/bottom 5%,and the distribution is artificially impacted to prevent people from leaving the game because of poor results? ..  

[sheriffsherman]

Nutellanism, on 05 February 2015 - 08:51 AM, said:

 

(Not that I care) That would also affect people like me. I didn't really know what I was doing until 6000 games and never platooned. Now, I would say I improved and I platoon a lot now. That means itll show that I had much more defeats while going solo rather than platooning because of the noob days. Make sense?

 

very true, i would just like to know for my own fyi kinda thing.  Truth is i've heard thru the grapevine, 'don't platoon with sheriff he's a parasite'  i just want know if its true.  

[sheriffsherman]

On a serious note, i would like to see a small asterik or penalty for former pc users.  I think my numbers would be much hiher if I'd come in with a ton of experience on how to massage the numbers etc.  i know atleast one player with an impressive resume that is completely clueless, eventually if he keeps playing his numbers will even out, but anyone comparing his resume to mine would automatically come to conclusion he is a better player and that is simply not the case.

[mbrcmf]

Since platooning normalization is highly debated, can we also get normalization for the tier V/VI tanks that see +2 MM.  

 

 

;-) (mostly) 

[SkiFletch]

Nutellanism, on 05 February 2015 - 04:51 PM, said:

 

(Not that I care) That would also affect people like me. I didn't really know what I was doing until 6000 games and never platooned. Now, I would say I improved and I platoon a lot now. That means itll show that I had much more defeats while going solo rather than platooning because of the noob days. Make sense?

 

I too was just like Nutellanism here.  Didn't really know what I was doing early no and didn't platoon much.  Since learned a LOT about tanks AND platoon a lot so my metrics have been going up.  Chicken or the egg?  Am I doing better because I downloaded armor inspector and memorized my tank's weaknesses and where the ammo racks are on all tier 8+ tanks?  Am I doing better because I know how to generate specific angles for sidescraping now?  Am I better because I load my equipment or consumables out differently now?  Or is it all because of my higher platoon rate?

 

Chicken or the egg?

[SkiFletch]

mbrcmf, on 05 February 2015 - 05:01 PM, said:

Since platooning normalization is highly debated, can we also get normalization for the tier V/VI tanks that see +2 MM.  

 

 

;-) (mostly) 

 

Look at the per-tank averages.  There's really nothing there to normalize for at a quick glance.  There's some good tanks, some bad tanks within those tiers just like higher tiers.  We may rant about the roughly 20% time you are top tier but it doesn't seem to have an effect on win rate.  Averaging tier 6 win rates, you get 50.76875% win rate.  That's actually higher than the norm. You don't just loose because you're in tier 6.  You just don't have the ability to exert your will over the game because you're in a Usually inferior tank.

[NevirSayDie]

SkiFletch, on 05 February 2015 - 05:07 PM, said:

 

Look at the per-tank averages.  There's really nothing there to normalize for at a quick glance.  There's some good tanks, some bad tanks within those tiers just like higher tiers.  We may rant about the roughly 20% time you are top tier but it doesn't seem to have an effect on win rate.  Averaging tier 6 win rates, you get 50.76875% win rate.  That's actually higher than the norm. You don't just loose because you're in tier 6.  You just don't have the ability to exert your will over the game because you're in a Usually inferior tank.

 

It does have an effect on very good or very poor players. Very poor players actually have a higher WR in tier V and VI, because they're less likely to drag their team down. Very good players have a lower WR, because they're less likely to be able to carry their team. 

 

Also, wotbstars still doesn't have a massive database. For example, my tier V-VI stats are good, but I played a lot of my V/VI matches last summer. After the MM change, I noticed that my WR in V and VI tanks went down, even though I was playing even better than before. Case in point: I played the KV-1s last summer, did about 1200 damage/match, which earned me something like 69% WR. Then I did the M6 after the MM update, dealt 1350 damage/match, and my WR went down to 64%. Why? Because I was bottom tier more often. 

 

But that's all circumstantial. I'm excited to see this rating system take shape! I think the parameters listed in the OP are good. The numbers can be tweaked if need be, but the main thing is that this is a fantastic project. +1 to face! 

Edited by NevirSayDie, 05 February 2015 - 05:19 PM.

[originalmadkilla]

I was thinking how come some of the average win rates for tanks are suspiciously high, and that got me thinking, so I joined a couple of tier two games last night, to get a random test sample, and then I checked the stats of all tier two players on both teams to see if I could find anything suspicious.

 

And to my surprise, I found a few players who had very high win rates in their tank, I'm talking about in the seventy percent range. So I took a closer look at their stats, and one of the players only had like nine battles!

 

Could it be that players with not many battles are artificially skewing the results? I mean if a noob has only played 4 or 5 battles in a tank, it's totally possible for that player to have been lucky and have a 100% win rate. And if their data is given equal weighting compared to the data of another player who has hundreds of battles, then might this explain how certain tank winrates are higher than they should be?

 

I'm no expert on statistics, but should data from players who have less than a certain number of matches played for a certain tank have their results discarded, since it's not reliable? I'm not sure what that number should be, but if somebody has only played a few matches in a particular tank, then their data is not useful or worthwhile at all.

[johndd1_winner]

sheriffsherman, on 05 February 2015 - 05:01 PM, said:

On a serious note, i would like to see a small asterik or penalty for former pc users.  I think my numbers would be much hiher if I'd come in with a ton of experience on how to massage the numbers etc.  i know atleast one player with an impressive resume that is completely clueless, eventually if he keeps playing his numbers will even out, but anyone comparing his resume to mine would automatically come to conclusion he is a better player and that is simply not the case.

Hmm. Not sure how someone who is "completely clueless" also comes with such a good resume?

Also, do tell how people with PC experience can "massage the numbers"? I'd really like to know.

[mbrcmf]

originalmadkilla, on 05 February 2015 - 05:19 PM, said:

I was thinking how come some of the average win rates for tanks are suspiciously high....

And to my surprise, I found a few players who had very high win rates in their tank... 

 

Could it be that players with not many battles are artificially skewing the results?....... 

I'm no expert on statistics, but should data from players who have less than a certain number of matches played for a certain tank have their results discarded, since it's not reliable? I'm not sure what that number should be, but if somebody has only played a few matches in a particular tank, then their data is not useful or worthwhile at all.

Your concern would be if face was taking an average of the averages. I highly doubt that, and is just using the total wins and matches of each to arrive at the tank average. If the distribution across users is not normal, then the first step would be to look at the median, then identify the distribution (bi-modal etc).. And so on.. 

[sheriffsherman]

johndd1_winner, on 05 February 2015 - 09:23 AM, said:

Hmm. Not sure how someone who is "completely clueless" also comes with such a good resume?

Also, do tell how people with PC experience can "massage the numbers"? I'd really like to know.

I'm telling u it's true, it surprised me more than anything to see a player with 60 plus WR doing some really nutty complete noob stuff while platooning.  And it happened in more than one game on different occasions!  I avoid platooning with him altogether now.  I've also noticed that his stats are going down. 

 

It doesn't necessarily have to be someone that came in an avoided certain tank lines, exploited platooning since tier 1 or used certain premiums, etc...  Perhaps it's a case of someone borrowing daddy's account, endulging in substance abuse or even a case of folks purchasing strong profiles.  I've heard the profile thing happens.  There was even a guy spamming on here about selling accounts.

[voyager35]

Actually this platoon thing is not a big issue. Just make it less important when calculating your WN8, or whatever you want to call it in Blitz.

In pc wot, the win rate is not the only factor for WN8 calculation, and nor the most important one. Average damage, spots, and kills actually contribute more to WN8, if I interpret the formula right.

Also, besides platooning, lots of things increase your win rate or efficiency. A forumer averagely can do better than the ignorant. Are we penalizing the forumers? Use gold shells can also increase the ave. damage, and use FE to skip stock tank can also make your numbers look better. Are we penalizing these people?

In the end, it's just a number. If a guy can do 65% win rate without gold and platooning, I will pay him more repect than the one spam gold and do 90% platooning but has the same WR.

[sheriffsherman]

voyager35, on 05 February 2015 - 09:42 AM, said:

Actually this platoon thing is not a big issue. Just make it less important when calculating your WN8, or whatever you want to call it in Blitz.

In pc wot, the win rate is not the only factor for WN8 calculation, and nor the most important one. Average damage, spots, and kills actually contribute more to WN8, if I interpret the formula right.

Also, besides platooning, lots of things increase your win rate or efficiency. A forumer averagely can do better than the ignorant. Are we penalizing the forumers? Use gold shells can also increase the ave. damage, and use FE to skip stock tank can also make your numbers look better. Are we penalizing these people?

In the end, it's just a number. If a guy can do 65% win rate without gold and platooning, I will pay him more repect than the one spam gold and do 90% platooning but has the same WR.

Plus 1, strong post, very true

[SkiFletch]

originalmadkilla, on 05 February 2015 - 05:19 PM, said:

I was thinking how come some of the average win rates for tanks are suspiciously high, and that got me thinking, so I joined a couple of tier two games last night, to get a random test sample, and then I checked the stats of all tier two players on both teams to see if I could find anything suspicious.

 

And to my surprise, I found a few players who had very high win rates in their tank, I'm talking about in the seventy percent range. So I took a closer look at their stats, and one of the players only had like nine battles!

 

Could it be that players with not many battles are artificially skewing the results? I mean if a noob has only played 4 or 5 battles in a tank, it's totally possible for that player to have been lucky and have a 100% win rate. And if their data is given equal weighting compared to the data of another player who has hundreds of battles, then might this explain how certain tank winrates are higher than they should be?

 

I'm no expert on statistics, but should data from players who have less than a certain number of matches played for a certain tank have their results discarded, since it's not reliable? I'm not sure what that number should be, but if somebody has only played a few matches in a particular tank, then their data is not useful or worthwhile at all.

 

How to break this down.  Given the volume of data we have, it's easy to make up a normal set for every tank which face has already done.  That's done by averaging all victories and defeats and the rest of the metrics over all players.  When you create an average, you're "smoothing" out the data so that those outliers like me with a 100% win over 4 games in the E50M don't alter the stats.  My 4 wins are thrown into a pool with tens of thousands of other wins so who cares about them?

 

Remember folks, on a large enough bell curve data base with a 0 sum, for every uber platoon like nuke/dq, there's a steaming pile of potato toon.  It all evens out.

[LittleLovelyGuyFromSpace]

>>A forumer averagely can do better than the ignorant. Are we penalizing the forumers?

 

Please get the point - rating is a trying to determine your performance not the performance of two players who are part of the platoon.

 

We are trying to compare apples and apples. Let me make it very clear what the difference between person who plays solo and who plays platoon in numbers

 

Person who plays solo has the following control on the game inputs:

1. choice for one player win rate and a choice of one tank - nominally it is 14% of the distribution

2. choice for one top tank / one top tier tank operator winrate when seeded as a top tier - nominally it is 25%-50% of the distribution as there are between 2-4 top tier tanks in one game

 

Person who plays platoon:

1. choice for two players win rate and a choice of two tank - nominally it is 28% of the distribution

2. choice for two top tanks / two top tier tank operators winrate when seeded as a top tier - nominally it is 50%-100% of the distribution

 

So platoon players have twice as much of control and in case of when seeded as top tiers they have a full control of the choice of who is going to be the top tier and which tank they are going to play.

 

It allows to influence matchmaking in your favor.

 

It would be unfair to not recognize this advantage and add a provision for players who improve their stats by tweaking matchmaking in their favor trough introducing one tank and one player as they wish.

 

If a solo player can have such a control and say to matchmaking:

 

Ahh I play a Panther, this tank sucks. Please always give me a team mate with T29 and the same winrate as mine.

 

It would be nice. A solo player does not have this luxury when a platoon player does.

 

Conclusion: Fair rating will make a provision to this.

 

Edited by face4stas, 05 February 2015 - 06:40 PM.

[originalmadkilla]

face4stas, on 05 February 2015 - 06:37 PM, said:

It would be nice. A solo player does not have this luxury when a platoon player does.

 

Conclusion: Fair rating will make a provision to this.

 

Well said! I think that If I could duplicate myself and have two of myself in every match, then I have no doubt that my win rate would increase significantly, probably anywhere from 15%-30%, and the same goes for many other solo only players.

 

Certain platooners here are simply not being objective or honest with themselves.