Jump to content
You need to play a total of 5 battles to post in this section.
wjp120

Is "average" PR and WTR really average?

21 comments in this topic

Recommended Posts

Members
92 posts
5,342 battles

I'm wondering if the sampling of wows-numbers and warshipstoday is truly representative of the player base. Or, does it only count players who are actively looked up? If so, would the averages change significantly if every active player was counted?

Share this post


Link to post
Share on other sites
205
[PT8TO]
[PT8TO]
Members
384 posts
13,745 battles

Average would be even lower since not every player release their stats. Ofc, stats are not super accurate either; there are sync drop accounts, dummy CVs, bots, etc. that mess with the server average.

Share this post


Link to post
Share on other sites
647
[1STBG]
Members
1,245 posts
5,228 battles

The WR component is probably 50%, the others (survival rate, frags, etc) probably somewhat subjective. I think it's geared towards competitive/CB's, where "average" is considered somewhat below average competitively. 

Share this post


Link to post
Share on other sites
Beta Testers
100 posts
3,446 battles
9 minutes ago, wjp120 said:

I'm wondering if the sampling of wows-numbers and warshipstoday is truly representative of the player base. Or, does it only count players who are actively looked up? If so, would the averages change significantly if every active player was counted?

Both sites only show players that have been looked up. That’s how the WG API works so they don’t really have a way to do anything else. So I would guess that if you had true server numbers things would generally trend lower as most players probably aren’t looking themselves up to check their stats.

Share this post


Link to post
Share on other sites
1,091
[NGAGE]
Beta Testers, In AlfaTesters
2,931 posts
10,485 battles

I think instead of looking at the mean wr, we should look at the mode. Most players are going to be at 48-49% wr

  • Cool 1

Share this post


Link to post
Share on other sites
2,795
Alpha Tester
7,112 posts
3,722 battles

Practically every player gets looked up thanks to the stat monitor program people run. The number of people not looked up isn't going to be significant.

  • Cool 3

Share this post


Link to post
Share on other sites
4,015
[GWG]
[GWG]
Alpha Tester, In AlfaTesters
17,998 posts
10,482 battles
32 minutes ago, DouglasMacAwful said:

The WR component is probably 50%, the others (survival rate, frags, etc) probably somewhat subjective. I think it's geared towards competitive/CB's, where "average" is considered somewhat below average competitively. 

Average WR for WoT is considered to be 48% - 52% and would place it there for this game too. The other numbers such as PR are not very reliable.

Share this post


Link to post
Share on other sites
Members
40 posts
63 battles

the average rating that is on the stats website isnt really the average.  that website has certain levels to it. if you look closely the rating system you can see that for every 200 points it changes the level that you are on. 2000 points is Unicum, 1800 is Very good, 1600 is good and stuff like that. PR has different levels too. I might be wrong about the numbers as I am on my mobile and cant check the actual numbers. you can do that for yourself though

Share this post


Link to post
Share on other sites
896
[XBRTC]
Members
2,437 posts
8,474 battles
2 hours ago, DouglasMacAwful said:

The WR component is probably 50%, the others (survival rate, frags, etc) probably somewhat subjective. I think it's geared towards competitive/CB's, where "average" is considered somewhat below average competitively. 

 

CB stats aren't even accessible publicly. The standard numbers that people see when they look someone up are for random battles only.

Share this post


Link to post
Share on other sites
3,985
[-K-]
Supertester
3,124 posts
6,984 battles

I think its hard to talk about average performance.  I think median is a more meaningful description of what we perceive as average.  I.e., there are an equal number of players that play better and worse than this particular metric.  Which is why I like the WOWS Stats and Numbers site.  It has ship performance broken down into averages of the top 50%, top 25%, top 10% and top 5% of players.  I think if you want to compare your performance and find out if you're average or how much above or below average you are, this is the way to do it.

I think the overall PR or WTR numbers are pretty good reflections of player skill.  There are flaws, but a player with a PR of 2600 is generally going to be better than one with a PR of 1900.  A 1750 WTR player is generally better than a 1500 WTR player.  

Edited by Pope_Shizzle

Share this post


Link to post
Share on other sites
38
[TATER]
Members
85 posts
1,327 battles
8 minutes ago, Pope_Shizzle said:

 

I think the overall PR or WTR numbers are pretty good reflections of player skill.  There are flaws, but a player with a PR of 2600 is generally going to be better than one with a PR of 1900.  A 1750 WTR player is generally better than a 1500 WTR player.  

There may be a positive correlation between the PR score and skill (using W/L as the base measure of "skill"), but I suspect it's not a particularly good predictor...or even a bad one.  These are multiple reasons why that model due to overweight and underweight of certain stats like damage, sinkings, players deaths, etc.  In fact, for a number of my best (and worst) ships there is a negative correlation using the PR number. 

Share this post


Link to post
Share on other sites
3,985
[-K-]
Supertester
3,124 posts
6,984 battles
Just now, YouSatInGum said:

There may be a positive correlation between the PR score and skill (using W/L as the base measure of "skill"), but I suspect it's not a particularly good predictor...or even a bad one.  These are multiple reasons why that model due to overweight and underweight of certain stats like damage, sinkings, players deaths, etc.  In fact, for a number of my best (and worst) ships there is a negative correlation using the PR number. 

Like I said, I think OVERALL PR and WTR are pretty good reflections of player skill.  When you start to break it down by individual ship, the flaws in the rating become more visible because of the massively smaller sample size.  Outliers in performance start to have a more significant impact.  But I stand 100% behind what I said, that a 2600 PR player is probably going to be better than someone at the 1900 level.  It's not a measure of potential, but one of consistency.  

Share this post


Link to post
Share on other sites
737
[HEROS]
Members
1,509 posts
12,830 battles

Win rate, yea, it counts.. WTR how ever is a composite stat.  And I don't consider it a very trustworthy one.  It rewards damage farming more the frags and winning a match.. You can have a high WTR and have a sub 50% WR .   More importantly is a player WR in a specific ship.  Also of particular importance is is that WR from solo play or divisioned play.  Player that only play divisioned will tend to have a WR 55% or better.  

If you really want to know how good or bad a player is go spend 20 matches with them... NOT divisioned with them and see not only how they do, but what they do and when they do it.

Stats are where you start with evaluating how well some one plays.  The mark One eyeball is still the better tool.

Edited by TL_Warlord_Roff
  • Cool 1
  • Bad 1

Share this post


Link to post
Share on other sites
130
[-K-]
Alpha Tester
363 posts
10,688 battles
27 minutes ago, TL_Warlord_Roff said:

Win rate, yea, it counts.. WTR how ever is a composite stat.  And I don't consider it a very trustworthy one.  It rewards damage farming more the frags and winning a match.

Yeah, no. For starters, both include win rate as one component of the rating. Secondly, doing damage is an integral part in winning matches; someone has to actually damage enemy ships enough for them to die. Every team has a finite amount of health that's set at the beginning of the match. If you reduce the red team's health pool to zero before yours goes to zero (barring a points or capture victory), you win. There is no such thing as bad damage; however, there is such a thing as damage which could be better applied elsewhere. Even fire damage on BBs is useful because every point of fire damage they have to spend a heal on is a point of AP damage that doesn't get healed immediately. Even if they eventually heal up all the fire damage, taking a BB out of the fight for a few minutes so they can hide and heal is also tactically advantageous. Could the formulas used to calculate WTR/PR be improved? Absolutely. But it is laughable to assert that they have no correlation with winning.

27 minutes ago, TL_Warlord_Roff said:

You can have a high WTR and have a sub 50% WR .

Prove it. Find me one player with a green WTR and sub-50% w/r. I'm going to guess that you can't. Even if I disagree with how things are weighted, the sorts of things you have to do to get good WTR/PR are the exact same sorts of things you need to do to win more matches.

27 minutes ago, TL_Warlord_Roff said:

Player that only play divisioned will tend to have a WR 55% or better.  

No. Divisions amplify the quality of player used to make them. If you throw three average players in a division, they might win slightly more often than average due to the increased teamwork, say 52% maybe. If you throw three 45% players together, they are going to lose far more often than 45% because you have now replaced one quarter of your team with extremely sub-par players.

  • Cool 2

Share this post


Link to post
Share on other sites
737
[HEROS]
Members
1,509 posts
12,830 battles
16 hours ago, Dianeces said:

Yeah, no. For starters, both include win rate as one component of the rating. Secondly, doing damage is an integral part in winning matches; someone has to actually damage enemy ships enough for them to die. Every team has a finite amount of health that's set at the beginning of the match. If you reduce the red team's health pool to zero before yours goes to zero (barring a points or capture victory), you win. There is no such thing as bad damage; however, there is such a thing as damage which could be better applied elsewhere. Even fire damage on BBs is useful because every point of fire damage they have to spend a heal on is a point of AP damage that doesn't get healed immediately. Even if they eventually heal up all the fire damage, taking a BB out of the fight for a few minutes so they can hide and heal is also tactically advantageous. Could the formulas used to calculate WTR/PR be improved? Absolutely. But it is laughable to assert that they have no correlation with winning.

Prove it. Find me one player with a green WTR and sub-50% w/r. I'm going to guess that you can't. Even if I disagree with how things are weighted, the sorts of things you have to do to get good WTR/PR are the exact same sorts of things you need to do to win more matches.

No. Divisions amplify the quality of player used to make them. If you throw three average players in a division, they might win slightly more often than average due to the increased teamwork, say 52% maybe. If you throw three 45% players together, they are going to lose far more often than 45% because you have now replaced one quarter of your team with extremely sub-par players.

And the other issue with those who swear by WTR/PR is it's arrogantly assumptive.   The formula for WTR/PR fits what the creators think is important, but the stats are all about the player not the team.    

This is first and foremost a team play game.  You may play it "singly"  but even crappy me has had matches where I performed outstandingly and still lost  (go to you tube.. from years back, search for "Fortress Fuso").

Quite simply being good in this game is not just how good you are at dealing damage but how good you are at damaging the RIGHT target and stats, especially composite such as WTR and PR stats only barely hint at that.  

I've been wargaming over 50 years at this point.  I do know what I'm talking about.  But I'm not trying to sell anyone on a fundamentally flawed composite stat.   Something I have seen time and again in gaming over the last decade.  Honestly you younger ones think and act like this is all something you invented but it's not.   Far from it.   Stats have been tracked in competitive tabletop play for over 50 years to the best of my knowledge starting in the late 60's, early 70 at the Origins convention that was annually put on by Avalon Hill Games in Baltimore MD.   

What I will give you is both WTR and PR are "close" but I will NOT use them as a yardstick for a player level of skill.  Nor should you.  Only as a place to start.  As a stated previously if you really want to know how good someone is or can be you have to go in match with them.  Trust your eyes balls.   Numbers can be manipulated, and are, especially when someone is trying to prove some particular point.  Especially when they're numbers that are used to say. 

"I'm better then you are"

....  and who knows, you might be better than I.  But if you run into ME in a match and I'm "on" that particular day.  you'll know you've been in a fight.   

Once the fight starts stats become meaningless.  In "pro" play they mean less then nothing.  They're there to give announcers something to chatter about when the game pace is slow and to fill dead air while players/teams sort them selves out for the next phase of a match. 

If you don't treat every single last player in a match as gods gift to gaming you'll be handed your shot up ship in a silver platter.

The worlds best swordsman only truly fears the worlds worst swordsman because they will do the unexpected that no good swordsman would ever consider.

The video I said go look up shows it very well.  The game was still new and only the IJN and the USN had full lines, and I.. well I frankly in retrospect stunk as a player at the time, but the players that were in there with me stunk even worse.  6 kills and a lost fight.    It's a rather entertaining fight as I did pull off a little trick that is very rarely seen even today years later.   You'll recognize it when you see it  (not that the reviewer doesn't point it out, he does)   I adapted to the situation and that is the mark of a good player.  

Stats only hint, they do not tell in any other area then the extremes and outliers.  For middle of the pack players such as myself stats just don't talk to you real well.  You guys that seem to consider stats the end all be all need to learn this.  

 

  • Cool 2
  • Boring 1
  • Bad 1

Share this post


Link to post
Share on other sites
35
[TASC]
Members
210 posts
6,709 battles
19 minutes ago, TL_Warlord_Roff said:

And the other issue with those who swear by WTR/PR is it's arrogantly assumptive.   The formula for WTR/PR fits what the creators think is important, but the stats are all about the player not the team.    

This is first and foremost a team play game.  You may play it "singly"  but even crappy me has had matches where I performed outstandingly and still lost  (go to you tube.. from years back, search for "Fortress Fuso").

Quite simply being good in this game is not just how good you are at dealing damage but how good you are at damaging the RIGHT target and stats, especially composite such as WTR and PR stats only barely hint at that.  

I've been wargaming over 50 years at this point.  I do know what I'm talking about.  But I'm not trying to sell anyone on a fundamentally flawed composite stat.   Something I have seen time and again in gaming over the last decade.  Honestly you younger ones think and act like this is all something you invented but it's not.   Far from it.   Stats have been tracked in competitive tabletop play for over 50 years to the best of my knowledge starting in the late 60's, early 70 at the Origins convention that was annually put on by Avalon Hill Games in Baltimore MD.   

What I will give you is both WTR and PR are "close" but I will NOT use them as a yardstick for a player level of skill.  Nor should you.  Only as a place to start.  As a stated previously if you really want to know how good someone is or can be you have to go in match with them.  Trust your eyes balls.   Numbers can be manipulated, and are, especially when someone is trying to prove some particular point.  Especially when they're numbers that are used to say. 

"I'm better then you are"

....  and who knows, you might be better than I.  But if you run into ME in a match and I'm "on" that particular day.  you'll know you've been in a fight.   

Once the fight starts stats become meaningless.  In "pro" play they mean less then nothing.  They're there to give announcers something to chatter about when the game pace is slow and to fill dead air while players/teams sort them selves out for the next phase of a match. 

If you don't treat every single last player in a match as gods gift to gaming you'll be handed your shot up ship in a silver platter.

The worlds best swordsman only truly fears the worlds worst swordsman because they will do the unexpected that no good swordsman would ever consider.

The video I said go look up shows it very well.  The game was still new and only the IJN and the USN had full lines, and I.. well I frankly in retrospect stunk as a player at the time, but the players that were in there with me stunk even worse.  6 kills and a lost fight.    It's a rather entertaining fight as I did pull off a little trick that is very rarely seen even today years later.   You'll recognize it when you see it  (not that the reviewer doesn't point it out, he does)   I adapted to the situation and that is the mark of a good player.  

Stats only hint, they do not tell in any other area then the extremes and outliers.  For middle of the pack players such as myself stats just don't talk to you real well.  You guys that seem to consider stats the end all be all need to learn this.  

 

good points sir, i too have been gaming for a long time and have helped companies get a game ready for market. ( Europa series ) for 1. and yes we have tracked stats and argued over what one meant more for a long time.  i like your idea because in the end we are here to have fun and if someone isn't fun to play with i don't want to be on their team. stat shamers are high on my list of " not on my team" . i know i stink  what of it? i'm having fun and i have had some very fun games that were losses but just wonky in how they went. 

i hope to see you out there sometime. 

Share this post


Link to post
Share on other sites
2,195
[-K-]
[-K-]
Members
3,781 posts
13,643 battles

Because a fair number of people now use software that automatically looks up everybody in a given match, it's likely that most players who play mid-to-high-tier Random Battles have gotten added to the databases as some point. I would expect a systematically lower rate of inclusion of co-op-heavy and low-tier players, simply because they are less likely to be in a match with someone running those mods.

Additionally, sites like wows-stats specifically exclude players with less than 20 games played in a ship from their global averages. So the "average" displayed is biased toward players with more playtime, and/or who play specific ships more, as opposed to getting a T10 and playing it once before starting on the next line grind.

In terms of whether arithmetic means of WR, WTR, and PR actually match up with the semantic definition of "average"... yes, with big caveats. I expect those systems work decently well in the realm of "average", but become increasingly unreliable toward the high end of the skill range.

Share this post


Link to post
Share on other sites
130
[-K-]
Alpha Tester
363 posts
10,688 battles
2 hours ago, TL_Warlord_Roff said:

The formula for WTR/PR fits what the creators think is important, but the stats are all about the player not the team.    

Yes, and? Over a large enough sample size, say a couple thousand battles or so, any sort of team effects are statistically insignificant. Unless you're one of those conspiracy theorists who think that Wargaming is out to get you and consistently saddles you with worse than average teams.

2 hours ago, TL_Warlord_Roff said:

You may play it "singly"  but even crappy me has had matches where I performed outstandingly and still lost  (go to you tube.. from years back, search for "Fortress Fuso").

Again, so what? A sample size of one is meaningless.

2 hours ago, TL_Warlord_Roff said:

I've been wargaming over 50 years at this point.  I do know what I'm talking about.  But I'm not trying to sell anyone on a fundamentally flawed composite stat.   Something I have seen time and again in gaming over the last decade.  Honestly you younger ones think and act like this is all something you invented but it's not.   Far from it.   Stats have been tracked in competitive tabletop play for over 50 years to the best of my knowledge starting in the late 60's, early 70 at the Origins convention that was annually put on by Avalon Hill Games in Baltimore MD.   

What I will give you is both WTR and PR are "close" but I will NOT use them as a yardstick for a player level of skill.  Nor should you.  Only as a place to start.  As a stated previously if you really want to know how good someone is or can be you have to go in match with them.  Trust your eyes balls.   Numbers can be manipulated, and are, especially when someone is trying to prove some particular point.  Especially when they're numbers that are used to say. 

Doing something for a long period of time is not in any way the same thing as doing things well. If I live to be 150, I will never sculpt as well as Michelangelo or be able to defeat Magnus Carlsen in chess. I will almost certainly improve from where I am now, maybe even become something resembling good at it. It is, on the other hand, entirely possible that I will continue to suck at these things. Trying to play the "I've been doing this far longer than you have card" is exactly the same thing as stat shaming, only it's using something completely irrelevant to try and shut down the conversation.

Furthermore, this whole stats vs. eye test argument is so incredibly boring at this point. Major league sports teams primarily use stats to inform their hiring decisions. The reason they do this is because they are in the business of putting out the best possible teams and (spoiler alert) stats do this better than the eye test. We've seen this happen in baseball, there are big pushes to include more and better stats in both hockey and soccer, and even football is becoming more data driven. I can't tell you that someone with a 750 WTR is a better boats player than someone with a 700 WTR, not for certain. But I can tell you that someone with a 1000 WTR is almost certainly better than either of those players without playing with any of them.

2 hours ago, TL_Warlord_Roff said:

Stats only hint, they do not tell in any other area then the extremes and outliers.  For middle of the pack players such as myself stats just don't talk to you real well.  

This is just patently false. It's more of the same tedious stats denial that has been going on since pretty much day one of World of Tanks. And I'm not even going to touch the middle of the pack bit since I value my ability to continue posting.

Share this post


Link to post
Share on other sites
978
[OO7]
[OO7]
Members
1,806 posts
6,219 battles
4 hours ago, TL_Warlord_Roff said:

....  and who knows, you might be better than I.  But if you run into ME in a match and I'm "on" that particular day.  you'll know you've been in a fight.   

I doubt it.  I spend half of my games alt-tabbed to begin with.  It doesn't make a difference if you feel 'on' today or not.  The game is boiled down to be as simple as tic-tac-toe for us, and placing your 'O' in a corner is neither shocking nor difficult.  

Share this post


Link to post
Share on other sites
Supertester
36 posts
8,145 battles

The average numbers on the stat sites are definitely inflated due to bad players hiding their stats far more often than good ones - just do the math for winrates among a particular class and tier and you'll see the average recorded winrate is actually slightly over 50%, even though it should be just a little under 50% due to Draws.

If players with hidden stats had their games counted, the stats of the "average" player would drop.

Share this post


Link to post
Share on other sites
352
[KRAB]
Members
732 posts
7,221 battles

It doesn't help that the stat distributions are asymmetrical. 

There are a LOT of players who are just slightly below what we would consider average (50% WR) offset by the extended tail of people maintaining 60%+ 

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×