Jump to content
You need to play a total of 50 battles to post in this section.
Blixies

Do you care about your PR? Do NOT buy the Black ships.

84 comments in this topic

Recommended Posts

Players
367 posts
6,306 battles
7 hours ago, ColonelPete said:

Who cares? I mean: I can throw a dice and get twenty sixes in a row. Possible is everything. But we are talking about statistics so cherry picking some outlier to argue against the expected values is missleading. 

Share this post


Link to post
Share on other sites
[OMPG]
Beta Tester
258 posts
5,382 battles
On 11/29/2019 at 10:46 AM, veslingr said:

But still they insist on PR......most common reason is that WR can also be farmed in division where other 2 player caries you.

 

Not many bad players have high WR and PR (and both are required for good clans)

Solo WR can not be farmed by anyone else but you. :cat_bubble:

 

Too bad default WR is almost always presented in stats programs by your overall WR not solo WR.

Share this post


Link to post
Share on other sites
[CATS]
Players
21,142 posts
12,903 battles
49 minutes ago, The_Noob_Commander said:

Who cares? I mean: I can throw a dice and get twenty sixes in a row. Possible is everything. But we are talking about statistics so cherry picking some outlier to argue against the expected values is missleading. 

While some of it may be statistics, there is also the human factor of damage farming.

 

We are just saying that solo WR is more trustworthy than PR. 

 

 

  • Cool 1

Share this post


Link to post
Share on other sites
[ADRIA]
Players
119 posts
7,504 battles
2 hours ago, ColonelPete said:

We are just saying that solo WR is more trustworthy than PR

Agree with that. On the other hand, you can use both to achieve a more precise evaluation.

 

I just wish WG didn't include premium time XP bonus in the Base XP. Then Base XP would be a much better KPI than PR is now.

Share this post


Link to post
Share on other sites
Players
367 posts
6,306 battles

How does it come that people reject the importance of doing damage that much? I take as an example the Desmo,  a cruiser I personally dont play for damage at all. So lets take all people with at least 80 battles. Because I wanted to safe time I left out the last page (197) and thus the sample consists of 19600 players. As one can see, there is a straight forward linear relationship with r = 0.76. WR and Damage follow a normal distribution (see below) and thus probabilities for "High Damage/ Low WR" or "Low Damage/ High WR" are supposed to be neglectable. For instance, there are 14 people with a WR > 60% and a damage below the first quartile (55697). On the other side, people with a WR > 65% have an average damage (median) of 99700, which is one standard deviation (!) above the average (arithmetic mean) of the total sample. So yeah, Damage is am important factor, even in ships like the Desmo.   

 

 

 

DesmoCorr.jpg

DistriDesmo.jpeg

  • Cool 1

Share this post


Link to post
Share on other sites
[CATS]
Players
21,142 posts
12,903 battles
31 minutes ago, The_Noob_Commander said:

How does it come that people reject the importance of doing damage that much?

That has nothing to do with rejection. We know that damage and damage is not equal.

You do unimportant damage and farm 100k damage on two BB in 20 minutes and not sink one or you sink two DD in in the first 5 minutes and do 40k damage....

The damage stats do not tell us what is what.

 

The guys on one line did the same damage:

shot-19_11.29_14_34.06-0049.thumb.jpg.6e6f5b6c462552ddd2ef1722f668d39f.jpg

I know which ones I prefer on my team.

  • Cool 2

Share this post


Link to post
Share on other sites
Players
367 posts
6,306 battles
11 minutes ago, ColonelPete said:

That has nothing to do with rejection. We know that damage and damage is not equal.

You do unimportant damage and farm 100k damage on two BB in 20 minutes and not sink one or you sink two DD in in the first 5 minutes and do 40k damage....

The damage stats do not tell us what is what.

 

The guys on one line did the same damage:

 

I know which ones I prefer on my team.

Yes, but thats what statistics is NOT about: There might be people with extremely high damage and low win rates, but the odds speak against it. Take Montana (picture below, n = 22.000 players) as an example. To get to the Top 10%/ Top5% Top1% based upon damage you need 106.064/113.076/ 125.897 Damage. So what about the likelihood to meet guys with WR < 60% in those percentiles? There are 1117/ 427/ 47 people fitting that criteria (!) out of 2200/ 1100/ 220 people in those percentiles, which givse us 50.8%/ 38.8%/ 11.0%. You see a pattern?  And 60% WR is actually pretty high - what about the average Dude with a WR of below 50%? The relative frequency values expressed in percentage are 4.1%/ 1.8%/ 0.9%.  And that IS whats PR as as statistical measurement is about. And it does its job very good. If you take the picture below which shows the three parts of the PR value, you can see strong correlation between all of them. Therefore, you can literally mathematically prove that they measure the same thing (latent construct ~ -> PR). So yeah, actually not much to debate. 

 

 

 

Montana.jpeg

Share this post


Link to post
Share on other sites
[CATS]
Players
21,142 posts
12,903 battles

Statistics are relevant when looking at many players.

When I am looking at one player, I want to know what he can do. High damage numbers are nice, but there is a chance that he is just a farmer.

With a player with a 60% solo WR, it is much more unlikely that he is a farmer.

As your graph showed, the variance between players with the same average damage can be 30 %-points of WR.

  • Cool 1

Share this post


Link to post
Share on other sites
Players
1,481 posts
9,661 battles
26 minutes ago, ColonelPete said:

Statistics are relevant when looking at many players.

When I am looking at one player, I want to know what he can do. High damage numbers are nice, but there is a chance that he is just a farmer.

With a player with a 60% solo WR, it is much more unlikely that he is a farmer.

As your graph showed, the variance between players with the same average damage can be 30 %-points of WR.

Yet it also showed there is a clear correlation between pr and wr. Unless you are looking at a single player (which spreadsheetPete rarely does). 

Share this post


Link to post
Share on other sites
[CATS]
Players
21,142 posts
12,903 battles
1 minute ago, GarrusBrutus said:

Yet it also showed there is a clear correlation between pr and wr. Unless you are looking at a single player (which spreadsheetPete rarely does). 

I said that I AM looking at a specific player. I want to know if he is good or not, not gamble.

Share this post


Link to post
Share on other sites
Players
367 posts
6,306 battles
20 minutes ago, ColonelPete said:

I said that I AM looking at a specific player. I want to know if he is good or not, not gamble.

But here is the point where we can meet. One should, of course, not fall into the classical trap of statistical reasoning. But if you ask the question "Does PR values show you differences between the skill level of players?" you are not converned with a single data point in one of those scatter plots but rather about the player base as a statistical group. And here is the answer a straight forward YES.

Share this post


Link to post
Share on other sites
Players
367 posts
6,306 battles
3 minutes ago, ColonelPete said:

But I do not look at the PR of a large amount of players. I might compare two.

For sure, but the question concerning the validity of PR can not be answered when comparing those two players.

Share this post


Link to post
Share on other sites
Players
367 posts
6,306 battles
On 12/2/2019 at 5:06 PM, ColonelPete said:

With a player with a 60% solo WR, it is much more unlikely that he is a farmer.

 

By the Way Pete, Solo WR is an heavily biased measurement ;)

Share this post


Link to post
Share on other sites
[CATS]
Players
21,142 posts
12,903 battles
36 minutes ago, The_Noob_Commander said:

By the Way Pete, Solo WR is an heavily biased measurement ;)

Not when I use it...

Share this post


Link to post
Share on other sites
[FJAKA]
Players
2,785 posts
438 battles

Is it really possible that we need math and statistic to realize purple WR and purple PR = Cheater? :)

Share this post


Link to post
Share on other sites
[CATS]
Players
21,142 posts
12,903 battles
14 minutes ago, The_Noob_Commander said:

I forgot, you are such a brilliant mastermind...

You do not need brilliance to understand stats,  but for some people it may seem that way.

Share this post


Link to post
Share on other sites
Players
367 posts
6,306 battles
Just now, ColonelPete said:

You do not need brilliance to understand stats,  but for some people it may seem that way.

Funny. Especially as you proved throughout the complete debate that you get things wrong, not to mention that I teach stats .. :Smile_trollface:

Share this post


Link to post
Share on other sites
[CATS]
Players
21,142 posts
12,903 battles
9 minutes ago, The_Noob_Commander said:

Funny. Especially as you proved throughout the complete debate that you get things wrong, not to mention that I teach stats .. :Smile_trollface:

I was not talking about you, unless you consider yourself brilliant....

Share this post


Link to post
Share on other sites
Players
367 posts
6,306 battles
9 hours ago, ColonelPete said:

I was not talking about you, unless you consider yourself brilliant....

OK. Than explain to me how you deal with the bias of solo WR if a certain assumpotion is violated. Because, for a usefuel comparison of solo vs. div WR one has to assume that they are temporally at least somewhat identically distributed. In an ideal case,  a person having 10 games a day, that means he plays 5 games solo and 5 games in div. But what if that is not given? And you have most of the solo games in the past, therefore, most of the games resulting in the value are from old games you now use to compare with battles from recent games? From wows-numbers, you can check this only for the last 21 days so how do you know?

Share this post


Link to post
Share on other sites
[CATS]
Players
21,142 posts
12,903 battles
12 minutes ago, The_Noob_Commander said:

OK. Than explain to me how you deal with the bias of solo WR if a certain assumpotion is violated. Because, for a usefuel comparison of solo vs. div WR one has to assume that they are temporally at least somewhat identically distributed. In an ideal case,  a person having 10 games a day, that means he plays 5 games solo and 5 games in div. But what if that is not given? And you have most of the solo games in the past, therefore, most of the games resulting in the value are from old games you now use to compare with battles from recent games? From wows-numbers, you can check this only for the last 21 days so how do you know?

I can see on WoWs numbers the number of solo matches for the past months.

Share this post


Link to post
Share on other sites
Players
367 posts
6,306 battles
12 minutes ago, ColonelPete said:

I can see on WoWs numbers the number of solo matches for the past months.

Thats, true but just how many games overall so you got no more details about the ship types/ classes and so on. You know what I mean? And what would do if someone hasnt played Solo the last 12 months?

Share this post


Link to post
Share on other sites
[CATS]
Players
21,142 posts
12,903 battles
6 minutes ago, The_Noob_Commander said:

Thats, true but just how many games overall so you got no more details about the ship types/ classes and so on. You know what I mean?

Yes, but I can see which ships he played solo overall and the last weeks.

If someone plays only division, then his stats a hard to judge anyway as everything is the result of division effort.

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×