Jump to content
You need to play a total of 20 battles to post in this section.
Snargfargle

Another example of why winrates don't always tell the whole story

26 comments in this topic

Recommended Posts

3,628
[CAAT]
Members
6,313 posts
8,036 battles
Quote

Another example of why winrates don't always tell the whole story

On a somewhat unrelated but kinda related note, this is also why winrate isn't necessarily directly indicative of a ship's performance (or lack thereof).

In other words, winrate (especially on its own)really doesn't tell the whole story! Gotta look at ALL the variables.

On topic though, I mean, it was Hornet vs Saipan, and Saipan isn't exactly in the greatest place right now in terms of performance....so even though that Saipan player was unicum or whatever, Saipan probably let them down a bit here. Maybe? I'm not really sure tbh XD

Edited by SaiIor_Moon
  • Thanks 2

Share this post


Link to post
Share on other sites
778
[RSL]
Members
897 posts
13,701 battles
13 hours ago, arch4random said:

i think he is comparing the 2 cvs

Ah.

Well, I still don't understand the point of the post. bad players can have good games, and good players can have bad games. That's why you always need to have a large enough sample size. So just posting one match and saying "See! W/R doesnt matter" really doesn't prove anything. 

Edited by SirPent13
  • Cool 10

Share this post


Link to post
Share on other sites
4,970
[ARS]
Beta Testers
8,353 posts
9,639 battles
7 minutes ago, SirPent13 said:

Ah.

Well, I still don't understand the point of the post. bad players can have good games, and good players can have good games. That's why you always need to have a large enough sample size. So just posting one match and saying "See! W/R doesnt matter" really doesn't prove anything. 

There are a number of MMM users who will immediately give up if the skill level is significantly mismatched.  It is rather self fulfilling really.

  • Cool 4

Share this post


Link to post
Share on other sites
Beta Testers
156 posts
5,186 battles

Statistics like this are good for analyzing long-term trends. On a battle by battle basis, anything can (and often does) happen. That's one reason why tools like MMM are a bit silly.

Share this post


Link to post
Share on other sites
1,302
[PAT]
Members
1,258 posts
15,083 battles

I mean, it depends which WR you're referring to.

The discrepancy in ship to account WR suggests the losing CV typically plays in a division, and probably isn't that strong of a player individually.

Share this post


Link to post
Share on other sites
3,861
[TDRB]
Members
7,373 posts
16,186 battles
17 minutes ago, Helstrem said:

There are a number of MMM users who will immediately give up if the skill level is significantly mismatched.  It is rather self fulfilling really.

I seen many post in this forum confirming you comment is factual.

17 minutes ago, jediknight120 said:

Statistics like this are good for analyzing long-term trends. On a battle by battle basis, anything can (and often does) happen. That's one reason why tools like MMM are a bit silly.

True ^^^^^^ In baseball it is not rare that we see the statistically better players fail & the statistically poorer player get the game winning hit.

Share this post


Link to post
Share on other sites
Members
653 posts
3 hours ago, SaiIor_Moon said:

.so even though that Saipan player was unicum or whatever

He is not a unicum (55%), so he is whatever.

Share this post


Link to post
Share on other sites
3,628
[CAAT]
Members
6,313 posts
8,036 battles
14 minutes ago, Lazy_Joe said:

He is not a unicum (55%), so he is whatever.

Ooops, didn't see his winrate when I wrote that, so whatever it is! ;P

tbh, I just saw the purple and assumed unicum lols

Edited by SaiIor_Moon

Share this post


Link to post
Share on other sites
2,772
[SR-_-]
Members
5,505 posts
53,162 battles
6 hours ago, SirPent13 said:

Ah.

Well, I still don't understand the point of the post. bad players can have good games, and good players can have good games. That's why you always need to have a large enough sample size. So just posting one match and saying "See! W/R doesnt matter" really doesn't prove anything. 

That would go on the presumption that people are different and yet they are actually the same except for the circumstances of the situation.

It is like that scene in "History of the World: Part I" when Princess Nympho is selecting her escorts...

"No, no, no, no, yes, no, no, yes, no, yes, no, no, no, no, no, yes, no, no, no, yes, no, no, no, yes, oh my yes!" To paraphrase. 

Phalanx! Forward Harch!. 

Nympho: "I always loved Forward Harch!"

You are right, we should have a thread of Unicums having bad games. Lots of them, scores of them. 

Would that make a big sample? Or a small one?

LOL

 

Edited by SteelRain_Rifleman
  • Haha 1

Share this post


Link to post
Share on other sites
1,441
[TMS]
Members
4,222 posts
43,514 battles
6 hours ago, jediknight120 said:

Statistics like this are good for analyzing long-term trends. On a battle by battle basis, anything can (and often does) happen. That's one reason why tools like MMM are a bit silly.

MMM looks at all the battles.

Total account battles Winrate and Total winrate battles for the current ship played. 

Edited by Final8ty

Share this post


Link to post
Share on other sites
2,867
[ARR0W]
Members
6,168 posts
32,942 battles
11 hours ago, SirPent13 said:

Ah.

Well, I still don't understand the point of the post. bad players can have good games, and good players can have good games. That's why you always need to have a large enough sample size. So just posting one match and saying "See! W/R doesnt matter" really doesn't prove anything. 

Its a contrarian statement; we are supposed to give up on a sound general predictor because of outliers. 

Share this post


Link to post
Share on other sites
12,714
[PSP]
[PSP]
Members
19,325 posts
30 minutes ago, Pugilistic said:

Its a contrarian statement; we are supposed to give up on a sound general predictor because of outliers. 

Many people will see that the other team has a 53% average winrate while theirs has a 51% average winrate and think "Well, it's no use trying, we are going to lose anyway," especially if they see that the "big, mean CV" on their team has a significantly lower winrate in his ship than the other CV driver has in his. However, a few percentage points difference in overall winrate is probably not even significant and there may be several reasons why one player has an excellent winrate in a ship after playing only a few battles in it while another has a low winrate in a similar ship. The purpose of this thread is to show that thinking that you are going to lose because the matchmaking monitor "tells you" so is oftentimes a self-fulfilling prophesy.

Edited by Snargfargle
  • Cool 3

Share this post


Link to post
Share on other sites
208
[GROGS]
Members
481 posts
13,815 battles
11 hours ago, Helstrem said:

There are a number of MMM users who will immediately give up if the skill level is significantly mismatched.  It is rather self fulfilling really.

Yes indeedy. I know a few people like this, MMM loads and they'll start by predicting a loss, then focus on who to blame.

One of the first things I learned playing sports is that you play every game to win. Even if you've lost the last 4 games, you gotta keep your chin up and expect to win the next. If you're defeated in your mind, you'll be defeated on the field.

Share this post


Link to post
Share on other sites
2,867
[ARR0W]
Members
6,168 posts
32,942 battles
12 hours ago, Helstrem said:

There are a number of MMM users who will immediately give up if the skill level is significantly mismatched.  It is rather self fulfilling really.

 

1 hour ago, Snargfargle said:

Many people will see that the other team has a 53% average winrate while theirs has a 51% average winrate and think "Well, it's no use trying, we are going to lose anyway," especially if they see that the "big, mean CV" on their team has a significantly lower winrate in his ship than the other CV driver has in his. However, a few percentage points difference in overall winrate is probably not even significant and there may be several reasons why one player has an excellent winrate in a ship after playing only a few battles in it while another has a low winrate in a similar ship. The purpose of this thread is to show that thinking that you are going to lose because the matchmaking monitor "tells you" so is oftentimes a self-fulfilling prophesy.

Being on the short side of a -11% MM just makes me hitch up my britches. it may be predictive, but I'm giving my best effort. There are also statistical anomalies from low numbers of games played as mentioned, these can usually be spotted and taken into account. 

Edited by Pugilistic

Share this post


Link to post
Share on other sites
398
[F4E]
Members
605 posts
10,580 battles
1 hour ago, Snargfargle said:

Many people will see that the other team has a 53% average winrate while theirs has a 51% average winrate and think "Well, it's no use trying, we are going to lose anyway," especially if they see that the "big, mean CV" on their team has a significantly lower winrate in his ship than the other CV driver has in his. However, a few percentage points difference in overall winrate is probably not even significant and there may be several reasons why one player has an excellent winrate in a ship after playing only a few battles in it while another has a low winrate in a similar ship. The purpose of this thread is to show that thinking that you are going to lose because the matchmaking monitor "tells you" so is oftentimes a self-fulfilling prophesy.

I'm not a fan of MMM, or SBMM, etc,

...but you do realize the folly of posting one example and saying "Look, I just disproved all the other evidence!!"

 

I sank a Petro once, thus it must not be an OP ship!

I played DD once, and the CV didn't kill or contribute to my death, so balanced!!

I was in a game with subs (not in a sub myself) and I had fun, so subs are okay and make this game better!!!

 

Any other theories we can disprove with one example?  Or would some of them require at least two examples??

 

  • Cool 1

Share this post


Link to post
Share on other sites
206
[P_P]
Members
500 posts
1 hour ago, Snargfargle said:

Many people will see that the other team has a 53% average winrate while theirs has a 51% average winrate and think "Well, it's no use trying, we are going to lose anyway,"

i don't understand how a 2% difference in w/r or even a 5% could make much difference when they each have barley played any battles in those ships... 43 to 25? am i reading that right?

even the avg ships is only 82 to 111 for the teams. I must be reading that wrong yeah?

Is Account battles just random games or also include PvE?

Share this post


Link to post
Share on other sites
12,714
[PSP]
[PSP]
Members
19,325 posts
3 hours ago, Retarie said:

Is Account battles just random games or also include PvE?

In the matchmaking monitor you can move a slider to get the values for Random battles or just the battles in the mode that you are currently playing (e.g., Ranked). I usually leave it on Random battles to get the largest PVP data set's results. WOWS Stats and Numbers has everything broken down in all the categories except for Clan Battles.

Many people, myself included, have vastly different winrates in certain ships than they do overall. If you see me in my Massachusetts then you probably have some reason to be concerned if you are on the other team. However, if you see my in my Worcester, and you are on the other team, then you can sit back and relax as I've never been able to hit anything very well with it.

 

  • Funny 1

Share this post


Link to post
Share on other sites
11,355
[SALVO]
Members
16,508 posts
10,225 battles
4 hours ago, Retarie said:

when they each have barley played any battles in those ships... 43 to 25? am i reading that right?

Kek, I have like only 5 ships with over 50 matches... How do we read that? 

14 hours ago, Retarie said:

what surprises me, is that person has 26 Karma left after playing cv's

40 Bearn matches still left me a net positive karma :p 

14 hours ago, Lazy_Joe said:

He is not a unicum (55%), so he is whatever.

Yikes... That ball spiked close :Smile_trollface: 

 

  • Cool 1

Share this post


Link to post
Share on other sites
206
[P_P]
Members
500 posts
5 hours ago, Snargfargle said:

In the matchmaking monitor you can move a slider to get the values for Random battles or just the battles in the mode that you are currently playing (e.g., Ranked). I usually leave it on Random battles to get the largest PVP data set's results. WOWS Stats and Numbers has everything broken down in all the categories except for Clan Battles.

Many people, myself included, have vastly different winrates in certain ships than they do overall. If you see me in my Massachusetts then you probably have some reason to be concerned if you are on the other team. However, if you see my in my Worcester, and you are on the other team, then you can sit back and relax as I've never been able to hit anything very well with it.

 

Haha, yeah we all have ships like that.

Thanks, i haven't ever used MMM and to be honest, this is the first time I've really looked and tried to understand what it says

Share this post


Link to post
Share on other sites
206
[P_P]
Members
500 posts
6 hours ago, ArIskandir said:

Kek, I have like only 5 ships with over 50 matches... How do we read that?

40 Bearn matches still left me a net positive karma :p 

Yikes... That ball spiked close :Smile_trollface: 

 

Heh, don't mind me, Im probably just a slow learner. Takes me a good few more games then that in a ship before i feel that i understand how to play it.

edit - I might have played for a while, mostly solo and never checked someone else's stats, But I've only recently come to the forums and it just surprises me sometimes how quick other people can pickup a ship. :)

And grats on keeping Karma in the Bearn, no idea how you have done that.. :cap_haloween:I loose Karma when i play certain Mid tier DD's..:Smile_hiding:

 

Edited by Retarie
  • Funny 1

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×