If MMR were accurate

[Legatus.3608]

> @"Chum Chum.7234" said:

> > @"Legatus.3608" said:

> > > @"Chum Chum.7234" said:

> > > > @"saerni.2584" said:

> > > > Too easy for some to manipulate.

> > > >

> > > > But it does hurt when you lose with 60%+ of your team’s offensive score.

> > >

> > > That is my main complaint. The highest rated player in the match can contribute the most to the match and subsequently lose the most MMR after a loss. The performance means nothing and the system consistently moves higher rated players towards the rating of the players in the queue at the time.

> > >

> > > The 'if you're better than you'll win more matches' argument is invalid as there is usually a superstar on the other team as well trying to carry their poor performers. The only way to win consistenly is to duo queue with the other superstar until your ratings are too high to duo queue anymore...or the other way which shall not be named

> >

> > Well, if everyone on your team was good, you wouldn't have 60% of the stats would you? And if people were taking more of the stats the chances are pretty good they know how to do something useful in the game and contribute to a victory, thus if you don't own all the stats the odds are pretty good you'd win?

> >

> > I don't see anything there that should be surprising to anyone, but if you're basically saying "It sucks I lost when I did all this stuff" well yeah it does but that's life, sometimes you do everything right and things still don't work out in your favor.

>

> Sure, that's life.

>

> Let's use some logic to determine if MMR is an accurate measure of skill and i dividual performance though. Please let me know if any of my step or assumptuons about the system are wrong, as I don't know everything about the ANET system.

>

> 1- Observation- When trying to carry a bad team, you'll end up with more than average stats (greater than 20% of team contribution)

> 2- Logic - Based on logic step 1, we can assume better players tend to generate more stats.

> 3- Assumption - The system doesn't evaluate stats.

> 4- Observation - The system takes more MMR from the highest rated player on the losing side and gives more MMR to the lowest rated player on the winning side.

> 5- Logic - Observation 4 acts to move the highest rated players MMR closer to the bad players that they carried or failed to carry.

> 6- Logic - If you played a couple dozen games with the same bad team, logic step 5 dicates your MMR would move towards the other players and eventually you all would end up with the same or similar MMR

> 7- Logic - The outcome of logic step 6 happens despite the fact that you are by far the best player as known based on logic step 1 and 2.

> 8- Logic - Because logic step 7 is true despite actual skill and individual performance, MMR does not accurate represent individual skill or performance.

 

I like how you laid that out lol.

 

Regardless you kind of rigged it though. Your first 8 steps are true up until the point you described. But as you continue to play:

9 - Observation - Eventually the guys you played with will find themselves on the other team

10 - Logic - Your past allies, now enemies, have higher than normal MMR whilst you have lower than normal MMR

11 - Observation - The system will grant more points for winning or take less points for losing as a result of your MMR being closer than it should

12 - Logic - The result is that your MMR's errors will eventually correct themselves as you play against your former allies

[Chum Chum.7234]

> @"Legatus.3608" said:

> > @"Chum Chum.7234" said:

> > > @"Legatus.3608" said:

> > > > @"Chum Chum.7234" said:

> > > > > @"saerni.2584" said:

> > > > > Too easy for some to manipulate.

> > > > >

> > > > > But it does hurt when you lose with 60%+ of your team’s offensive score.

> > > >

> > > > That is my main complaint. The highest rated player in the match can contribute the most to the match and subsequently lose the most MMR after a loss. The performance means nothing and the system consistently moves higher rated players towards the rating of the players in the queue at the time.

> > > >

> > > > The 'if you're better than you'll win more matches' argument is invalid as there is usually a superstar on the other team as well trying to carry their poor performers. The only way to win consistenly is to duo queue with the other superstar until your ratings are too high to duo queue anymore...or the other way which shall not be named

> > >

> > > Well, if everyone on your team was good, you wouldn't have 60% of the stats would you? And if people were taking more of the stats the chances are pretty good they know how to do something useful in the game and contribute to a victory, thus if you don't own all the stats the odds are pretty good you'd win?

> > >

> > > I don't see anything there that should be surprising to anyone, but if you're basically saying "It sucks I lost when I did all this stuff" well yeah it does but that's life, sometimes you do everything right and things still don't work out in your favor.

> >

> > Sure, that's life.

> >

> > Let's use some logic to determine if MMR is an accurate measure of skill and i dividual performance though. Please let me know if any of my step or assumptuons about the system are wrong, as I don't know everything about the ANET system.

> >

> > 1- Observation- When trying to carry a bad team, you'll end up with more than average stats (greater than 20% of team contribution)

> > 2- Logic - Based on logic step 1, we can assume better players tend to generate more stats.

> > 3- Assumption - The system doesn't evaluate stats.

> > 4- Observation - The system takes more MMR from the highest rated player on the losing side and gives more MMR to the lowest rated player on the winning side.

> > 5- Logic - Observation 4 acts to move the highest rated players MMR closer to the bad players that they carried or failed to carry.

> > 6- Logic - If you played a couple dozen games with the same bad team, logic step 5 dicates your MMR would move towards the other players and eventually you all would end up with the same or similar MMR

> > 7- Logic - The outcome of logic step 6 happens despite the fact that you are by far the best player as known based on logic step 1 and 2.

> > 8- Logic - Because logic step 7 is true despite actual skill and individual performance, MMR does not accurate represent individual skill or performance.

>

> I like how you laid that out lol.

>

> Regardless you kind of rigged it though. Your first 8 steps are true up until the point you described. But as you continue to play:

> 9 - Observation - Eventually the guys you played with will find themselves on the other team

> 10 - Logic - Your past allies, now enemies, have higher than normal MMR whilst you have lower than normal MMR

> 11 - Observation - The system will grant more points for winning or take less points for losing as a result of your MMR being closer than it should

> 12 - Logic - The result is that your MMR's errors will eventually correct themselves as you play against your former allies

 

I love it. I'm glad to meet you.

 

I'd argee with your counterpoints except that

 

13 - Assumption - The system tries to level teams rather than completely shuffling the deck each match.

 

And

 

14 - Logic - The same effect that I described in 5-7 takes place relative to the pool of players in the queue.

 

15- Observation - Playing at 7:00 am EST (when I play) there aren't enough players for random matches, so the same 10 players keep showing up.

 

16- Observation - The system balances the team more or less the same until MMR falls far enough.

 

17- Observation - Each match I get 3 or all 4 of my previous teammates back.

 

18- Observation - The other teams superstar is always still on the opposing side.

 

19 - Logic - The other superstar and I bleed MMR to the lesser skilled players as noted in 5-7.

 

20- Observation - This is reinforced by my MMR surging 200 or more by simply queueing at 7:00pm EST.