49 replies to this topic

[Gnume]

Vlad Ward, on 22 August 2015 - 01:46 AM, said:

Why would it be anything but a normal distribution centered at tier 3 with delineations at each standard deviation?

Well, it depends on how the implement their Tiers.
Are they unlimited in the amount of players that can be assigned to a certain Tier or are they going to be limited in that only X% of players are ever in any particular Tier based on the top PSR scores at any given time?

Are the Tiers going to have Hard Number breaking point? Once you reach that number you move into/out of a Tier?

If there are no limits to how many can be in any given Tier, then a Standard Distribution would likely not apply since theoretically, everyone could be actually end up in a single tier (not very likely).

Given the way you can change your PSR ... perform well or extremely poor (but not rewarded nor punished for "average" play), then at some point, Tier 1 players will likely normalize and remain within their tier always where those in lower tiers can still keep raising their PSR by performing well. Tier 1 players will reach a point where they find it harder and harder to have "High Performing Games" to raise their PSR while at the same time, remaining "average" within their tier and not lowering their PSR.

Tier 5 should be fairly small as new players become accustomed to the game and begin performing better...and given that new players start out at Tier 4, they will likely only drop to tier 5 if initially facing lots of matches against higher tiers while they are still learning.

So, I think you will see the majority of the player base in the Tier 2 and 3 with a fair amount in 4 that transit between 3 and 5 for a while.

No actual math or statistics was used in this post Just my "gut" take on it

Edited by CaptRosha, 25 August 2015 - 12:09 PM.

[AlphaToaster]

StillRadioactive, on 22 August 2015 - 01:44 AM, said:

It should be:
Tier 1 can face off against Tiers 1 and 2.
Tier 2 can face off against Tiers 1, 2, and 3.
Tier 3 can face off against Tiers 2, 3, and 4.
Tier 4 can face off against Tiers 3, 4 and 5.
Tier 5 can face off against Tiers 4 and 5.

Absolutely. This is closer to a skill distribution that would match players within their skills more closely. However, due to population, Tier 1 and 5 would have really long wait times.

It's a catch 22. That balance between acceptable wait times and quality matches.

[Fate 6]

Kira_Onime, on 21 August 2015 - 05:59 PM, said:

You're more likely to see 80-85% of the player within Tier 5 through 3.
T2 would be the good players while T1 should be pretty much comp level players.

^
And if it's 5% at the top it's the actually good comp players. Plenty of lower comp players are vastly outskilled by the better comp players. I would view about a third of comp players as T2

[Eglar]

FupDup, on 21 August 2015 - 05:32 PM, said:

It probably looks like this:

just let me add:

[Hit the Deck]

According to these two threads, the distribution seems to be normal:

• http://mwomercs.com/...ts-for-science/
  
• http://mwomercs.com/...n-tier-seid-ihr

The first one in the GD could be more biased (right sided) because there's no poll. The second one has a poll.

Tier mechanics aside, I am pretty pleased because so far, it creates a normal distribution!

Edited by Hit the Deck, 25 September 2015 - 08:41 PM.

[INKBALL]

My thinking: Hi guys, i just generalize there...

But as PGI thinks: ''eventually everyone will bubble up to Tier 2 or Tier 1''
http://mwomercs.com/...-tiers-and-psr/

I have to go up front and say that i think im not alone to not want top 10-15% players to play with average/under average players. I'm not speaking for the top 10-15%, and i'm not speaking for the average/underaverage players... but i think that both sides dont wanna see series of steamrolling games... (or just ****/farn noobs (or get insta kill showing his face over a cliff))

So here is what i dream about: Elo system based on Individual match score (a match score who is more evenly balanced for every class weights ( closer to a dmg/3 and more points for components destructions/flanking/scouting/spotings/UAVing/etc. and still a minus points for loses).So top players could be really ranked top as they play more Meta/Cautious/Strategically.

I'd make it work with a standard deviation on match score, lets say -2 pts for bottom 20%,match score, -1 for bottom 20% to 30%, 0 point bottom to top 30% match score, +1 for top 20%-30% and finally +2pts for to 20% match score.

Let's say you put it 7 tiers, and still working with the 2 brackets range. Then Top 14% players would at worst engage over average players (T3, who are 43th to 28th percentile.).

And of course, game still balanced (with as much as possible ) equal points on both sides.


Am-I so wrong? I think its a damn descent point of view, no ?

Edited by INKBALL, 25 September 2015 - 08:54 PM.

[Hit the Deck]

INKBALL, on 25 September 2015 - 08:53 PM, said:

...
But as PGI thinks: ''eventually everyone will bubble up to Tier 2 or Tier 1''
http://mwomercs.com/...-tiers-and-psr/

You must not cherry pick and need to quote the whole thing so people don't get misinformed. Paul said it provided that there's no influx of new players (a closed system). Our Tiering system has a fixed value for each tier so it stands to reason that when people get better and better over time, the majority will be on the top because their skill/experience are equally the same (assessed by the system).

INKBALL, on 25 September 2015 - 08:53 PM, said:

So here is what i dream about: Elo system based on Individual match score (a match score who is more evenly balanced for every class weights...

I don't think that you mean "ELO" but just a scoring system based on individual performance (irrelevant of win/loss). Am I right?

[sycocys]

Davers, on 21 August 2015 - 06:02 PM, said:

With Elo you have the player base divided into a parabola



The highest point would be players who rated as average, with the numbers falling off at the extremes.

But Tiers makes it sound like a pyramid, such as everyone's favorite



Which would imply that there would be more Tier 5 players than Tier 4, more 4 than 3, and so on until you reached the smallest number of players in Tier 1.

Which one are they using?

It sounds like they are using the bottom one with the fever dream that the pyramid is actually upside down.

[Hit the Deck]

sycocys, on 25 September 2015 - 09:49 PM, said:

It sounds like they are using the bottom one with the fever dream that the pyramid is actually upside down.

It sounds like you didn't read the thread?

[Tarogato]

I don't know what the ideal distribution *should* be, but I'm starting to get an idea of what the distribution actually is. Thread: http://mwomercs.com/...ts-for-science/


130 samples.

Tier 1: 23 ... 18%
Tier 2: 35 ... 27%
Tier 3: 45 ... 34%
Tier 4: 16 ... 12%
Tier 5: 11 ... 8%



If ya'll could help me out to get more samples, that'd be great.