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

So 40% Max Bottom tier is it? Hapa, care to comment?

50 comments in this topic

Recommended Posts

999
[X-PN]
Members
1,578 posts
12,164 battles

Ever since I did a 100 something battle log with Mainz, I noticed bottom tier battles exceeding 40%.  At that time I learned from Hapa that it was strictly based on that last 20 battles of that tier, so in other words, you can't have more than 8 in 20 battles...allegedly. 

Out of curiosity I started tracking my T6 battles.  Different ships, but since Hapa never mentioned how divisions might influence things, none of my battles were in divs.  Also, no waiting times of more than a minute happened so I assume so relaxed MM rules.  I only am up to 31 but results are already "surprising".  (Also my, 100 something Mainz battles, showed similar results, but I wanted a current log that I knew was right).

BTW - I was diligent in making sure this was accurately logged.

image.png.666bfa381a2f5bd82ea0e2317834b512.png

As you can see, the 40% percentage limit appears to have some issues.  Before I say WG's programming is wrong or they are misleading us, I would like Hapa to please explain what could be going on.

At first, I thought maybe it was a greater than or greater than or equal to math issue in the coding where at the 40% limit, one more bottom tier battle was allowed... but the 23rd and 24th battle still shouldn't have been allowed to be bottom tier.  So, I'm at a loss...

Hapa, care to comment?

 

 

Edited by YouSatInGum
  • Boring 4

Share this post


Link to post
Share on other sites
1,150
[DRFTR]
Beta Testers
3,919 posts

divs break it, queue dumps break it...

Share this post


Link to post
Share on other sites
3,062
[ARS]
Beta Testers
5,753 posts
6,003 battles

If you are playing a Tier VI ship and get in a match with only Tier VI ships, how are you counting it?  How about a match with only Tier VI and VII ships?

I also assume that you have long since played through the initial 20 match calibration period and those games are not included in the dataset.

  • Cool 1

Share this post


Link to post
Share on other sites
3,331
[SALVO]
Members
6,954 posts
5,510 battles

I understand it as a trend towards 40%, with punctual variations being possible. Some sort of corrective mechanism that kicks in once you are outside the fence, not a fence by itself. 

  • Funny 1

Share this post


Link to post
Share on other sites
4,263
[-K-]
[-K-]
Members
5,726 posts
21,805 battles
6 minutes ago, Helstrem said:

If you are playing a Tier VI ship and get in a match with only Tier VI ships, how are you counting it?  How about a match with only Tier VI and VII ships?

I suspect this is the most likely explanation, unless you've already account for it. I think the rule internally reads "T6 ships will only be placed in games with T8s <=40% of the time". So even if you're bottom tier in a T6-7 game, it counts as a non-T8 game.

Share this post


Link to post
Share on other sites
3,062
[ARS]
Beta Testers
5,753 posts
6,003 battles
1 minute ago, Edgecase said:

I suspect this is the most likely explanation, unless you've already account for it. I think the rule internally reads "T6 ships will only be placed in games with T8s <=40% of the time". So even if you're bottom tier in a T6-7 game, it counts as a non-T8 game.

Being a Tier VI in a match with only Tier VIs and VIIs counts as mid tier.  A Tier VI in an all Tier VI match is top tier.  Bottom Tier is only when there are ships two Tiers above you in the match.  Tier IX cannot be bottom Tier.

Edited by Helstrem

Share this post


Link to post
Share on other sites
1,399
[WOLFC]
Members
2,560 posts
11,041 battles
7 minutes ago, Edgecase said:

I suspect this is the most likely explanation, unless you've already account for it. I think the rule internally reads "T6 ships will only be placed in games with T8s <=40% of the time". So even if you're bottom tier in a T6-7 game, it counts as a non-T8 game.

I suspect this as well. I always understood it as bottom tier = +2 MM. Also, the majority of the abnormalities are 45%, which is 9/20 instead of 8/20, so I suspect the aforementioned ambiguity of what “bottom tier” means combined with queue dumps/other players in divisions could easily explain the discrepancy.

Share this post


Link to post
Share on other sites
4,263
[-K-]
[-K-]
Members
5,726 posts
21,805 battles
7 minutes ago, Nevermore135 said:

I suspect this as well. I always understood it as bottom tier = +2 MM. Also, the majority of the abnormalities are 45%, which is 9/20 instead of 8/20, so I suspect the aforementioned ambiguity of what “bottom tier” means combined with queue dumps/other players in divisions could easily explain the discrepancy.

FWIW, I went back to check the source directly, and noted that the official wording is "with ships two tiers higher".

Quote

Example: for Tier VIII ships, there's a maximum limit of 40% of battles with ships two tiers higher. If you played Tier VIII ships in your 20 most recent battles, a maximum of eight battles would have included Tier X ships.

That means any 2-tier or single-tier games automatically won't count.

However, the OP says that he's already not counting those as "bottom", so the result should be the same.

Correction: It wasn't the OP, maybe we found the reason?

17 minutes ago, Helstrem said:

Being a Tier VI in a match with only Tier VIs and VIIs counts as mid tier.  A Tier VI in an all Tier VI match is top tier.  Bottom Tier is only when there are ships two Tiers above you in the match.  Tier IX cannot be bottom Tier.

 

Edited by Edgecase

Share this post


Link to post
Share on other sites
999
[X-PN]
Members
1,578 posts
12,164 battles
12 minutes ago, Edgecase said:

I suspect this is the most likely explanation, unless you've already account for it. I think the rule internally reads "T6 ships will only be placed in games with T8s <=40% of the time". So even if you're bottom tier in a T6-7 game, it counts as a non-T8 game.

All T6 is rare and didn't happen in this set.  But if it did, I would do as I did with my Mainz data set and tag it accordingly and not count it as bottom tier.

I think T6 and T7 only games are even more rare and probably only happen in a queue dump. 

None the games had me waiting long enough for a dump, but if a dump happens with you waiting a normal time period so you don't even realize it.... well that sort of defeats the purpose of having MM rules if they are relaxed all the time without you even knowing about it.

Share this post


Link to post
Share on other sites
999
[X-PN]
Members
1,578 posts
12,164 battles
18 minutes ago, Helstrem said:

Being a Tier VI in a match with only Tier VIs and VIIs counts as mid tier.  A Tier VI in an all Tier VI match is top tier.  Bottom Tier is only when there are ships two Tiers above you in the match.  Tier IX cannot be bottom Tier.

Interestingly enough, I suspect T9 is the best tier for being top tier most often, and part of the reason why T8 and T7 are now the worst... also T9 seems to get less CVs too.  No numbers to back it up but I'd still bet my lunch money with anyone that it is the case.

Edited by YouSatInGum

Share this post


Link to post
Share on other sites
4,327
[CNO]
[CNO]
Members
6,341 posts
19,353 battles

I've done this several times out to over 100 games.  On every test, I've discovered the 40% rule simple DOES NOT WORK as described.  

Even before they announced the 40% rule, I simply accepted being bottom tier as part of the random nature of match making.  And I don't really mind it, as I tend to play well when placed in those positions.

After they announced the rule, I figured I'd check it out just to see it if worked.  As noted above and several of my previous posts since they initiated the new rule, it DOES NOT WORK!!!  VERIFIED.  Not speculation.  Ran the numbers.  Multiple times.  It doesn't work.

It doesn't really matter to me from a play standpoint, because it has little effect on how I play or what tiers I play.  I just treat it as status quo, and press on.

But I can confirm to OP that what they are seeing is not abnormal.  It's not as WG has described...which I concluded a long while back.  Go with what you see and observe (keeping good records).  That's the reality.  And the reality is bottom tier is NOT protected by the 40% rule as described.  

Edited by Soshi_Sone
  • Cool 1
  • Thanks 1

Share this post


Link to post
Share on other sites
3,331
[SALVO]
Members
6,954 posts
5,510 battles

You guys are assuming it is a hard rule for <=40%, in order for that to happen, the code needs to be predictive and enforce action (most likely restricting you from the appropriate queue pools) before you get over the fence (40%). That would be more complicated than having it a soft rule that restrict you from determined queue pools once you go over the fence (40%) so you don't need the predictive calculation, just keeping track of a variable. Good code is always the simplest possible, I bet my lunch the 40% is a soft rule.

Share this post


Link to post
Share on other sites
1,150
[DRFTR]
Beta Testers
3,919 posts

i tracked the numbers for awhile last yr and they were working out... but i believe i am hearing things have been tweaked since

  • Cool 1

Share this post


Link to post
Share on other sites
7,068
[WORX]
Members
12,638 posts
19,907 battles
1 hour ago, YouSatInGum said:

Ever since I did a 100 something battle log with Mainz, I noticed bottom tier battles exceeding 40%.  At that time I learned from Hapa that it was strictly based on that last 20 battles of that tier, so in other words, you can't have more than 8 in 20 battles...allegedly. 

Out of curiosity I started tracking my T6 battles.  Different ships, but since Hapa never mentioned how divisions might influence things, none of my battles were in divs.  Also, no waiting times of more than a minute happened so I assume so relaxed MM rules.  I only am up to 31 but results are already "surprising".  (Also my, 100 something Mainz battles, showed similar results, but I wanted a current log that I knew was right).

BTW - I was diligent in making sure this was accurately logged.

image.png.666bfa381a2f5bd82ea0e2317834b512.png

As you can see, the 40% percentage limit appears to have some issues.  Before I say WG's programming is wrong or they are misleading us, I would like Hapa to please explain what could be going on.

At first, I thought maybe it was a greater than or greater than or equal to math issue in the coding where at the 40% limit, one more bottom tier battle was allowed... but the 23rd and 24th battle still shouldn't have been allowed to be bottom tier.  So, I'm at a loss...

Hapa, care to comment?

 

 

Few things...

  • Top, mid and low descriptors are clearly not defined properly.
    • Even if they were according to you... WG might have a different definition for those variables... If this is the case... Your project already has a high percentage of deviation (which we do not want).
  • You do have other factors that can push/skew the numbers...
    • Mainly, DIV and queue drops...
  • Other factors may apply such as
    • Time of day you played
      • Was it consistent or varied?

If it was me... I would designate a time slot... Then, only count matches that fit a certain (above) criteria from that time slot.

I suspect, there's a trigger mechnism.. When triggered, MM compensate... If this is the case... It does so for every single player in queue...

No wounder queue times are high...

 

Edited by Navalpride33

Share this post


Link to post
Share on other sites
1,126
[CO-OP]
Members
2,822 posts
28,397 battles
20 minutes ago, YouSatInGum said:

None the games had me waiting long enough for a dump, but if a dump happens with you waiting a normal time period so you don't even realize it.... well that sort of defeats the purpose of having MM rules if they are relaxed all the time without you even knowing about it.

FWIW, queue dumps occur when someone is waiting an extended period of time.  You might only be in the queue for 30 seconds, but if someone else has been waiting for several minutes, MM could pull you in to form a battle.

Share this post


Link to post
Share on other sites
1,150
[DRFTR]
Beta Testers
3,919 posts

Here's numbers for all the T7 random matches i played  over 3 months last yr...     mainly hiliting there is an app for this..

Using mxstats it can pull this info pretty quickly 

image.png.cf7d13096f4c8ab1e4b5c73e8ca02b9c.png

 

Share this post


Link to post
Share on other sites
8,689
[GWG]
[GWG]
Alpha Tester
27,876 posts
14,882 battles

@YouSatInGum, that is a target over time and not a hard limit over short periods and bottom tier means bottom tier in a three tier match. Divisions should have little impact except for "fail divisions" and that is self inflicted.

Share this post


Link to post
Share on other sites
7,870
[WG]
Administrator, Developers, Community Department, WG Staff, In AlfaTesters
4,574 posts
15,079 battles
2 hours ago, YouSatInGum said:

Ever since I did a 100 something battle log with Mainz, I noticed bottom tier battles exceeding 40%.  At that time I learned from Hapa that it was strictly based on that last 20 battles of that tier, so in other words, you can't have more than 8 in 20 battles...allegedly. 

Out of curiosity I started tracking my T6 battles.  Different ships, but since Hapa never mentioned how divisions might influence things, none of my battles were in divs.  Also, no waiting times of more than a minute happened so I assume so relaxed MM rules.  I only am up to 31 but results are already "surprising".  (Also my, 100 something Mainz battles, showed similar results, but I wanted a current log that I knew was right).

BTW - I was diligent in making sure this was accurately logged.

image.png.666bfa381a2f5bd82ea0e2317834b512.png

As you can see, the 40% percentage limit appears to have some issues.  Before I say WG's programming is wrong or they are misleading us, I would like Hapa to please explain what could be going on.

At first, I thought maybe it was a greater than or greater than or equal to math issue in the coding where at the 40% limit, one more bottom tier battle was allowed... but the 23rd and 24th battle still shouldn't have been allowed to be bottom tier.  So, I'm at a loss...

Hapa, care to comment?

 

 

Still need A LOT more information.

Like what time of day was this? Is it during peak server times? Is it in the AM, evening? What class ship were you playing? When you dropping into queue if you were in a DD, where there 40 DDs in queue and only 4 BBs? Or 1 DD and 30 BBs? All these things play in to MM trying to build a match, because we have it set to try to make the class lay outs of the teams the same. Did you play ONLY tier VIs and then record that? Or did you do Tier IV then V, then VI then VII and only record the VIs?

Beyond knowing all those variables we have always said MM will try to keep uptiering to no more than 40% not it WILL do it, because there are A LOT of variables that go into MM and if you stay in queue for too long, MM will actually override its own rules.

When matching teams, the algorithm also takes your battle history into account, i.e. the number of battles you've played with ships of the same tier as yours and lower, with ships one or two tiers higher, and where all ships were of the same tier. We also collect separate statistics for each battle type and ship tier you play. After a short calibration period, the matchmaker will ensure that the percentage of battles with ships of the specified tier doesn't exceed the set limits, so as to prevent long series of battles fighting against ships of higher tiers. When a player joins the queue for their next battle, the algorithm takes the player's previous 20 battles into account when assembling their team, thus ensuring that the set limitations are met.

That being said, if you're in queue long enough to override the rules you could be uptiered quite often.

-Hapa

 

Share this post


Link to post
Share on other sites
4,642
[RLGN]
Members
15,645 posts
27,371 battles
2 minutes ago, Hapa_Fodder said:

That being said, if you're in queue long enough to override the rules you could be uptiered quite often.

-Hapa

Hapa,

I’ve always also presumed it’s less about a specific tier, (like 20 different T6 ships,) and more, (what OP and others seem to indicate,) about specific ships.

So someone should track 20, or a 100, or more games in Farragut, or Gaede, or Normandie or Izmail or whatever to see if the 40% limitation is actually ‘working.’

Is that right?

Share this post


Link to post
Share on other sites
7,870
[WG]
Administrator, Developers, Community Department, WG Staff, In AlfaTesters
4,574 posts
15,079 battles
4 minutes ago, Estimated_Prophet said:

Hapa,

I’ve always also presumed it’s less about a specific tier, (like 20 different T6 ships,) and more, (what OP and others seem to indicate,) about specific ships.

So someone should track 20, or a 100, or more games in Farragut, or Gaede, or Normandie or Izmail or whatever to see if the 40% limitation is actually ‘working.’

Is that right?

I mean I suppose if you played ONLY Izmail or whatever for 100 battles you'd know better, or the test would be better, but the problem is that there still are so many variables, its never going to be exact.

Like case in point, I've played on the weekend for 4 hours, not been uptiered ONCE, then 2 hours on a Monday evening and had EVERY battle be uptiered....

So maybe if you ONLY played a specific ship at a specific time of day every day... but still, it is NOT exact.

-Hapa

Share this post


Link to post
Share on other sites
4,642
[RLGN]
Members
15,645 posts
27,371 battles
2 minutes ago, Hapa_Fodder said:

I mean I suppose if you played ONLY Izmail or whatever for 100 battles you'd know better, or the test would be better, but the problem is that there still are so many variables, its never going to be exact.

Like case in point, I've played on the weekend for 4 hours, not been uptiered ONCE, then 2 hours on a Monday evening and had EVERY battle be uptiered....

So maybe if you ONLY played a specific ship at a specific time of day every day... but still, it is NOT exact.

-Hapa

Didn’t mean ‘exact’ so much as just that the restriction(?) was tracked more by individual ships, than by tiers as a whole.

Share this post


Link to post
Share on other sites
2,362
[PVE]
Members
7,858 posts
23,833 battles
1 hour ago, Navalpride33 said:

Few things...

  • Top, mid and low descriptors are clearly not defined properly.
    • Even if they were according to you... WG might have a different definition for those variables.

 

In the article at the time the mechanic was implemented WG clearly defined bottom tier as uptiered by 2 tiers & in a battle w/only 2 tiers there is only top & middle tier ships in that battle.

W/all same tier ships everybody is top tier.

The highest tier in the battle (even if it is only 1 ship) defines the tier of the battle.

Share this post


Link to post
Share on other sites
7,870
[WG]
Administrator, Developers, Community Department, WG Staff, In AlfaTesters
4,574 posts
15,079 battles
1 minute ago, Estimated_Prophet said:

Didn’t mean ‘exact’ so much as just that the restriction(?) was tracked more by individual ships, than by tiers as a whole.

Yes it is by ship, not just tier. Specifically because of the limitations of how many X, Y and Zs you can have in each battle.

-Hapa

  • Cool 1

Share this post


Link to post
Share on other sites
287
[GLF]
Members
1,150 posts
18,139 battles

I'm convinced its all placebo, and the queues are the same as they always were. 

It just served the purpose to stop everyone from complaining while they instead collect data & do math and statistics calcs.   :Smile_glasses:

 

 

Share this post


Link to post
Share on other sites
971
[4HIM]
Beta Testers
2,251 posts
16,107 battles
25 minutes ago, Hapa_Fodder said:

Still need A LOT more information.

Like what time of day was this? Is it during peak server times? Is it in the AM, evening? What class ship were you playing? When you dropping into queue if you were in a DD, where there 40 DDs in queue and only 4 BBs? Or 1 DD and 30 BBs? All these things play in to MM trying to build a match, because we have it set to try to make the class lay outs of the teams the same. Did you play ONLY tier VIs and then record that? Or did you do Tier IV then V, then VI then VII and only record the VIs?

Beyond knowing all those variables we have always said MM will try to keep uptiering to no more than 40% not it WILL do it, because there are A LOT of variables that go into MM and if you stay in queue for too long, MM will actually override its own rules.

When matching teams, the algorithm also takes your battle history into account, i.e. the number of battles you've played with ships of the same tier as yours and lower, with ships one or two tiers higher, and where all ships were of the same tier. We also collect separate statistics for each battle type and ship tier you play. After a short calibration period, the matchmaker will ensure that the percentage of battles with ships of the specified tier doesn't exceed the set limits, so as to prevent long series of battles fighting against ships of higher tiers. When a player joins the queue for their next battle, the algorithm takes the player's previous 20 battles into account when assembling their team, thus ensuring that the set limitations are met.

That being said, if you're in queue long enough to override the rules you could be uptiered quite often.

-Hapa

 

I can attest that as playing from Alaska, I have some late play times, and I find that the later I play the more likely I will be in a mixed tier games.  If I play in the late afternoons which is early or later evening for everyone else, I get more same tier games. (bigger population numbers)  In fact the late games can be really bizarre line ups, and if it is lower tier, I am not surprised to see a few bots manning the ships.

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×