<< Back to Clans Forum   Search

Posts 1 - 30 of 102   1  2  3  4  Next >>   
Clan War matchmaking: 4/8/2021 18:45:35


Sephiroth
Level 60
Report
The problem
Clan War matchmaking, as it is implemented right now, has a crucial flaw that benefits mid-low tier clans, as it's in fact happening now: the rating of these clans would be low enough that their players won't get matched with players from the top rated clans and sometimes it would be even as low as to grant these clans free wins (TSFH, the #1 clans for number of wins in this moment had 11 free wins, and MH #2 had 2) - and if that wasn't bad enough by itself, it adds up the fact that getting a free win doesn't even cause the rating to increase.
Clans with higher matchmaking are powerless in front of this, as the only thing they can do is watch the mid-low tier clans get their (free?) wins while being matched against players from the other clans with high ratings. There's simply no way to interact in a system like this that does everything to privilege low rating clans.

Here's some data about the top 4 clans (ordered by winrate):

Clan                | Wins | Losses | Played | Winrate | Rating | Free Wins | Winrate w/o free wins
--------------------|------|--------|--------|---------|--------|-----------|----------------------
MASTER Clan         |  103 |     61 |    164 |     63% |  255.1 |         0 |          63%
Python              |   72 |     46 |    118 |     61% |  345.8 |         0 |          61%
Two Steps From Hell |  108 |     99 |    207 |     52% |  117.6 |        11 |          49%
M'Hunters           |  106 |    101 |    207 |     51% |  216.9 |         2 |          51%



The simple solution
Since number of wins is the only value that matters towards Clan War victory, that number itself is the value that better expresses performance and, hence, the one that should be used for matchmaking.
With the other artificial rating, the system will always reward clans like TSFH and MH.

Edited 4/9/2021 06:31:04
Clan War matchmaking: 4/8/2021 19:05:41


Ralph 
Level 59
Report
Clan wars was initially launched to be an event for idle players to start playing more warzone classic. Making it into a skill-based competition would counteract that. Besides, there are already plenty of options for competitive play, so no need to change the CW matchmaking imho.

Edited 4/8/2021 19:15:01
Clan War matchmaking: 4/8/2021 19:12:41


Kratt 
Level 61
Report
Hey, watch who you call mid-low tier!
Clan War matchmaking: 4/8/2021 19:15:40


Beep Beep I'm A Jeep 
Level 64
Report
With the other artificial rating, the system will always be exploited by clans like TSFH and MH.


Just playing the game is not the same as exploiting it, so I think that's a kinda rude statement.

Also, the important measurement quite clearly is games played, and TSFH and MH lead those categories, so there's nothing wrong with them being on top. I would even go so far to say that these stats suggest the system works nicely, because Masters and Python are able to make their skill count, too (they score significantly above 50%). It's balanced.

Edited 4/8/2021 19:52:33
Clan War matchmaking: 4/8/2021 19:19:54


Sephiroth
Level 60
Report
the important measurement quite clearly is games played

The only measurement that counts towards victory is the number of wins
Clan War matchmaking: 4/8/2021 19:22:51


Kratt 
Level 61
Report
Not sure you realize how the world actually works but okay.
Eventually most (except the very best and the very worst) clans will have the clan rating that gives them around 50% winrate.

So yes, the amount of games you play is what really counts.
All in all it's a casual participation competition and it should remain that way.
Clan War matchmaking: 4/8/2021 19:41:18

Legolas 
Level 60
Report
Clan War matchmaking: 4/8/2021 23:02:45

Widzisz • apex 
Level 61
Report
It's not really an "exploit" if they cannot control it, is it?

I can't think of an easy way to avoid "free wins", and some clans will get more simply because of their paticipation rate.
Your solution of "not counting free wins" is unfair to them, it's like it's free loss instead of free win.

Matchmaking is also fine, especially with more timeslots and templates in current season.



My only beef is about the free win not giving you any rating points, it should be counted as a win against average opponent I would say.
I don't think this would change much though, most likely the free wins would be distributed more evenly (percentage-wise) among the lowest rated clans, which I believe is more fair.

I'm not smart enough to easily check how the free wins have been distributed so far among ALL the clans, but would definitely like to see that (I mean number of free wins vs games actually played).
Clan War matchmaking: 4/8/2021 23:11:30


joeydavis761 
Level 59
Report
I thought it was based off players are matched with other clans with similar rating and fr there it's based on levels. It made sense that a high rated clan like MASTERS played 2v2 they might get the free win while the other 2, more closely related would face each other which gives the advantage to the higher ranked clans. It's more challenging to win territories so they're just essentially granted a free win.
Clan War matchmaking: 4/8/2021 23:18:02

Plausible Deniability
Level 10
Report
should be counted as a win against average opponent I would say.
+1 This many free wins should not be going to TSFH. They are winning the season. The only clans that should be getting free wins at a consistent and predictable rate are the ones who are near the bottom in terms of skill and need handouts in order to stay motivated.
Clan War matchmaking: 4/9/2021 03:54:02


Kenny • apex 
Level 59
Report
Yeah I think saying that they're exploiting this is not right.

As the system works right now, if Python or Master queue up to any template and there is a possible opponent, we will get that opponent since our rating is so high. TFSH rating isn't high so during timeslots that higher rated clans play and an odd amount of players queue up they're likely to get a free win. TFSH also seems to play lightly every timeslot while Master/Python seem to prefer certain timeslots.

It's not an easy solution - but I think some sort of weight needs to be given to total territories or wins to force TFSH to play instead of a 2-3 territory higher rated clan. It unfortunately removes a lot of the competitiveness of the clan wars concept if highly ranked clans get free wins.

If Master had maintained their low rating at the beginning of season 1 and carried that over to season 2 it'd be us in the position of TFSH right now which I don't think would be fair. Then again, we're just built different so it didn't become a problem.
Clan War matchmaking: 4/9/2021 03:59:12


Norman 
Level 58
Report
With the other artificial rating, the system will always be exploited by clans like TSFH and MH.

We only got 2 free wins. One was an unexpected one which I got as only 2 other players joined that template from Blitz and Python who were both ranked higher at that time. The other free win every clan could have grabbed since only M'Hunters players joined that template in that time slot and nobody else.

However I have already pointed out in the Discord that the strict Elo based matchmaking isn't good and that in fact it could be beneficial for clans in a certain Elo spot to throw games against way higher ranked clans on purpose. It's just that the arguments in the opening post are relatively poor ones to challenge the matchmaking.

Regarding "gaming" the system I have a certain hunch that Python and Masters players tend to avoid each other. Not as in a combined approach but I believe they have both a tendency to play their niches.

Edited 4/9/2021 04:05:21
Clan War matchmaking: 4/9/2021 04:25:16


Kratt 
Level 61
Report
Isn't there a very easy fix for the problem?

First, free wins giving rating, which could equal to beating their own clan, rating wise.
And second, when matchmaking use clan rating x number of territories as the formula to pair up.
Clan War matchmaking: 4/9/2021 04:34:59

DonkeyTeeth
Level 60
Report
Or maybe other clans should motivate more of their members to participate. Sorry that we have relatively good turnout.

I will say that in many timelsots TSFH has an entrant for each template. This increases our odds of free wins as well.
Clan War matchmaking: 4/9/2021 04:41:23


Beren Erchamion 
Level 64
Report
Sephiroth was wrong to use the word "exploit" since I don’t think either MH or TSFH can really do anything different from what they are doing.

However, I don’t see what clan rating has to do with CW. In the ladders it makes sense to match people using their ratings, since that is the metric by which the competition is judged. You want the people competing at the top to play each other, and people in the middle tiers to play against each other.

In CW, rating has absolutely nothing to do with anything. Wins are the only thing that matters. So why wouldn’t matchmaking be based on wins?

Another issue with the existing matchmaking, which I don’t think has been mentioned is that Masters can’t even do anything to "play defense" against TSFH. In the ladders, you could at least hope to play against your main competition and defeat them, but since Masters have a rating so much higher than TSFH, they have very little opportunity to play against them to beat them.
Clan War matchmaking: 4/9/2021 04:54:45


{TLA} LND
Level 59
Report
Yeah bad arguments in the first post, sensible fixes suggested - make free wins earn rating, and factor in number of territories somehow.

And I also think the idea of matchmaking based on wins instead of rating is much better.

Edited 4/9/2021 04:56:49
Clan War matchmaking: 4/9/2021 04:54:52


Edge
Level 63
Report
Obviously you can say, that the matchmaking via clan rating is a problem. Probably in every season the rating could start at 0 again, but that also has it flaws as a poor win rate in the beginning can gain you a significant advantage in the mid game, if you're a good clan, when you get matched up with weaker clans. That was seen last season as well, as we were right at the top from the beginning and got matched for the first week or even 10 days against players from One, when available etc. while Masters started with a rating in the lower top 10 rankings, which was obviously to low for them to get rated accordingly to their normal skill level. They obviously didn't got matched up with the top clans as often as we did in that time period. They did end up the season with the highest rating i think, so they earned their win, but it shows, that the method of starting at 0 again also has it flaws, through i do think it would be more fair to reset the rating for every season, than keeping it forever. In the end it's indeed a participation competition, rather than a strategic competition.

If you're looking at it like that, than TSFH and MH played more games than any of the other two clans. It doesn't matter if they received free wins or not, if any of the other two clans had played as much games or just came close to the amount of games the other had played, they'd be leading by a good margin of wins, since their winrates are higher. That was always obvious to be the case in this competition. So if you want to beat them, you neet to rise your participation. Otherwise just accept, that this comeptition isn't a competition for strategic purposes, but rather casual clan game.

Ranking the clans on the clan page due to the competition, might not be the best, because when i look at it, i think the best clans should be ranked at the top, rather than the ones who just won a participation competition. So if the matchmaking works via clan rating, which never sets back, i think that's a more according stat to use to rank clans, than the final rankings of a participation competition. But i'm also not a friend of an ever changing list and like to have a steady list, but in the end it's Fizzers side and his to decide.

I'm also worried, that the competition with these kind of templates, will not be appealing for that much longer, so i expect a slow participation drop, at least from the strategic clans, as i think few clans even bothered to play the first season and even less bother to play in the 2nd season. I dunno how the real numbers look like. Maybe i'm wrong or it appeals the casual clan player to participate to balance out or even outgrow the loss of strategic players. I just think the strategic scene will lose interest and so i think you'll see the strategic clans drop in the rankings.

Edited 4/9/2021 04:58:50
Clan War matchmaking: 4/9/2021 05:03:48


{TLA} LND
Level 59
Report
Yes more template variety than just MME or SE is needed, if the strategic clans are going to be kept engaged in Clan Wars.

My suggestion is that at least 2 non-MME/SE templates are randomly selected (or chosen based on popularity, whatever works for Fizzer) from Quickmatch for each timeslot; if there are concerns about there not being enough interest in these templates, also implement a QM-style template selection system. Instead of choosing just 1 template out of the 4 to play for that timeslot, give each template tickboxes like in QM, so people select every template they are willing to play - this way you will minimise the number of "leftover" people who get free wins, as well as being able to increase the template variety.
Clan War matchmaking: 4/9/2021 05:24:15


Norman 
Level 58
Report
Here some problems I see with the matchmaking / Elo system:

----------

[20] clan has only 1 player playing for them. He has 4 total games out of which 3 are free wins. The player playing for them is actually an excellent player and probably wants to actually play his games. He also can't really improve his clans Elo if he keeps only getting those free wins.

----------

I believe 2v2 currently does not count towards the Elo but correct me if I'm wrong.

----------

When you insert the Elo of 2 clans for example here http://sandhoefner.github.io/chess.html you get the expected win rates. If I type in Pythons + M'Hunters current Elo I get the result that Python should win 70% of their games against us. Intuitively I assumed that most clans apart from the best and worst corner cases end up with exactly a 50% win rate + some noice. However the problem is that if M'Hunters would only play Python and we win the expected 30% of times against them, neither ours nor their Elo would change. If there weren't other clans "shielding" us from Python and Python would have a way higher participation rate than us, we would play mainly Python and would thus score way below 50%. In such a case our optimal play would be to throw games on purpose so another clan would take the heat. I believe a system where you can make a mathematical argument that the optimal play is to lose on purpose is inherently flawed.

----------

For most clans the expected Clan Wars success is 100% determined by their participation with neither wins nor losses having any effect. My clan mates get mad at me when I tell them that it's completely irrelevant whether they win or lose. I can't really be that happy about my own wins knowing that I would have helped my clan equally if I had just surrendered on picks.

----------

I believe there is a pretty easy solution which addresses every problem, namely:
- free wins
- clans are stuck winning 50%
- currently more theoretical argument that clans should drop games on purpose
- we want to reward both participation and skill
- we still want clans to play against their peer clans in terms of skill so both parties get a fun game

My solution would be to use a weighted random matchmaking. For example:
1) randomly pick a player
3) randomly pick 5 (or whatever) possible opponents
3) choose the opponent out of the pool from above closest to that player in terms of clan Elo
4) repeat

Edited 4/9/2021 05:53:16
Clan War matchmaking: 4/9/2021 05:42:51


SkekUng
Level 58
Report
Great suggestion LND - if the community could vote in a template or 2, or a random shuffle of QM templates, then everyone would be happy. Getting:
a) Variety;
b) Strategic templates that the community WANTS and LIKES (As opposed to some MME templates which I know a lot of people aren't fond of);
and c) Subsequently keeping clans and people involved and motivated, waking up every morning thinking, "What templates will it be today?"

Also like the 'checkmark' idea, would definitely reduce free wins, which I'm sure most clans and people would be happy about.
Clan War matchmaking: 4/9/2021 06:11:53


alexclusive 
Level 64
Report
I think that pretty much all suggestions that were made here so far would notably improve the CW experience.
Clan War matchmaking: 4/9/2021 06:21:05

Legolas 
Level 60
Report
@Norman

Isn't all fair matchmaking systems biased towards finding matches that try to give 50% win rate for both teams? I think getting 50% win rate is a sign that the system is working.

Edited 4/9/2021 06:21:16
Clan War matchmaking: 4/9/2021 06:35:50


Sephiroth
Level 60
Report
Yeah 'exploit' wasn't exactly the word I was looking for, but you got my point
Clan War matchmaking: 4/9/2021 07:09:48


Norman 
Level 58
Report
@Legolas: In the 1v1 ladder a 50% win rate is great. If you improve as a player you still only win 50% but you get rewarded by a higher rating. Surrendering on picks certainly isn't a winning strategy there. However with the currently inflated Python rating I have a hunch that I would help my clan currently more if I just surrendered all games against Python on picks. Of course I'm not doing that but winning against a clan with a way higher Elo feels wrong to me, given that we then get "punished" with a big Elo gain from which we will need maybe 3 losses to recover. The really valuable games are the ones against low ranked clans where we lose a lot of Elo when losing and gain next to no Elo when winning.

So in other words:
- In the 1v1 ladder you celebrate your victory against that number 1 record holder.
- In Clan Wars you want to avoid that top Elo clan and if you run into them you might actually prefer a loss.

Edited 4/9/2021 07:10:38
Clan War matchmaking: 4/9/2021 07:14:43


Sephiroth
Level 60
Report
Apparently exploit was not the wrong word after all
Clan War matchmaking: 4/9/2021 07:15:58


SkekUng
Level 58
Report
strategy
Clan War matchmaking: 4/9/2021 07:35:31


Norman 
Level 58
Report
No "strategy" or "exploits" but merely stating the fact that when you need 3 losses to "recover" your Elo from a single win against Python, the win feels much like a Pyrrhic victory. I got a honest loss against Masters today while trying my best and I never saw anyone losing games on purpose so far.

IMO the the basic idea should be very simple:
- If you win against a clan at exactly your Elo, you need 1 loss against an equally ranked clan to make up for the Elo gain.
- If you win against a clan above your Elo, you need more than 1 loss against an equaly ranked clan to make up for the Elo gain.
- If you win against a clan below your Elo, you need less than 1 loss against an equally ranked clan to make up for the Elo gain.

Edited 4/9/2021 07:53:41
Clan War matchmaking: 4/9/2021 07:40:04


Master Meldarion 
Level 63
Report
Many useful things are already said;

To me the easy fix is this:

- Start at the top of the number of wins, meaning first TSFH, MASTER, MH, Python, to "reserve" slots for those players, go through everything in the algorithm, and give the clan with the lowest number of wins a free win in the case there needs to be one. After that go through the algorithm as it is, still basing the essential match-ups on clan rating, so there's still a +/- 50% winrate.

This helps with another problem as well, namely the one TLA had last season; by going from top to bottom without regard for amount of players in a clan, they often had a lot of people that all received no match-up, because they were the lowest clan in warrating. By going through it from the top, it should "reserve" slots for clans with more players and low war rating.

It also means that free wins counting towards war rating no longer matters, as the war rating isn't the decisive factor in getting those free wins.
Clan War matchmaking: 4/9/2021 07:54:43


Sephiroth
Level 60
Report
Yeah whatever we want to call it, the fact is that the current system makes an approach based on keeping a low rating the winning strategy and is allowing for this.

The goal would be a matchmaking that matches clans at the top with other clans at the top, and clans at the bottom with other clans at the bottom, and since win condition === number of wins it comes trivial that it should be the only criterion taken into account for a matchmaking to be as fair as it can get. Let clans with the most wins fight against each other.
Clan War matchmaking: 4/9/2021 10:09:14


Johnny Silverhand 
Level 57
Report
The other problem with clan rating, is it might make clans have close to 50% winrate, but it doesn't make players have close to 50% winrate unless they are close to the average skill of their clan. Anyone who is below average of their clan becomes a punching bag. The reverse is also an issue. Ursus is the best player in TSFH by a mile. But because his clan has low rating he can't get engaging games where he is challenged by his opponents.
Posts 1 - 30 of 102   1  2  3  4  Next >>