--------------

Edited 9/16/2021 00:28:15

I like it :D Most Masters actually want solid games rather than easy wins.

+1 Tacky

@Crazy
To be clear, you're saying you prefer it this way despite knowing that others are disadvantaged in order to give you those 'solid games'?

Edited 9/15/2021 19:45:20

While True:
	if playersLeft()==False:
		break
	thisClan=getClanWithTheMostFreeParticipants()
	if randint()%2 == 1:
		if checkIfClanWithBiggerRatingHasFreeParticipant(thisClan) ==True:
			pair(thisClan, clanWithClosestBiggerRating(thisClan) #pairs 2 random participants from these clans
		else:
			if checkIfClanWithLowerRatingHasFreeParticipant(thisClan) ==True:
				pair(thisClan, clanWithClosestLowerRating(thisClan) #pairs 2 random participants from these clans
			else:
				giveFreeWin(thisClan)
				break
	else:
		if checkIfClanWithLowerRatingHasFreeParticipant(thisClan) ==True:
			pair(thisClan, clanWithClosestLowerRating(thisClan) #pairs 2 random participants from these clans
		else:
			if checkIfClanWithBiggerRatingHasFreeParticipant(thisClan) ==True:
				pair(thisClan, clanWithClosestBiggerRating(thisClan) #pairs 2 random participants from these clans
			else:
				giveFreeWin(thisClan)
				break		

Time complexity: O(n^2)

Edited 9/15/2021 17:24:06

+1 Tacky

Yes. The matchmaking system is trash.

our Lord fizzer, hear our prayer

I hope fizzer does something

Helping new clans with Clan Wars? Fizzer does it all the time! Take a look at Harmony for example; TSFH was disbanded and Fizzer allowed players from TSFH to play for Harmony the same season. He even transferred their clan rating, so they could be matched with relevant opponents. He will surely listen to you too.

Edited 9/15/2021 17:43:49

+1

For anyone that doesn't know ... καλλιστηι is being facetious. None of that happened. Anyone that played in TSFH could not play for the remainder of S5 in any clan, Harmony included. No ratings were transferred. Had to rebuild the clan spirit (TSFHell) entirely in a new clan home (TSFHarmony).

- - - - -

Just saw Jacob's post below. I'll just tack on an LMAO here instead of reposting.

Yes you're probably right. He did comment that it could take 10+ months to stabilize and become truly competitive. So ... yea, season 9.

Edited 9/15/2021 18:01:28

i have a feeling this will fall into the category of "lets see how it plays out"

so dont worry yolo members, you will get plenty of games in season 9~

I was going to write a more "charitable" post about how the current matchmaking algorithm was only designed to work with more "organic" clan growth. But now that I think about it, those concerns are completely irrelevant.

How is a low ranked clan supposed to improve if they can't get games? Ours is an extreme case, but it highlights a problem that would befall any low ranked clan with more than ~5 players not willing to jump ship.

Honest question: What would be the drawbacks of a matchmaking system that prioritized maximizing games, then pairing people with comparable skill levels?

Krinid should save your rating dw

Does being a jerk come naturally Lox or have you practiced it extensively?

Honest question: What would be the drawbacks of a matchmaking system that prioritized maximizing games, then pairing people with comparable skill levels?

Those two aren't mutually exclusive. Balthromaws algorithm looks pretty good to me and I also like the part with randomly chosing whether to look for a stronger or for a weaker clan. The algorithm should be sufficient so that at max 1 player does not get a game, it pairs equally rated players and it neither gives an inherent advantage nor disadvantage for big stack clans.

The algorithm is a bit weaker in pairing equally ranked clans than the current one. Let's say we have the players A > B > C > D in terms of clan skill. The algorithm from Balthromaw can then pair B with C and then it needs to pair A with D.

Edited 9/15/2021 18:30:13

I think it is even possible to get very good algorithms with a O(n) complexity. With some proper administration, Balthromaws algorithm would have a time complexity of O(n).

The algorithm is a lot better than that we have now, but there still some problems. For example, let clan A and C have 2 players and B and D both 1. If they have the ratings:

A - 400
B - 250
C - 200
D - 100

Then a possibility is that we get the match-ups A&C, A&D, B&C while A&B, A&C, C&D would be optimal (That is, eveyrone gets a game, and the max rating difference is minimized.

I would propose the following algorithm:
Initial pairings are according to the current algorithm. For every two unmatched players, unmatch the lowest rated match without players from the clan of the two players. Then pair player from each of these unmatched matches with a player that initially had no match. This is also O(n), but is slightly easier to get in this complexity.

An exact approach would be to use a variant of the Hungarian algorithm, but that has O(n^3) complexity and is likely a little overkill for this situation.

+1 Tac(ky)tical

+1

Another dumpster fire of a matchmaking effort just now. 8 teams were queued for strategic 2v2: 4 of them were yolo. We got no matches…just a free win. Boring.