A follow-up post, the problems with league matchmaking and ranking system.

This is solely to promote discussion where people might disagree with me or to find out if it's a common agreement that matchmaking isn't effective at placing teams together.

So to be sure that there isn't any confusion with what points I'm trying to make I'd like to define a few key terms. MMR - Matchmaking Rating, or MMR is a value that determines the skill level of each player. This value is used in matchmaking. Winning increases a player's MMR, while losing decreases it. Players with the highest ratings are listed on the world Leaderboards. rs in competitor-versus-competitor games such as chess. Win Rate - Win Ratio is the The win/loss ratio is a ratio of the total number of winning games to the number of losing games. It does not take into account how much was won or lost simply if they were winners or losers. MMS - just a shortened term for matchmaking system.

Why I don't think MMR is a good system for ranked matchmaking:

• MMR is determined by how well a player has done in the past 20 games.
• It does not take into account the players rank or overall skill level.
• MMR changes after every game. Going only off of MMR can lead players to be placed against opponents that greatly out skill them, or vice versa to be placed vs opponents greatly under their own skill level.
• MMR and rank are completely independent so that a player in silver 1 can play against someone from platinum 5 if their mmr has a great enough difference from where it should be.

The main reason mmr shouldn't be used for matchmaking in a ranked environment is because you should only be playing against players in your division, regardless of skill. This might seem counter-intuitive at first but let me explain. By using recent games to influence the process of drafting the team, you're creating a system where players climbing have to face progressively tougher opponents, and players falling face progressively weaker opponents. This causes a players ranking to normalize by going on long win streaks or lose streaks. This is also why it seems that when you go on a win streak into your promos, your team mates suddenly seem to get much worse. This has to do with the mms giving you weaker players so that your team average is brought down, but the other team being all around better than your own team because the skill level is more evenly balanced. The biggest tell tale sign that this is how the mms is set up is that many players will go on long win streaks or lose streaks. My solution: Only rank should be taken into consideration when drafting ranked teams. Ranked is where you try to climb the ladder, if you have a greater overall skill level than everyone from silver and below, you should be put into gold. You shouldn't go against golds to prove that you belong in that division.

To prove this point we'll go through a few situations, using the old system of using mmr to draft teams, and using only rank to draft teams.

Old system

1. Assuming you have the approximate skill level of gold 4 but the ranking of silver 1. Your mmr is inflated and the game places you against gold 5, 4, and 3 players during your promos. You maintain a 50% winrate in silver 1 but constantly play against players higher rated than you. The system failed because it requires you to consistently beat players your skill level just to get to the division they're at. Your skill level and opponents don't match your current rank because the system directly forces you to play against players your skill level, WITHOUT granting you their rank.
2. Assuming you DON'T have the skill level of a higher rank you maintain a 50% winrate in your current division. This seems ok until you take into consideration players who are better than you stuck at your division that you will occasionally have to play against or alongside. You aren't very much of a presence because these players account for the majority of where the actual game is taking place. Whichever team has a more of these players are more likely to win to no fault of your own. Conclusion, the system has failed twice. Once in letting players advance to the rank of the players they're playing against, and twice where players in the same division ave little control over the game they're playing because of players commonly referred to as "hard stuck."

New system

1. Assuming you don't belong in in the next division, you only get placed vs people in your own division and maintain roughly a 50% winrate. You can't argue against it because you've only played alongside people of similar skill level and you win the same number of games that you lose. Conclusion, the system is successful in preventing a player who doesn't belong in the next division from advancing.
2. Assuming you do belong in the next division, you play against people who are the same rank, but lower skill level, causing your win rate to inflate and you proceed to the next division. Conclusion, the system is successful in promoting a player who belonged in the next division.

Whether you agree or disagree with the specific examples is besides the point, playing progressively tougher opponents without anything to show for it is a failure of the system. Rank alone should draft the teams, in an ideal system. If a player doesn't belong at that rank then they climb or fall. If a player does belong at that rank then they maintain a 50% winrate.

THIS ALSO SOLVES THE ISSUE OF PLAYERS CLAIMING THEY'RE IN "elo hell." Elo hell is when a player believes that their teams prevent them from climbing into a further division. Elo hell definitely exists, but it's really just called league of legends. Every elo is an elo hell because of how the matchmaking system works. Players who are clearly better than the rank they are have to maintain a positive winrate against the division they belong in just to get to that division. Meaning that if a player would maintain a 50% winrate in gold 3, they would have to have a greater than 50% winrate against those players just to get to that rank, which is what doesn't make sense. If players only play with and against peoples in their own division than you can't claim you belong in a higher division because the only people you play against and with are players worse than you (assuming that player actually does deserve to get promoted.)]

So what to do with mmr. Mmr has a purpose but it doesn't belong in the process of drafting team. MMR's home is determine the gain and loss of lp for any given match. If your mmr is greater than your division, then you have a net gain for one win and one loss. Meaning, if I am currently placed in gold 3, but I have the mmr of a gold 1, then I would possibly get +22 lp for a win and -18 lp for a loss. The actual numbers can be changed but I hope you get the concept.

Rank determines the drafted teams, mmr determines the net gain/loss from a win/lose.

What about queueing up with multiple people? This is what the caused the shitshow known as season 6's dynamic queue. The value of any 2 or 3 players queueing up together can't be quantified so you shouldn't change the remaining players. Giving the duo queued players a worse team is unfair to the other players on their team. They didn't queue up with a friend so they shouldn't be punished. But playing against a good duo queue isn't fair because you didn't queue up with a friend so you shouldn't be punished for not doing so. Personally I see 3 solutions to this.

League is a social game. League is a social game but ranked isn't League is a social game, but playing solo is harder than with a friend.

If league is considered a social game then we get season 6 all over again with dynamic queue. A highly abusable system that failed and should never come back. If league is a social game but ranked is NOT considered social then we get a regular solo or duo queue system like season 5 and before. We've experienced both of these systems and nobody is quite happy with the outcome so the final solution is:

League is a social game, but playing solo is much harder than with a friend, and the gain/loss of lp should reflect it. Playing with a trusted friend is much easier to win in league, but currently players receive the same amount of lp for doing what is indisputably less work. My proposed solution is that when duo'd with a friend, your gain is reduced and your losses are increased or normalized to compensate for the ally. Better yet, you would receive bonus lp for going on winstreaks in soloqueue. If you aren't with a friend and you're consistently winning games then you should receive more lp to show for it. (This balances it out making it so that queueing up with a friend grants more consistent wins, but queueing up solo grants more consistent lp, as long as you earn it.)

Unranked players and normal games. Where the new system can't apply the old one would still take place. Players in roughly the same rank would be drafted against each other but if you've never played ranked then only your mmr applies. The only change I would make would be to put a soft floor for ranked players in normals so that they couldn't play against people incredibly worse than them. In summary your ranked mmr is completely independent, but your normals mmr is influenced by the ranked mmr. This is only because I've been in too many games with pre level 30 players that have to play vs platinum + players. (the worst case I ever saw was 2 pre 30's on my team vs 2 diamonds and a masters player on the other side.)

I think that's it. Just curious about anyone else's opinion on the matter.