League of Legends General Discussion Thread

[VoiceOfUnreason]

Lateral yank, as I think this becomes more discussion than humble brag....

90, 93, 95, 96... ugh. Very frustrating. One day I'll get to Gold V.

I've been watching Senpai's support experiment. (Executive summary: Platinum I summoner SenpaiNoticedMe has rolled himself a smurf, and is climbing the ladder playing exclusively support champions).

He's been logging the results here: http://bit.ly/SenpaiSupportExperient

The thing that has really caught my attention is that he's been absolutely cruising through league points. 55 games since he was originally placed in Silver III, 4 promotion series, and there's nothing in the log that looks like a throttle.

Tells me that I don't really understand how the current rating system really works under the covers.

[v8mark]

I think his MMR was high enough from the successful placement games to send him straight up. I noticed he leapfrogged divisions regularly as well, probably for the same reason (Silver III, Silver I, Gold V, Gold III, Gold I).

[spellman]

cross-moving since this is getting into ELO arguments:

Riot have said time and again that matchmaking is determined by MMR only and has no connection whatsoever to your League standings. And for all I know, MMR works just like the old Elo, except you can not see it anymore. I see no reason to not believe them.

I'm a tad confused why they would decouple how they setup the matchmaking and their league system. Assuming MMR is accurate and decoupled from league ranking, then you should trend towards 50-50 win-loss no matter your League ranking. Which means promotions are pretty much useless since you should trend toward 50-50 wins. Eventually 50% gets you 2-out-of-3 and 3-out-of-5.

I would expect something closer to the following. Use League Ranking as the MMR, and for promotion series seed them as if they were the tier above (or 2).

In theory MMR and League Ranking should trend towards the same point. So why would it be decoupled?

My primary thought is they needed a way to unify over Normal versus Ranked games. But then why not just have 2 rankings, MMR for Normal and the League Ranking for Ranked? Similarly unique ones for TT and Dominion.

Anyone have quotes/links as to their reasoning?

[VoiceOfUnreason]

In theory MMR and League Ranking should trend towards the same point. So why would it be decoupled?

One possible answer: they want people playing ranked, rather than sitting out to protect their rating.

By decoupling the ranking from the match making, you can then choose to pair the ranking with a ratchet mechanism, so that there is no "risk" involved in playing more games, and without being forced to compromise on matchmaking players of equal "estimated current skill" into the same game.

I've collected no data to test this particular hypothesis.

[abuzeus]

With both LCS and Diablo being down, seems I have time to respond to this (Can't blame Riot for widespread internet issues, but sure can blame them for not having a LAN server in EU. Arrrg)

Here is what I know and/or assume: MMR is the old Elo, except you can not see it. I think everybody knows how Elo works. Furthermore, the most important thing for matchmaking is to create a balanced match with as short a queue time as possible. So the matchmaker does not know about the leagues and just tries to make somewhat balanced teams (and puts the stronger team on purple side). So you are correct that you eventually will get about 50-50 odds in your matches, once MMR reaches your true skill.

Then there is the league system. I remember Riot saying that the purpose of the leagues is to have a smaller succession of goals, instead of a number that goes up and down and for most players will largely stay at the same point. Added on that is what spellman said: you can have players permanently keep tiers, e.g., for season end rewards, without matchmaking being affected.

Over the long run and assuming constant player skill, MMR and League standing should converge. If your MMR is close to a tier or division divide, the law of larger numbers does indeed dictate that, eventually, you will get promoted into a new tier. However, if you do not truly improve in skill, you will be stuck in division V of your new tier, or, in case of the promotion being to a new division, just be demoted again.

Also, you being in a promotion series does not have impact on matchmaking (I think). In particular, in most of your promotion games, the rest of the players will not be in a promotion game. I think it would be largely unfeasible to change that, because finding a game where all players are up for promotion may not be easy in a reasonable window of time off-peak for the lesser populated skill tiers. But since some players in your promotion game just play a regular ranked game, it is of course unfair to have the fact that you are in a promotion series have an impact on matchmaking. At least, that is my theory.

Also, as far as I know, normal and ranked MMR are not the same (same goes for Dominion and 3s). I have seen a red post about ranked "splashing" rating on the other modes though, but I am not sure whether that is live, or if they scrapped that idea. I also think there are good reasons for normals and ranked to have separate ratings.

[Jowy]

Wow demotions are pretty brutal.. I was at 12 points before a game, I lost, and got demoted to the earlier division.

[spellman]

Curious on people's thoughts on Nautilus. Seems primarily a tanky ganker jungler meatshield with some nice root/knockup.

How well does he fare top? Feeling on what comps he works well in?

I'm trying to decide on a tanky guy to buy. Or I could actually become competent at Singed I suppose. Or buy Lux. She seems fun.

[v8mark]

Naut is excellent. Really, really strong ganking jungler, who I understand can clear the jungle now as well. I know Bruce plays a lot of Nautilus and consistently raves about him. He has so much CC that he’s useful in almost any situation, and if you get remotely strong you’re this ridiculous unkillable tank capable of zoning two or three people at a time in teamfights.

He works in most comps, but particularly excels at catching people. So a pick comp, with someone like Ashe and a high-damage mid is ideal. But he’s a bit like Cho’Gath in the jungle… ideal because he doesn’t need to be fed to be useful, and offers A LOT of tankiness, utility and CC.

Nautilus top is actually really good against people who don’t know how to play against him. Basically you abuse the fact that your shield is completely OP, because it both shields you from lots of damage AND does damage around you in an AoE. You’ll generally outdamage most bruisers while your shield is up for this reason.

Any smart enemy will disengage as soon as you pop your shield, then re-engage a couple of seconds later when you can’t do anything. But because Nautilus top isn’t really a thing, and because it takes a couple of games to figure out what to do, it’s unusual to see others do this. Being out-traded is a terrible thing though, as you have no prospect of getting any sort of sustain in lane.

In other news, I’m sad about the Lulu Q mana cost nerf. She really didn’t need nerfing… she had something like a 46% win rate as it was, I understand. Time to spam Janna games (she’s the next FotM support, I think)

[Jowy]

My Lulu has a 30% win rate as it is, I could have used a buff on her haha.
Janna is really good if you need a disengage. Also hard counters Leona in lane.

[v8mark]

My Lulu record's pretty good (13-6 or something I think) but overall she wasn't a successful pick, as the statistics showed. She has a relatively high skillcap; you need to be able to focus and think quickly in teamfights in order to be effective with her (in contrast to Sona, Taric, Thresh, for example).