Skill Ranking and Matchmaking Enhancements

[nordag]

Oh and about your whining : I never really played quake until ql released and all i did was joining the highest tier servers possible. Now look at my stats.

... well, i dont remember you being any good at all. actually you are one of the worst railwh0res i've ever seen. Your teamplay sucks, no support, 2much +b ... e.g. on ca1 you keep camping on the upper ledge (tele-exit left), railing from distance, while your mates get their asses pwnd at top ...
play lame-ass-duel or insta-stuff, but stop talking about beeing high-skilled with 100 ca-rating and stuff ... sure you are rated high when you keep playing vs nubs. That's how (ql)ELO works.

[Yakumo]

Played in yet another game today ruined by one extreme slummer. My question is this: skill matching is determined by the average skillrating of the server compared to your own skillrating, yes? So say my skillrating is 50. 80, 70, 40, and 20 join a server. I'll see that as skillmatched because the average skillrating is 52(.5). Yet that game is going to be a **** fest depending on the shuffle, and even if the shuffle is even, it's just going to be watching the two beasts go at it, while the rest of us are meat shields. The guy with 20 skill tiered up so he could "learn better" (supposedly), but he's going to spend the whole game getting *****ed out because he has 9 points, and is 0 and 8.

This is incorrect, the (non duel) servers have had fixed skill level to display to all players since the game came out of beta:

Try a different game type?

but there will always be better players than you no matter what tier you are in, and they are not necessarily in the tier above, but could just be the top of your tier.

Having a look at a hud modification that shows sv_skillrating on your screen at all times could help relieve frustration (my hud from the guide in my sig has this) as you would know the average skill of the players on the server then.

I've had people screaming at me for tier slumming several times now when I've not only known I've joined a skill matched game, but the server rating shows 90 or higher so it isn't even made up of players that have joined a game labled 'difficult' or 'very difficult' for them.

The angriest player of all was on a 97 server.


Partly this can be a problem because if there are no games at Y level, then people do join the next server down just to get an active game, this can end up with a lot of X+1 (equal to Y) level players on an X server, and that doesn't change the reported level of the server. So then X level people keep joining not knowing the higher players, and go away unhappy at being stomped.

Servers used to change their reported level based on the sv_skillrating average of players on (including specs), but a huge number of players complained about this and campaigned for set-tier levels for servers instead, and id eventually relented, this does mean that if people are at least roughly following the guidelines there should be more servers showing for each skill level, but when the above problem happens the lower users are at a loss.
But going back to all dynamic level serves would make lower level servers too thin on the ground.


Hopefully at some point the scoreboard will show all users the server level, or even individual player levels, as then they can make a more informed decision on if they wish to keep playing as maybe the leader just got lucky, of if they're outclassed and they may prefer to move on.


The worry is that would stop too many players from even trying, far too many players are just defeatists, the numbers put them off and they don't try even when they have no idea what the difference really is, I got half my fun over the years trying to prevail even once against hopeless odds, trying to frag Thresh or any big names even just the once on US servers on 250ms, winning a game wasn't even a consideration.

you can check the skillrating at any time with /serverinfo in the console, look for 'sv_skillrating' that it reports, use the pgup/pgdwn keys to scroll the info. You will soon learn the numeric value for the games you feel most comfortable with.

nb. duel uses Elo and sv_skillrating only shows the rating of the highest player on, not an average.

[Stormcrow__]

This is incorrect, the (non duel) servers have had fixed skill level to display to all players since the game came out of beta:

Well, in game the sv_skillrating is still the average of all connected players. So the server goes through the trouble of computing the average skill of connected players and then shows people at the match browser... some... made up arbitrary number that is in no way indicative of the actual player skill you'll find on that server?

Yeah, matchmaking works great!

[*Cham3l3on]

So basically it's a number. Does the scheme take into account various game modes? It would be interesting to see the breakdown. Obviously raw skill is one measure of an individual. However, how that number applies to various game modes is limited. I have a feeling this is part of the issue with a generalized approach.

[Dyslexsick]

GET RID OF ****IN SKILL MATCHING!! IT IS INACCURATE AND IT DIVIDES THE COMMUNITY!!!!

I havent played a game of CA in months since NON are ever my skill level (lower skilled players being equal to me, or better).

The ranking system is busted,a fallacy since the beginning. I have itraised so high due to my sparingly amount of play and high accuracy... and I can never lower it.... ***!?

It's decided FOR me to quit supporting id on this venture.

I wish quake would have been better out of beta, but its just a ****ing joke to get money.

Cant even play any ****ing games with any friends, cant get a refund, or anything. ****ing thieves.

[AXxXxO]

this skill matching is pure crap!!!
didn't played normal ctf from beginning of that worthless skill comparing......return the damn skill like it was in beta,at least i could play on my level,now i play with morons and idiots....when i make team with friends and play,we win 20 games in row.why?idiots join high skill and want to learn how to play ctf.......what im trying to say REMOVE IT BECAUSE IT IS PURE **** and skill rating is just broken in pieces!

[Waperboy]

I'm a little confused - some people talk about "my skill is stable on 29", and "tier 5" and such things. Nowhere on the QuakeLive site can I find any data remotely relating to "skill" or "tier". The only indication of my supposed skill level is the server browser telling me if the match is lower, higher or same skill as me.

Someone care to elaborate?

[stownr]

I'm a little confused - some people talk about "my skill is stable on 29", and "tier 5" and such things. Nowhere on the QuakeLive site can I find any data remotely relating to "skill" or "tier". The only indication of my supposed skill level is the server browser telling me if the match is lower, higher or same skill as me.

Someone care to elaborate?

Go into an empty server and type /serverinfo in the console. look for sv_skill, that is your skill rating for that particular game type.
Or do it in a server with players to see the average of all the players i.e. the servers skill rating.

[ray]

Have you guys considered weighting the rankings differently based on how stable the teams were during the game?

I don't know how many times I've seen a wave of quits, switches and joins once a team goes up by a couple caps. At that point, the game is quite often just a waste. Should such "garbage" games be weighted less than a game where everyone plays straight through?

[*Cham3l3on]

Have you guys considered weighting the rankings differently based on how stable the teams were during the game?

I don't know how many times I've seen a wave of quits, switches and joins once a team goes up by a couple caps. At that point, the game is quite often just a waste. Should such "garbage" games be weighted less than a game where everyone plays straight through?

It's a good idea, but I think the issue would be determining what constitutes a garbage game? If someone's connection drops that's considered a quit, so you may need to keep track of reconnection attempts or something along those lines. On that topic, I played a 2v2 last night and the people except the one other person on the opposite team kept dropping so I have no idea how that ties into things. It forced me into a bad situation (not that I care all that much, not complaining but it made me think) but as far as statistics go, it's skewed for that match. In other words, a 2v1 in TDM (unless the 1 is a much better player) sounds like it would artificially skew the stats unless they have that already taken into account.