Jump to content
Sign in to follow this  
Qbit

Ranking System Discussion

Recommended Posts

Listen, dont talk down on the Halo kids. they just want the game to be as close to Halo 1 as possible, with no bullshit like weapon timers put into the game. They want it to be as skill based as possible and use weapon timers on iPods like tru competitive gamers. Dont need that fucking handicap built in.

No sure if you're referring to someone earlier in this thread, but this sarcasm doesn't apply to most as when it was announced, and still, the vast majority of people here are for in-game weapon timers. No skill in knowing the hidden time the sniper will spawn at is 11:56.

Share this post


Link to post

Not quite true. Who you lose to and who you win against also matters in TrueSkill and ELO, and that's one key difference between them and Zero Sum, in which wins and losses truly are the only thing that matter.

Yes this is why they result in a faster accurate ranking of players.  In all cases wins and losses matter but the trueskill and elo determine the quality of win by looking at opponents rank.  Without that a ranking system is prone to dodging where players only play when the competition online is weak since quality of opponents does not affect them.

 

 

Funnily enough, this isn't necessarily true either. The reason for that is that every player in Zero Sum starts at 0, which is basically the average player. If you have a rank of 0, your overall W/L spread is 0, so therefore you've won 50% of your total games and lost 50% of your total games, or haven't played any games. This is a good thing because it's a great idea to start off being pitted against the average player and either rank up from there if you're better than average, or rank down if you're below average. If you are an average player, you've instantly found your sweet spot from the word go. This is in contrast to a 1-50 where everyone is assumed terrible to begin with and starts off on the worst rank, rather than the average. Since it's Zero-Sum, if the combined W/L spread total of all the positive players is say, +10,000, that means the combined W/L spread total of the negative players is -10,000. If the highest ranked player has a spread of +100, you have to win at least that many games to get there (assuming they don't win/lose games in the meantime and rank up/down). The ranking system is therefore a ladder of all registered players, and their positions on it are determined by their W/L spreads. Each time they play they are matched against players with the same or closest W/L spread to them.

This is just implementation.  You could do the same thing with ELO or trueskill.  Just match the first match against the middle 50 percentile.  (also for clarification 1-50 is not the trueskill system but an arbitrary numbering system placed on top of it based off the trueskill rating system)  

 

 

However, keep in mind that this is the way it's done for parties and clans. For randoms vs randoms, the exact same concept applies but instead of the ladder positions being determined by the players' W/L spread, it's determined by the players' K/D/A spread. This is needed because otherwise the system suffers from the problems of poor players being constantly carried and good players constantly being dragged, leading to inaccuracy and randomness.

KDA is horrible metric.  If a rating/matching system is effective k/d of all players should migrate toward 1.0 for all players (except the best and worst) as they play enough games and if k/d was the only responsible factor for determining games.  Clearly it isn't.  It encourages poor play, and  shitty games.

Share this post


Link to post

Yes this is why they result in a faster accurate ranking of players.  In all cases wins and losses matter but the trueskill and elo determine the quality of win by looking at opponents rank.  Without that a ranking system is prone to dodging where players only play when the competition online is weak since quality of opponents does not affect them.

There can be simple metrics in place which prevent match-ups between teams who are a certain level of rank apart. Therefore, if no team who is close to your teams skill level is currently online (which would never happen in a game with a healthy population anyway) no such match would take place. Realize too that even in ELO and TrueSkill systems, matches like that are still bad. It's still going to be free rank for the good team assuming that they haven't run into a bunch of de-rankers or a team full of alts who are trying to screw over high ranked players, and no one wants to get continually creamed by much better teams or face teams that aren't a challenge to play against.

 

KDA is horrible metric.  If a rating/matching system is effective k/d of all players should migrate toward 1.0 for all players (except the best and worst) as they play enough games and if k/d was the only responsible factor for determining games.  Clearly it isn't.  It encourages poor play, and  shitty games.

You're not distinguishing between K/D/A spread and K/D/A ratio. The K/D/A ratio of the players in an effective ranking/matchmaking system does indeed tend towards 1.0 and that's exactly what it does in a K/D/A system. The K/D/A spread however, will always be positive for higher ranked players, because they consistently went positive against players less skilled than themselves and therefore ranked up until they reached the point where their average K/D/A spread per game was 0.0 (matched against opponents with same skill level).

Share this post


Link to post

Never once seen a player below gold. Even the thumbless.

I have a person on my friendslist that is a silver. I couldn't believe it when I saw it, but they exist. Now bronze......still haven't seen that. 

Share this post


Link to post

The 1-50 system worked so much better in my opinion and what I've experienced.  There's far too many games where teams are lopsided (to4 vs. randoms most likely) 

But in a perfect world say parties matched parties; randoms matched randoms. Would you guys say that this ranking/matchmaking system would work more efficiently that way, in making more competitive games?

Share this post


Link to post

All this ranking system talk is useless as long as there are not party restrictions and 343i ranks every single playlist. I won't be surprised if Action Sack is ranked (if it ever comes out)

Share this post


Link to post

You cannot drop out of a division once you're placed there, is that true?

Surely not for champion though?

Share this post


Link to post

You cannot drop out of a division once you're placed there, is that true?

 

Surely not for champion though?

You can drop out of champion, because it is the top 200 of onyx, so it's always changing. You cannot drop out of any other tiers, though.

Share this post


Link to post

You can drop out of champion, because it is the top 200 of onyx, so it's always changing. You cannot drop out of any other tiers, though.

What an absolute joke. It's like having a ranking system with training wheels on.

Share this post


Link to post

There are a few tweaks that need to be made to the Halo iteration, but overall, the modified ELO systems are the best systems that have been implemented to date.

Share this post


Link to post

There are a few tweaks that need to be made to the Halo iteration, but overall, fps games have played better without sprint to date.

Share this post


Link to post

Listen, dont talk down on the Halo kids. they just want the game to be as close to Halo 1 as possible, with no bullshit like weapon timers put into the game. They want it to be as skill based as possible and use weapon timers on iPods like tru competitive gamers. Dont need that fucking handicap built in.

Lol this was actually an argument against adding timers in game in league. Once they were implemented it didn't cause anyone to start winning like crazy and go from silver to challenger like some were making it out to be. All it did was make the game ever so slightly more competitive down in the shit tier of ranks.

Share this post


Link to post

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×

Important Information

By using this site, you agree to our Terms of Use.