• Please make sure you are familiar with the forum rules. You can find them here: https://forums.tripwireinteractive.com/index.php?threads/forum-rules.2334636/

ROLadder announces some of its features.

My focus with scoring systems is not using a system that purely lets teams battle for the top by climbing up and kicking other people down. My goal is letting people of comparative strength play each other, and for that a predictor gives you more accurate results than no predictor. Its a difference in goal in what we both want to obtain.

The issue I have with rung is that it essentially predicts that you loose against every team above you, and win against every team below you, as otherwise your rung is wrong. That is the basis of a rung ladder. The issue with that is that while you climb really fast teams after a while do not really settle somewhere. But continue going up and down which is why your ranking is only determined by your last match, and the team in the last match.

When you use ELO IRB Glicko TrueSkill etc, you measure someone's performance over multiple matches, allowing you to factor out differences of team performance due to averaging performances over multiple matches. Next to that with advanced systems like Glicko and TrueSkill that for instance you can take into account the inaccuracy of someone's score due to not playing for a certain time, or inaccuracy in someone's score due to never having played a match.

Although a rung based hop over system doesn't have a lot of issues with new players as if you win you simply get placed above the team you won against, without that team suddenly being worse than the other teams he previously won against. If you use the swap system then potentially the nr1 of the ladder can be swapped with the newest player in a ladder. Which is why it is more common to use a hop over rung ladder (at least if you can challenge any position above yourself) where if you win you get placed above the team you beat.

In the end its power to the people that play, that decide what system will be used, even though personally I'm not a big fan of rung based (unless its in small tiers to avoid rung based issues) but its not up to me to decide. With rung at the end of a season the score depends solely on the last match that a team that plays, rather than depending on the whole team that played matches. And that Is something I'm not a big fan off.

Since we can have different rating modules we can probably implement multiple systems as snuffel suggested and let them run next to each other (and gather statistics like which system converges faster, is the most stable and predicts the most outcomes correct), and rather than giving separate medals for every system letting the community say what they think about various systems. As discussing these things with clear facts is easier ;). This should probably at the least be possible with the mathematically more easy rating systems such as various rung based systems and elo/irb based systems.

Its exactly by taking into account multiple matches, that there is a damping in the system. Which means that things such as playing with different teams won't have you hopping around extremely in rank from match to match. So you would rather end up basing someone's score based on a teams results.
 
Last edited:
Upvote 0
I can only hope we see tournaments pop up in places like wanfest, alienware arena, and others like ESL. The last thing I'd ever compete in is some ego ladder.

Although discussions about rating systems might not be the most interesting discussions, but the talks hold truth for any league or ladder. If anything all of us here are discussing this in openly in here. People are free to step out on such discussions or not. If anything discussions like these are good to garner an idea of what people want in a scoring system.

But ESL for instance uses a basic Elo system, that game Admins cannot tweak in any way.
 
Last edited:
Upvote 0
What about a "global ROL ranking" that takes into account all Ladder, League and Cup matches in one system? Yet another of these things at least I've never seen. Taking it a step further would be to have that as the actual ladder ^^.

Would be fun (for me lol) to hear your point of view about that =).


(RO2 was not an option (out of 40 games, where 5 listed as "Uncertain release dates") in alienwarearenas poll about "most anticipated PC exclusive of 2011", so DOTA2 had to get my vote =(.)
 
Last edited:
Upvote 0
You need to have a valid ground of overlap for something like that to work. Aka enough teams playing in say the leagues should play in the ladder as well. Otherwise it becomes impossible to connect different teams.

As the average skill of say the league or the ladder is not the same. So you need enough overlap to be able to predict how much they differ from each other to be able to give global rankings that mean something. By the way point rating systems are setup you wouldn't need a lot of interconnected matches though.
 
Last edited:
Upvote 0
I support kinglol with his black box example. If a good player leaves, then everything will adjust on its own. Lets not tamper with predictions and formulas. And the best way to play is with leagues as the team gains a better sense of progression/achievement when it enters a higher league (as zets pointed out).
 
Upvote 0