Ranked Systems in games

I really want to discuss ranked systems in games. There are many systems out there, but which do you think are best?

I feel like an arbitrary system to measure skill will always be flawed, because there is just so much an algorithm can have as it's variables, and sometimes winning certainly isn't everything, since there are random factors that could have played into it, but that doesn't mean that they don't give a general idea of how skilled a player is, and the more games/play sessions that are taken into account, the best it gets to an actual representation of skill.

Now, online competitive videogames are obviously based around measuring skill (and having fun) and most of them have picked up ranked systems to give out an idea and to make equally skilled players play with/against each other. Not all games used this, but for example I remember games where you could enter a server with random people that had very different skill levels, i.e. Quake or CS. I think this had 2 effects, it could either discourage new players (constant stomps) or make them learn the ropes if they opt to watch closely what they do.

Before that, it wasn't terribly bad, but it had clear flaws that systems like the one on ICCup for Broodwar tried to fix, it gave a better look at how good or bad a player was based on some variables (and the win ratio), but it was still rudimentary. Warcraft III had a basic system too, where you got XP with wins and ranked up by having enough XP to reach the next level, and I would say that from there they tried to make the Starcraft 2 system, which is the best for 1v1 type of games since going from one rank to the other felt meaningful and rewarding. ELO in chess is probably very accurate and a great example too.

However, I think that team-based games are where ranked systems start to get kind of strange and probably not a good idea, not because they are wrong or always flawed, probably they are 100% right on determining the general stroke of skill; players higher on the ladder are better than those bellow; but it feels weird that team-games and the dynamics of it should go in hand with the the skill required for it. A good player is that which has the most skill or that who works best with the team or that which can lead his teammates? I think this makes games inherently frustrating and drains the fun of them, if MOBAs have a caustic community, it's in part because ranks and ladder matter so much that players start hating on each other as soon as they see a loss coming, and it doesn't help either that the matches are at least 30-mins long on average on the lower spectrum. Measuring solo-skill on a team-game is kind of weird as a premise.

Nowadays there are tons of team games and all of them seem to force their own flavor of rank and ladders, but I don't get why would they feel the need for them and who has the best system?

ranked systems is a mistake

I rank op didn't read/10

read the 4th paragraph and the final line then, the rest was more of a general talk about ranked systems in 1v1's and how I personally think that SC2 did it best (although the game was mediocre).

Brevity is a soul of wit nigger.
I'll read last line.

Because people need a straightedge ruler, so they can see how much they "progressed" since the last time and compare their epeen with other players.
This is why Experience, Levels, Upgrades and all that nonsense exist in pretty much every genre now.
Fuck basic score system in arcade cabinets is exactly that too.

Remember when leveling systems were called "RPG elements"?
It seems so long ago, but it was only 4 or 5 years

That's the issue though
In a TEAM-GAME where people play solo ladder, you can't expect the ranked system to give an accurate representation of how skilled a player is, or maybe it can, but it makes the whole premise of playing a team-game flawed.
Yes, you can train yourself to be super skilled and carry every game, but that in turn might make you less of a team-player and kill the whole experience for yourself and others. Instead of focusing on stuff like communication, leadership or team-strategy; I see people playing Lol, Dota 2, CS:GO or Overwatch try to win just by themselves and become caustic pieces of shit that no one wants to be around. Either that, or the whole game community is at each other necks every game, which again, kills the idea of a team-game.

How slow are you faggot? This shit took 2 minutes top to read.

People don't want to git gud, people want to be explicitly better than others, while putting as little effort as possible.
Those who want to be actually good team players don't play with randoms, they play with actual friends against randoms.


If person needs a wall of text to convey his idea, chances are he's either an idiot or tries to discuss too much on one thread. Or both.

Keep trying. Sorry it's not your usual template thread with old memes instead of actual discussion.

...

Kinda hard call people out for not discussing the topic when you wouldn't even know the topic if OP didn't tell you to read 2 sentences fam. You're only hurting yourself further at this point.

...

My favourite system is Dota's MMR system. It gives you a flat number where meaning is derived through the open statistics throughout the community.

The issue with ranking systems is one of sampling. Ranking systems sample players as individuals and make predictions about players relative to the population of individuals, however players are actually tested as nested within teams. This means that the data that is supposedly sampled from a population of individuals actually has less variance on a per-match basis that you would expect (i.e. players on a team are more similar to each other than randomly selected players by virtue of having played together). Inferences based on this constrained variance aren't as generalizable to other teams, which means predicted win/loss from matchmaking is incorrect.

In sum, matchmaking algorithms generally expect independent sampling of individuals, but get nested sampling of individuals in teams which inflates error in predictions

All the systems are pretty similar to one another. you have a ranking that gets you in matches with people of similar rankings. Ranks are changed based of outcomes and expected outcomes.
Only real differences is in the presentation. I would prefer a system that is transparent about hidden MMR and doesn't make a big deal out of it.

i got placed in Bronze overwatch, because of internet issues over the week.
After a few games i'm back up to diamond.
It's weird seeing how lowbies play the game. It's actually so peaceful to see people just run around and try to kill each other, without arguing in VOIP and saying "That's a bad pick"
Being able to play Junkrat and Mei without being REEEEE'd at by autismlords.
The higher you get, the more rigid the main COMP becomes.
Mfw one DC nets you -500 ELO.

ELO works in chess because both players start the entirely the same, always have the same opportunities, and can do the same things.
This doesn't translate well to video games at all.
1v1 rankings are better than team rankings because it is just one person against another but it also would need to assume each player is starting on a map that is essentially symmetrical/balanced in every way and they are playing the same/perfectly balanced character/race/etc which is almost never the case.
Team based games make this worse because they introduce dozens of ways that personal and team based ELO wouldn't ever be calculated properly because frequently the players will change, no start will be the same, no characters/races/etc, no maps are starting the same, etc.
At this point in time rather than having a ranking system in a team based game its more accurate to say they have a win percentile ranking. This won't change until it's possible to make an algorithm that accurately measures how much a person contributed to a win or a loss and I don't think it will be done in the near future.


That sounds like it's supposed to be a DC punishment that was given harshly for no real reason.

Fuck off and give me a leaderboard. Osu! does it best.

While Osu! might do it well, the scoring is fucking gay.

Leaderboards are for games where you play alone and then grade how well you did compared with everyone else.
Ranked is necessary because a new player might have a shot at winning against other new players but he'll lose 99% of the time against veterans.

What I would want out of a ranked system would be a simple breakdown on how well you play. Specifically:
Just because I think it would be funny as fuck to find players with high Individual Rank and very low Team rank. Because of rank works, they'd end up playing with other equalyl fucktarded no-team-no-play idiots and lose individual ranking in the process.

Only game I've played with a ranking system recently was Overwatch, so I guess you could measure team-rank based on:
The last one I'm not sure how it would work.
Individual rank would just take damage/kills/healling like normal.
This value gets compared to a "par value" calculated by your Rank. For instance, at rank 2500, if you're playing a Mercy it would expect you to heal Y damage in a match based off on your teammates health and damage taken.
If they take less damage than usual the "par value" goes down or if they change to more HP characters, the "par value" increases.
Slap a final bonus/malus based on wether you won or not and you could still rise in ranking even if you lost. You wouldn't rise as much, but avoids the "I played really well, the other 5 idiots didn't" situations.

The bad side of this is that a ranked system that doesn't focus solely on win/loss means you'll end up with a few players "farming" for points. They already sorta do this with normal ranking, it would be much worse if they figured "I can get Mercy, my friend gets Junkrat and we just alternate while racking up ELO".
Would either need moderation to step in (which Blizzard does but takes months to) or a very elaborate algorithm that impeded this.

Oh right, Tanks should proabably get ranking/score when they soak up damage.
Retarded tanks will take 400-500 damage and die.
Good tanks keep healed with medpacks and stay in the front soaking up 2000 damage before dying.

The main issue is that companies don't want to invest onto this, not because it doesn't make sense, but because it's so expensive (maybe not only on money, but on human resources) to make a good algorithm for a team game ranking system, that it's not worth it for them.
Although, a game that sells itself with an image of: "In our ranked system you might not win, but if you play really well, we will give you points or decrease the number of points you lose!" will certainly be popular among people.

Maybe a community effort to improve ranking systems would also be a good idea, but good luck having Valve, Riot or Blizzard disclosing the source code of their algorithms.
It's cheaper to just make cosmetics at this point.


Only giving you the number feels kind of "ascetic", when there are ways where they could improve. For example, giving the new percentiles of mmr or give info when you hover over your mmr in-game, i.e. "You are among the 15% best players of X server".
Another thing they could do is make the "Captain" role in pubs, based on the commend system (which would need to be actualized) and make it important, at least letting players know that they can follow the calls of a person within the team. I know this happens in CM, but it's more about who clicks first, and sometimes the retard on the team does click faster than the rest. Dota, more than other games of it's genre, needs to have a Captain.