this post was submitted on 17 Nov 2024
446 points (90.7% liked)

Greentext

4437 readers
1173 users here now

This is a place to share greentexts and witness the confounding life of Anon. If you're new to the Greentext community, think of it as a sort of zoo with Anon as the main attraction.

Be warned:

If you find yourself getting angry (or god forbid, agreeing) with something Anon has said, you might be doing it wrong.

founded 1 year ago
MODERATORS
 
you are viewing a single comment's thread
view the rest of the comments
[–] Zagorath@aussie.zone 18 points 5 days ago* (last edited 5 days ago) (1 children)

While Elo (and side note: it's a person's name, not an acronym) isn't perfect and systems like Glicko-2 are better even for 1v1s, is there a better system than Elo that could be used to rate players in team games? Especially if there's a mix of pre-made teams and random teams thrown together by matchmaking?

Edit: extra bonus if it can be applicable in games that have both 1v1 and team game components where there might be a desire for some form of bleed between the two. (e.g. AoE2 where your starting Elo in one of them is based on your Elo in the other, if you've played a lot of one type of game before trying the other.)

[–] chicken@lemmy.dbzer0.com 11 points 5 days ago (1 children)

I suspect games tinker with the formula behind the scenes, to accurately place people faster if nothing else. The more players the longer it could take for the skill of any one to show up in the numbers, so I bet they factor in other game specific metrics at least at first. There would be some risk of this being abused, but that's less if they keep it a secret and maybe the progress numbers shown to players aren't quite the same as the real numbers used to decide who to match them against.

[–] Kolanaki@yiffit.net 9 points 5 days ago (1 children)

Plenty of developers of competitive games with SBMM have said they actually make it more about keeping the player playing than actually giving a shit about their skill. They don't use straight up elo, but everything they do does derive from it. They also don't really disclose how they come to the numbers it assigns you; probably because they don't want to expose exactly how their skinner box works.

[–] vithigar@lemmy.ca 3 points 5 days ago* (last edited 5 days ago) (1 children)

Street Fighter 6 uses two systems. League Points are a "keep them playing" type, and Master Rate is pretty much pure Elo.

Everyone starts with LP only and initial placement matches put you into a league with progressively fewer guard rails as you live higher. Rookie league can't lose LP at all, there's a win streak bonus up to gold, and you can't demote to a lower league until platinum. Throughout it all there's very slight upward pressure on LP, you get slightly more more a win then you lose for a loss.

Finally you reach the topmost league, Master, the final guard rails fall away and you're given 1500MR to join in the net zero Elo ranking pool. You basically need to demonstrate that you have a willingness to keep playing before they will use that style of matchmaking. "Real" skill based ranking effectively begins there, with the lower ranks being made more to show dedication rather than just ability.

[–] Zagorath@aussie.zone 2 points 4 days ago

That sounds kinda similar to how AoE4 does it, albeit with fewer of the more complex guard rails.

The game only displays your "rating", which resets each season and is used to place you into a certain league. You have two ratings: 1v1 and Team Game. There's slight upward pressure so if you keep playing you should on average climb. At the end of the season you get some basic cosmetic rewards (profile pictures etc.) based on the highest Ranking achieved throughout the season.

The game also keeps a hidden MMR/Elo. That does not reset each season and is the tool actually used by the matchmaking system to decide who you will be playing against. It's a true Elo system, or possibly an Elo-like system such as Glicko. The game keeps track of separate MMR for 1v1, 2v2, 3v3, and 4v4.