The old ladder rating system also threw out any game over 3 months, causing (some) players to manipulate this timer in order to consolidate losses and wins and reach a higher peak than would be earned through normal play.

3 months is not very many games, especially when 2 of your 5 game slots might be played against stallers taking 3 days per turn plus vacations. You could easily defer a lost game for well over a month before realizing the loss.

1v1 Ladder has changed from a rating system that was decently accurate although couldn't distinguish between top players to an idle quest feature.

Anyone who thinks this is a reasonable rating system has only looked at the names at the top and said to themselves "good players are on top of the ladder must be good system," and not played through 30 games of junk just to get to the relevant part of the ladder, while staring down needing a 30 game win streak to over take someone whose been on the ladder a year longer than you.

It's not just that Rufus is untouchable at the top, it's that the players just below him also need like 80-90 games to catch. That is patently absurd for a MD system.

If it wasn't for the idle rewards I wouldn't bother playing it at all. Every few months I play some games to progress to the next idle league. Not sure I'll bother once I get the 600 one.

Skill issue

What’s the short version of what was wrong with the old ladder rating system? As far as I understand, it was favouring winning streaks.

Here's a shameless plug to one of my threads with a deep dive
https://www.warzone.com/Forum/451002-bayesian-elo-unfortunate-choice

Orcus, can you please explain to the class:

(a) Your last 8 ladder games include games against Rufus, Buns and adrenaline. Do you consider these opponents to be part of your “30 games of junk”?

(b) You have lost 4 of your previous 8 ladder games. What is your expected win rate over the “next 80-90 games” against these “junk” opponents? Surely it will be quite high in order to catch the “not-untouchable” Rufus?

(c) Why do you consider the 600 rating threshold to be trivial when only 11 players (Rufus, Krzyszstof, Crouton, Dom, adrenaline, Buns, Odin, Aika, Octane, QH, AI) have accomplished this? Do you consider yourself to be a peer to these players?

Dude, the system is giving you good games against comparable opponents. You’re going .500 in these games. What more do you want? You’re rated 700 points below Rufus because he’s that much better than you.

@TheGreatLeon

1) Rufus is not 700 points better than me. On the MTL where I have a 67% win rate (indicating I am currently underrated as all but the very top players should equilibrate to 50%) I am currently 454 points behind him. I also have a lifetime 1v1 record against Rufus of 3-8 (27.2%) if you discount auto-RoR games (2-2) 50% if you only count this account and (4-9) 33% if you count AutoRoR games but you really shouldn't. I know Fizzer isn't using elo exactly, but the expected win rate for an elo gap of 700 points is 1.75% which is way below my win rate vs. Rufus. Rufus is the best player in the game. He deserves to be #1, but if you think he's 500 points stronger than Adrenaline you're either ignorant or smoking something. Rufus is where he is on this ladder because a) he's the best player with skill that is approached by only a handful of players b) he has been on the ladder constantly, and c) he plays really fast because he can beat almost anyone without playing slow.

2) As for the players who have gotten 600 are they my peers? Keep in mind I got 2280 on the old 1v1 ladder with my old account and mid 1800s on the MTL. Rufus (no clearly better than me), Kryzy (yeah I think so but historically more accomplished to me so I wouldn't be offended by anyone saying chu chu is stronger), Crouton (9 places lower than me on Greece seasonal when he beat me head to head, no MTL so can't compare there), Dom (3 places above me on Greece seasonal, +28 on MTL currently), Adrenaline (clearly better than me), Buns (clearly better than me), Odin(IDK old player hard to tell but based on what I've seen probably better than me), Octane (clearly better than me), Aika (-29 from me on MTL currently), QH (same as Odin), AI (clearly better than me) . BTW you missed Gunslinger who I consider a peer. So yeah most of the players who have gotten 600+ are clearly better than me but I consider at least 5 (Kryzy, Crouton, Dom, Aika, Gunslinger) to be my peers. You noted I was 4-4 in my past 8 games, you failed to note I was 13-8 (59%) in my games since I re-started post 400 in January. 600 is not an obstacle that I'm afraid I can't get, but it is an annoyance because at 60% win rate it takes many games to climb 100 points. I halted for the time being because I'm playing MTL (13-2 last 15 games) and CL. I've played 60 games on the nu ladder. I've gone 42-18 (70%) in that time.

3) Even if you think I'm overestating my own skill. It doesn't matter. The rating system isn't about me. At +15 a game you have to go 45-0 to catch Kryzy, 71-0 to catch Rufus. It's way too many. And that's if you go freaking undefeated. At 70% (my overall win rate) and assuming equal rating change for wins and losses it's 113 to catch Kryzy (nearly double my games) and 178 to catch Rufus (nearly triple my games). At 60% (my recent win rate against stronger competition) it's 225 games to catch Kryzy's current rating and 355 games to catch Rufus' current rating.

The rating system is absolute garbage. Just because the best player in the game occupies the #1 spot, doesn't mean a rating system that requires hundreds of multi-day games isn't beyond stupid for accurately rating players.

Edited 3/1/2024 10:28:03

If I may refer back to my old thread: https://www.warzone.com/Forum/697546-new-ladder-rating-system-working

A 1500 rating difference here is designed to be 75% win rate, that's how Fizzer implemented it, so Orcus with his 27% win rate against Rufus should long-term expect to be rated roughly that much below him.

But that's a side fact, in general I do agree with Orcus, the rating system isn't great for the sole reason that it takes too long to stabilize (which is by intention BTW, which is something I also pointed out in the old thread).

Edited 3/1/2024 10:58:57

Thanks Beep, I kinda default assumed that the rating points were comparable to ELO which they're clearly not. In that case though everyone should be a couple thousand points higher. 1500 below Rufus is -300 and since I've won 70 percent on the ladder, average ladder player should be something like -1500.

Edited 3/1/2024 11:11:17

I made a lengthy post about the number of games on page 1. I don’t see it as a problem, especially in comparison with the other options.

It would be nice if we had a variable k-factor for the first ~20 games so people could equilibrate faster but the TrueSkill system uses something similar and I believe it’s working reasonably well.

Here’s the way I see it and you’re welcome to disagree:

1. There are maybe ~30-40 players who can equilibrate above ~500-600. I don’t know if you’re one of them, let’s assume you are.

2. It doesn’t take that many games to get to ~500 if you’re one of these players.

3. It doesn’t take that many games to get to ~300 where you are thoroughly outside of the junk region and playing decent games against decent players who can beat anyone with some luck.

4. In terms of catching Rufus, the fact that you’re talking about 60% win rate says it all.

Rufus is where he is because a 60% win rate would be atrocious for his standards. Yes, it would take an insane number of games, probably an infinite number of games, to get to ~1100 rating at 60% win rate. That’s the system working exactly the way it should. People are being shuffled to where they belong and a 60% win rate player doesn’t being on top.

Last, you’re out of your god damn mind if you think you’re better than Aika lmao. I would literally bet my house on Aika in a best of 9 on Strat 1v1.

(which is by intention BTW, which is something I also pointed out in the old thread)

Would be interesting to see from Fizzer's perspective whether he achieved his goal. Obviously, the goal was to make people play the ladder for a lot longer, rather than the previous play-20-games-to-get-rated-and-stop thing that a lot of people used to do.
Obviously for the people playing, this is working - but how many people are being deterred from playing the ladder at all, because they know it'll take too many games to reach their true rating?

Thanks Beep, I kinda default assumed that the rating points were comparable to ELO which they're clearly not. In that case though everyone should be a couple thousand points higher. 1500 below Rufus is -300 and since I've won 70 percent on the ladder, average ladder player should be something like -1500.

You would think so, but no.
I really think you would enjoy the insights from my old thread.
The average player will long term have a rating of 550.

Also Aika vs Orcus Bo11 for Leon's house when?

I mean, we’ve seen this before. The player who’s clearly worse with a weird insecurity about playing 60 MTL templates moderately well always gets thrashed by the player who’s clearly better and takes the time to figure out the one template in question.

Jack of all trades…

The GreatLeon, you don't really understand but that's ok. Rufus will eventually be rated MUCH higher than he is currently. Anyways thanks for giving me the motivation to grind out another 100 points. I've rejoined the 1v1 ladder.

Fantastic! The more players on the ladder, the better pairings for all of us. Plus with a little luck you might manage to even the score between us as well.

Happy you’ve rejoined orcus. Inspired, even. Maybe we should all rejoin!

I'm not sure there's much need to say anything more about the rating system than what was already said in the thread Beep linked. He did a very thorough analysis of the ladder settings, spelling out what they mean and the implications to the ladder. The analysis was good enough to prompt Fizzer to respond with a change to the parameters. TLDR; it's a good system for rating players, in principle, but it takes many hundreds of games for ratings to stabilize, which means that it will have a very sizeable activity component for almost all players - and this is by design, so it's not worth complaining about it.

Edited 3/1/2024 20:48:43

I agree that Fizzer did put much thought into tweaking the numbers so that the ladder works as he intended.

And I've always tried to be as respectful as I can towards Fizzer, but seriously, let's just look at the numbers:

3v3 ladder
Players currently participating: 157
Players currently in bronze league or higher: 30 (19%)
Players who currently have rating lower than the 100 they started with: 77 (49%)

In every game I've ever seen, the whole point of leagues is to motivate players. Basically everyone who cares should be able to at least reach silver and feel a little bit accomplished.

In Fizzer's version, half the players only see their rating fall down, and only 1 in 5 is able to reach bronze league.
Bronze league.
20%
BRONZE LEAGUE

New ladders have been running for 10 months, and it's been 8 months since the last tweak. I consider the ladders an abandoned project.
The only ladder that matters is the community driven MTL: http://www.warlight-mtl.com/

Edited 3/1/2024 21:20:00

MTL is best, I also recommend AWP and Seasonal Ladder though :)

This isn't a complaint of the new rating system but I do have a question of how it's displayed on my profile. Is there anyway to adjust the Y or X axis for the rating or ranking? if you look at my profile, it seems to me like a worthless graph