Competitive Solo Queue is where this game fails

[Amuro]

I have a background in data and work as a software developer, so I'm a bit familiar with how their system is most likely setup.

NetEase loves their data. This is known and while I do appreciate it, I think some of their design philosophy around ranked isn't working too well

1) solo queue has EOMM and its a bit apparent when you dig into win rates. You just know you're screwed when you get that black sheep player. Peeking at their profile on a website or in game career confirms everything. I've seen it all.

The 80 hour Lord Namor that has a solid 40% win rate over 200 matches played (LMAO)

Why does the system do this? Maybe its to enforce 50/50 or to prevent winning streaks. I notice my games are very "coin flip" and I'm sure everyone can attest to this. The enemy team has a bunch of 50% WR players and you get 1-3 40% WR players. They'te not trolling per se, but they're god awful and basically useless. I understand no system is perfect but the blatant "cointoss" match making is no coincidence. League of Legends, CS2, Valorant is and was never this bad.

2) No role queue means you might have to fill to bite the bullet and for the greater good of the team. If people don't fill or round out the team, it's an automatic wash. But wait..how can the game differentiate you and your performance? We don't know if it does or not.

3) getting SVP etc only reduces point loss by about 20% (-19/-20 versus -25 for example)

The game fundamentally is there to keep you happy, engaged, and bringing you back. A proper "ELO" system should roughly make games "50/50" if its tuned properly. The problem with this is different roles and champions. We don't know if NetEase secretly has an ELO numerical value assigned to you as a whole and each hero you play. But this is all irrelevant since your visual rating is all that "matters" to you as an end player.

The ranked experience is a big make or break for a game like this. While the reviews are "glowing", I suspect peoples minds will change - much like what happened to Overwatch. Currently the game is seeing a slow hemoraghing of players since launch. It's gone from a 644K player concurrency peak to a current average 327K 24 hour peak.

I've spent about $70 USD on this game and I'm fine with it. I got a little over 100 hours played but I think I'm going to stop and take a break. I'll come back eventually and try it for season 2 but the stability and solo queue grind isn't great. League, Valorant, CS2, OW2, etc all have better "solo queue experience"

[Amuro]

Originally posted by SecretSins187:

https://nos.netease.com/mg-file/mg/neteasegamecampus/art_works/20200812/202008122020238605.pdf

This is all that needs to be said.

Originally posted by Fruitseer:

Originally posted by Amuro:

trust me, they're not as coin toss. Ive played them all. You can see my CS2 hours.

Yeah, we're supposed to trust you, lol.

You dont have to trust me, you can read the paper on match making and draw your own conclusions. But the TLDR is the paper proves the coin toss hypothesis.


The paper from Net Ease claims their proprietary OptMatch Match making does the following:

This paper proposes a two-stage data-driven matchmaking frame-
work (namely OptMatch), which is applicable to most of gaming
products and has the least knowledge about the product required.
OptMatch contains an offline learning stage and an online planning
stage. The offline learning stage includes (1) relationship mining
modules to learn the low-dimensional representations of individ-
uals by capturing the high-order inter-personal interactions, and
(2) a neural network model to incorporate the team-up effect and
predict the match outcomes. The online planning stage optimizes
the gross player utilities (i.e., satisfaction) during the matchmaking
process, by leveraging the learned representations and prediction
model.
Quantitative evaluations on four real-world datasets and an on-
line experiment on Fever Basketball1 game are conducted to empir-
ically demonstrate the effectiveness of OptMatch.

I read the entire paper and this isn't the first time I've read something like this. It's all pretty legit but also a lot of $10 words to describe in plain english,

"We do online and offline modelling to try to keep it fair. We use minimal metrics and keep the system lean so it doesn't need to know much about the game to do match making."

Which I believe. But did you know a lot of players have mentioned "Log off if you lose a game or two games in a row, you'll get better games". That's gaming the offline learning system. If you log off, the system has time to model match making for you in it's offline mode. It'll most likely set you into a more fair matchmaking sequence when you log back in. Also if you lose a ton in a row, it usually serves you some "easy games". Nothing ground breaking but this proves the coin toss theory is built into the game. Read the paper's introduction and conclusion if you don't believe me.

[Steve]

Originally posted by SecretSins187:

https://nos.netease.com/mg-file/mg/neteasegamecampus/art_works/20200812/202008122020238605.pdf

This is all that needs to be said.

This is, in fact, the thread answer because it shows all the math.

What it doesn't show is a conspiracy theory.

[Amuro]

Originally posted by Fruitseer:

Originally posted by Amuro:

You dont have to trust me, you can read the paper on match making and draw your own conclusions.


The paper from Net Ease claims their proprietary OptMatch Match making,

> " The empirical results show that OptMatch gains an improvement
of 6% ∼ 17% in the offline prediction tasks for the match outcomes
on four real-world datasets, and an improvement of around 40% in
the match quality during an online experiment"


Wow, 40% improvement of match quality in online experiments? Yet coin toss is a very wide spread complaint. They claim their match making system (OptMatch) is game and product agnostic, and they use DOTA2 and NBA FEVER game data to run through it. Which is fine, but both games have set roles and a pre-selection draft system. Rivals does not.

I read the entire paper and this isn't the first time I've read something like this. It's all pretty legit but also a lot of $10 words to describe in plain english,

"We do online and offline modelling to try to keep it fair. We use minimal metrics and keep the system lean so it doesn't need to know much about the game to do match making."

Which I believe. A lot of players have mentioned "Log off if you lose a game or two games in a row". That's gaming the offline learning system. If you log off, the system has time to model match making for you in it's offline mode. It'll most likely set you into a more fair matchmaking sequence when you log back in. Also if you lose a ton in a row, it usually serves you some "easy games". Nothing ground breaking but this proves the coin toss theory is built into the game. Read the paper's conclusion if you don't believe me.

If this was true, then everyone would be at a high rank and there wouldn't be all this complaining about how hard it is to gain ranks.

The problem is they used DOTA2 and NBA Fever data which has pre-draft for DOTA2 and role for NBA FEVER. There isn't for Rivals. A Diamond DPS could be forced to play tank for the first time in their life and it would throw the match making algorithm completely off.

I've gotten games where you get 4-5 DPS that won't switch with one healer begging for some sanity. This paper and the current system has nothing to prevent those situations from happening. I saw it happen all the time in Platinum.

[Amuro]

Originally posted by Steve:

Originally posted by SecretSins187:

https://nos.netease.com/mg-file/mg/neteasegamecampus/art_works/20200812/202008122020238605.pdf

This is all that needs to be said.

This is, in fact, the thread answer because it shows all the math.

What it doesn't show is a conspiracy theory.

Theres no conspiracy. Their own match making paper used game data (DOTA and NBA) with role queue. Rivals doesn't have role queue.

Do you not see the issue with that?

[SecretSins187]

I didn’t read the entire paper, but from my understanding, the system models “team interactions” as well, and matches people up accordingly. In other words, the system KNOWS if you are an insta-locker or thoughtful; it knows if you stick with a character, or flex pick to fit your team. Rank is irrelevant. You can be a Gold teamed up with 4-5 GMs.

So yeah, that match where you got 5 Duelists and a Vanguard? Not an accident.

[Amuro]

Originally posted by SecretSins187:

I didn’t read the entire paper, but from my understanding, the system models “team interactions” as well, and matches people up accordingly. In other words, the system KNOWS if you are an insta-locker or thoughtful; it knows if you stick with a character, or flex pick to fit your team. Rank is irrelevant. You can be a Gold teamed up with 4-5 GMs.

So yeah, that match where you got 5 Duelists and a Vanguard? Not an accident.

prob because im willing to fill. decided to just uninstall and come back if things change.

[SecretSins187]

Originally posted by Amuro:

prob because im willing to fill. decided to just uninstall and come back if things change.

I feel your pain. I’ve learned to manage my expectations of these games since the F2P revolution, some 15 years ago. These games aren’t about skill anymore, as much as they are about luck, time, money, and learning to game the system. I still feel MR is a good game, but finding out about EOMM and the new de-ranking system is demoralizing to me.

The best thing about F2P games is to learn how to participate without spending money. I’ve had no desire to achieve max rank after maxing S0; maybe the 20-game participation penalty will increase my motivation, maybe not. It’s just another digital reward I’ll never use. Otherwise, it’s just QP for me.