Duelyst Forums

Talking of S rank being broken

#23

i reached s-rank on the 9th September and i won 3 games in a row at s rank because i was lucky so i ended up at rank 4, there is no way i am somewhere near to rank 4, skillwise ( i play just 60-70 games (wins) a month)

so if i decided to camp until the end of the month (which i didnt) i would be in the top 50 for sure, thats just dumb
i have some campers in my friendlist, must be fun to be on a list, which everyone knows is borken and doesnt give shit about

players with 100+ won games at s-rank with a winrate far above 50% dont end up in the top 50 but i do?
my own winrate is around 62% but i didnt play much because

winning a game against another player (mostly diamond) looses ranks?`

i feel like i am punished because i decide to play after i hit s-rank

the s-rank system is a mess, how is it possible that players who hit s-rank like on the 29th of the month, with 0,1 or 2 games at s-rank, have place in the top 50?

#24

I think you are way too harsh on yourself, if you manage to make to S-rank the 9th days of the season it is already a good accomplishment on its own. Winning 3 times in a row against actual S-rank players than it is also a big accomplishment because people who reach S-rank the 10 first days of the season are all very good players. Keep also in mind that there is very few people in S-rank so early which increase by a lot your chance to reach high rank right from the start.

You chose a pretty poor example to demonstrate your point.

Agree with everything else you said , the S-rank system is broken blabla … but I maintain my initial point: stating that most top50 players are"very bad" is just not true and even if they don’t necessarily deserve it you need to be good at the game to achieve it.

#25

It’s actually not that easy to get S early on because most of the time you’re actually grind through people that were diamond the previous season. Chances of encountering those good people who are also trying to get early S isn’t that common.

Anyway, it’s irrelevant to the discussion. Point is how anyone who gets S rank can easily end up in top10 just by getting few lucky wins. The situation is such that being top10 is just as relevant as being anywhere else on the S ladder. It just depends on how lucky you were in your first few S games.

#26

Firstly, I’m pretty sure they are talking about ranking people in S, where the rank floors are not relevant.

Secondly, an ELO system is not required to interact with sub-S ranking or even matchmaking. CP could track your ELO from silver to diamond with no issue and not use it for ranking purposes until S. This is because ELO is designed to fairly handle cases like a game between a grandmaster and a beginner. So if that game happens because of how divisions work it’s actually not an issue.

1 Like
#27

You replace the word “lucky” by “good” and I agree, at the end of the day Duelyst is a skilled based game from my experience luck is game defining roughly 20% of the time.

#28

Well there is also variance Thorrk. You could simply happen to win certain matches all in a row even if you have a 50% winrate. The games themselves wouldn’t be lucky wins, but the fact that you won them all in a row would be.

#29

Even if you look at it that way, it’s not okay for everyone who ever reaches S rank to have 20 percent chance to end up in top10 by sheer luck. And even the ones who are good enough to win those first couple of games in S, it still doesn’t show their skill level. Shit’s fucked either way.

#30

Part of the reason the system is broken because you only have MMR starting from S-rank, which requires crazy uncertainty values. This is the case because they can’t incorporate MMR from pre-S-rank,where there are rank floors.

#31

Of course they can. You can run both systems in parallel. You calculate MMR in the background. When the player reaches s-rank you reveal the mmr.

If the player has beaten powerful players to reach s-rank, he/she lands higher on s-rank. This way you also guarantee that the player has a certain number of games before mmr is important.

Unless decay/inflation is implemented tough, we still run the problem of people sitting on the top.

#32

Got ur spot 2 days ago, i climb to s at #43 won 1 game and i’m at 7 now. LOL

#33

I have PandaJJ on my friends list, everyone knows he is a very good player and just seing his match history. With a win streak btw, from the 34 last matches he played on ladder he got 32 wins vs 2 loses, the last time i talk with him he start at 100 or so, and until now he just move to 98 so far, like wtf…

#34

I’m confused, but from reading some replies. If I want to go S-Rank I just have to grind? Correct me if I’m wrong, new here.

#35

The process is really simple, you just have get to the rank 0 and you that by winning more games than you lose. If your winrate is just slightly above 50% it’s going to be a grind because you will need to play lots of games to get there. But despite the process being simple, not everyone can accomplish it, so you can’t just say “Oh I just need to play lots of games and I’ll get to S guaranteed”. But it is safe to say if you can get to rank 2 that you can also get to 0, just takes some patience.

#36

To add on, you need above a 50% winrate against diamonds in particular, which does take some skill. The absence of win streaks further cements the skill required to hit S.

#37

Also to aid though often not relevant at that level is win streaks. Still I do wonder what metrics is used to ascertain S rank because surely I am not 34.

#38

I won a match today and lost nearly 100 ranks. Since when is winning considered bad for ranking D: ?

#39

It’s important to make the distinction between your “skill”/expected win rate and your “luck”/spike win rate.

Due to the “chevron floor” and luck spikes, even with a 45% expected win rate of a “below average” Diamond player, if you play a lot (like 2,000 games, for example), then you still have a realistic chance (~18%) of making it; at an expected win rate of 48% and 1,000 games, you will make it ~44% of the time; at an even 50% expected win rate and 600 games, you can get there ~57% of the time; at 64% expected win rate, you can get there in 185 games or less almost every time (99+% chance).

#40

Well, for once i have to agree with you, i currently am at rank 9 in the s rank with a handful of games played where i won all of them, so according to the system i should be better than hsuku, unopro, zoochz, sylvermyst and whatnot, when i could only clean their shoes gamewise… Also if i will stop playing i will retain my rank so… Cp fix this please, as it is now the more you play the more you are penalized…

#41

Problem is that’s just a grind fest. Your ranking is mostly determined on just how much you play. Reverse those numbers and it’s more about skill.

Win +3 points, Loss -5 points

As for starting position. Game just tracks those numbers from the start, or once you hit diamond, so you have your win ratio roughly calculated.

But really ELO system is the best, because it basicly does that plus rewards or subtracts extra points based on who you beat/loose to. Win more for beating higher ranked loose more for loosing to lower ranked.

So take that +3 for win, and -5 for loss and now add +1 extra point if you beat someone higher rank then you. -1 point if you beat someone lower. -1 less for loosing to someone better and vice versa. Then their should be an equation to award/subtract extra points if there is a huge difference between your scores. But for the most part it should be working in much small numbers then it currently does.

Currently I think they have some sort of ELO system in place but it is a very poor one, as it is adding and subtracting wayyyyy to many points at a time, and seems to award a lot for win streaks.

Winning/Loosing streaks should have zero effect on amount of points gained. With ELO loosing will suck, but it should not ever drop you more then a rank or two. Where as you need a really good win rate to actually rank up.

Now the time it takes you to grind through diamond gives a fair estimate of your ladder placement. The ladder jumping around won’t be nearly as bad, and rating should more or less be accurate. And you can get there if you grind hard through bad luck, or just do really well with only some games. Your ladder placement should not change drastically wether you play 10 or 100 games, unless you legitimatly and significantly improve your self or fail to adapt to the meta.

1 Like
#42

lol, the current system is the ELO system but on crack. Win against a diamond/200+ player, stay the same rank or lose ranks (Had a 11 win streak over 24 hours and went down 20 ranks 80 to 100). Win against a top 100 you can actually gain some ranks. Losses put you down at least 10 ranks and quite often 50 or so if you were top 50. Since your ELO is also used to que you up against similar players once you fall to a certain point and queue only 150+ since everyone is camping, you are not climbing out of that hell hole. Then of course there are those placement matches that let any player entering S-Rank in the last 3 days a guaranteed top 50 spot if they win a game in S. A quick band-aid could be to remove placement matches entirely so that everyone gets screwed by the system equally