Rank hard stuck at Initiate 1?

[mtlancenese]

I understand how ranking systems work, I understand even me being in Initiate. I don't really care about the ranking but I have not even moved up to Initiate 2 for one game and then immediately went down. My total win rate is 45% from 60 hours of gametime. I played yesterday and won 4 out of 5 games and I didn't even bump up to Initiate 2? One of those games I carried with 30 kills and 70k souls. Even when my team loses, I am one of the top of the total souls collected. I'm not even this low of a rank in LoL and I barely play that game.

When I play in higher lobbies with friends, I still play well, positive K/D/high farm. I understand weighting wins/losses heavily but surely you can rank up without having to hyper carry 5 games in a row with 5 complete strangers.

 

[Leto]

Ranking seems more than a bit buggy at the moment, and I don't think fixing it is a priority for the dev team.

I take long breaks from the game, so my rank has reset a few times since the system was introduced. In previous instances I was always somewhere between high-alchemist to mid-arcanist. I haven't played since about February but after coming back a couple weeks ago, my rank has been permanently initiate 1 with absolutely zero change which strikes me as odd.

Maybe I'm just absolute dogwater, but it seems improbable given I've been playing other shooters and improving my aim and situational awareness in the time I was away from DL. I personally feel like I've been playing better than before, and the match stats reflect this, but my rank does not.

 

[Leto]

To add to this, I wanted to check out the match rank distribution. Not sure if we're allowed to link to any 3rd party tools so I'll just include screenshots of the charts.

Here's the lifetime rank distribution of matches:


And the last month:


Now this *could* be entirely due to new players, but it still seems kind of off? Unless there's thousands of people playing one match and never playing another, or only ever (and correctly) placing in initiate 1 and staying there, the second chart doesn't make sense to me.

But I guess time will tell ¯\_(ツ)_/¯

 

[nulltype]

That graph implies something is pretty broken. It sounds like @mtlancenese's win rate is fine though. @Leto do you have a win rate greater than about 75%? If not, I would guess that the displayed rank could be broken, but the actual underlying matchmaking rank could be correct (if the matchmaking rank was incorrect, it should reflect in the win rate).

 

[Leto]

Since I started playing again two weeks ago I have a 53% win rate, which is not too outlandish.

You're right that it may be my displayed rank that is incorrect, but that begs the question of what ELO metric is actually being used for matchmaking at the moment. I know a lot of this is obfuscated but if ranks were accurately being kept track of, why would it just be displayed incorrectly? And match ranks are based on player ranks, so you can't really have one be wrong and the other correct.

Could very well be that the current internal metric for matchmaking has a mismatch with the public ranking system, which would explain the discrepancy in the chart as well. I.e. maybe people are being skill matched correctly but a good amount of players AND games are being mislabeled "initiate 1" on the frontend.

 

[ExistingExpert]

It is not just about Initiate 1, every mmr bracket is insanely hard to climb up from. This is my win-loss trend for the past 200 matches I have played (Screenshot is my profile from statlocker).

This means that I have a 63% win rate over my last 200 matches. How much have I climbed from this? From Arcanist 2 at the beginning to Ritualist 3 now. And I am not the only one. It takes a lot of wins to climb up from one division to the next. If you say you need 4 wins to climb one subdivision and you are in initiate 1 and then you quarter that number to account for any winstreak mmr the game may or may not give you, that is still a 66 WINSTREAK that you need to get to climb to the highest ranks. Deadlock has the most amount of ranks of any similar game paired with a very slow rank mobility system. This brings us to:

Issue 1: So many good players are stuck in the lower ranks that the best Seekers-Alchemists-Arcanists-Ritualists are consistently playing as well as Emissaries-Archons-Oracles-Phantoms which makes it impossibly hard for actual Arcanists or Ritualists to climb higher ranks.
Issue 2: When the game matchmakes these "good" low elo players in the same team, it is a huge stomp. How many times have you been stomped by a team that is clearly much more skilled than you, but then you check the match screen and see that both teams are the same "rank"? These games are the worst, it is fun for neither the winning team and it is definetly not fun for the losing team. These are the games that make deadlock lose its playerbase.
Issue 3: When the game matchmakes these "good" low elo players against each other, the match is balanced and one of them is bound to lose, but then all this does is push the loser even farther than his actual rank, because the game does not take performance into account for its mmr calculations, it only cares about win/loss (and even that barely affects things). This just exacerbates the situation.
Issue 4: MMR decay is insanely overtuned. If you leave the game for a while then come back, the mmr system will place you down in the lowest depths of the ladder. You will stomp every game you come across (unless you face others like you) and it will take ages for you to climb back up.
Issue 5: 4 stacks, 5 stacks, 6 stacks just turn off the matchmaker and these teams get paired with literally anyone the game can throw at them. This means that the people playing in huge parties usually gain a massive advantage (no further words needed), along with perfect hero selection (one tricking is trivial).

TL;DR: Game has too many ranks but very low mmr mobility paired with very high mmr time decay, this makes good players stuck in lower ranks and they make situation worse for everyone including themselves. Couple that with huge party stacks and que the matchmaking complaints.

 

[lolihero14]

Yes, it's time to realize that the ranking system here is just for show. Since the game is still in a closed alpha test, it does not need to enter a ranking system yet, as everything will still be turned over several times. Just don't worry about the rank, but just play and have fun playing the game

 

[ExistingExpert]

Yes, it's time to realize that the ranking system here is just for show. Since the game is still in a closed alpha test, it does not need to enter a ranking system yet, as everything will still be turned over several times. Just don't worry about the rank, but just play and have fun playing the game

The issue is that the ranking system is NOT just for show. It is the system most influential in determining the outcome of every single game you play. The part that is "for show" is just the ranks it displays on your profile (which I believe to be wildly inaccurate for most people). Behind the scenes the matchmaker has placed so many good players in the lower ranks due to mmr decay and low rank mobility that the skill variation in lower elo games is unfathomable, even if the matchmaker promises that it is even. This makes games boring for the seasoned players and absolute hell for newcomers.

 

[nulltype]

> This means that I have a 63% win rate over my last 200 matches. How much have I climbed from this? From Arcanist 2 at the beginning to Ritualist 3 now.

That implies to me that it's working. Your win rate should approach 50% over a long enough time horizon.

> that is still a 66 WINSTREAK that you need to get to climb to the highest ranks

This doesn't make sense, if you win 3 out of every 4 games consistently, you'd have a win rate of 75% and that would imply the match maker is not working correctly.

> he game does not take performance into account for its mmr calculations, it only cares about win/loss (and even that barely affects things)

It's more noisy to only do win/loss, but it's more accurate than trying to measure performance. If win/loss did not affect things, then you would have some crazy 75% plus win rate.

 

[ExistingExpert]

That implies to me that it's working. Your win rate should approach 50% over a long enough time horizon.

Yes I agree that it is working, but I believe that it is not working fast enough. For a game with 66 divisions of ranks (compared to dota's and league's 40) it is just way too slow. If you want to climb you need a great win-rate, consistently, and you better not take any breaks ever to avoid mmr decay. I think it should both reward wins as well as penalize losses around 65% more to account for that many ranks.

This doesn't make sense, if you win 3 out of every 4 games consistently, you'd have a win rate of 75% and that would imply the match maker is not working correctly.

I meant that as an example to visualize how slow the ranking system is. 66 is my guess on how many straight wins someone with crippled mmr stuck in initiate 1 would need to get to eternus 6.
Is 75% win-rate the number when suspicion on the matchmaker arises? I think achieving that, fully solo queue, while combating phantoms+ who have rank decayed to your mmr, standing up against 4-5-6 party stacks, carrying new players in your team, and still dealing with smurfs/hackers (average moba experience), is a really tall ask just to climb a little higher.

It's more noisy to only do win/loss, but it's more accurate than trying to measure performance. If win/loss did not affect things, then you would have some crazy 75% plus win rate.

I do not understand why that is, but I would love to hit 75% win rate.
And here is the kicker, it only gets easier to hit that as you climb up. If you climb to archon from initiate (and this point may be even more relevant for the lower playercount regions such as asia and oceania), your games will just get easier. You will no longer have to contend with new players, or phantoms+ undergoing mmr decay, or new smurf accounts and you will breeze through your games.

 

[wsktke]

I understand how ranking systems work, I understand even me being in Initiate. I don't really care about the ranking but I have not even moved up to Initiate 2 for one game and then immediately went down. My total win rate is 45% from 60 hours of gametime. I played yesterday and won 4 out of 5 games and I didn't even bump up to Initiate 2? One of those games I carried with 30 kills and 70k souls. Even when my team loses, I am one of the top of the total souls collected. I'm not even this low of a rank in LoL and I barely play that game.

When I play in higher lobbies with friends, I still play well, positive K/D/high farm. I understand weighting wins/losses heavily but surely you can rank up without having to hyper carry 5 games in a row with 5 complete strangers.

Myself and a bunch of other people I know have tracked this. As long as you have an established account with 50 games within a 30 day period. You're on a +4 and -5 Win/Loss for ranking and deranking. If you said you went 4-5 in one day, then you arent going to rank up. You need a much more net positive win/loss in order to rank.

 

[wsktke]

To add to this, I wanted to check out the match rank distribution. Not sure if we're allowed to link to any 3rd party tools so I'll just include screenshots of the charts.

Here's the lifetime rank distribution of matches:
View attachment 58038

And the last month:
View attachment 58037

Now this *could* be entirely due to new players, but it still seems kind of off? Unless there's thousands of people playing one match and never playing another, or only ever (and correctly) placing in initiate 1 and staying there, the second chart doesn't make sense to me.

But I guess time will tell ¯\_(ツ)_/¯

The reason there have been an increase is because of the returning players.

 

[ExistingExpert]

Myself and a bunch of other people I know have tracked this. As long as you have an established account with 50 games within a 30 day period. You're on a +4 and -5 Win/Loss for ranking and deranking. If you said you went 4-5 in one day, then you arent going to rank up. You need a much more net positive win/loss in order to rank.

Is the data publicly available?

 

[nulltype]

Yes I agree that it is working, but I believe that it is not working fast enough. For a game with 66 divisions of ranks (compared to dota's and league's 40) it is just way too slow. If you want to climb you need a great win-rate, consistently, and you better not take any breaks ever to avoid mmr decay. I think it should both reward wins as well as penalize losses around 65% more to account for that many ranks.

Ah yeah that makes sense. In my experience though the displayed rank does move pretty quickly, I play basically only a single hero and my displayed rank has moved across a range of about 10 levels (up and down) despite my skill at this hero not changing that much.

As I mentioned it's more noisy to do win/loss, so it can take awhile for your rank to be accurate. I don't know if Valve uses something like TrueSkill, but if you look at the table here: https://trueskill.org/ it takes a large number of matches to estimate someone's skill accurately in a 4:4 setting (and their skill may change over time of course).

I meant that as an example to visualize how slow the ranking system is. 66 is my guess on how many straight wins someone with crippled mmr stuck in initiate 1 would need to get to eternus 6.
Is 75% win-rate the number when suspicion on the matchmaker arises? I think achieving that, fully solo queue, while combating phantoms+ who have rank decayed to your mmr, standing up against 4-5-6 party stacks, carrying new players in your team, and still dealing with smurfs/hackers (average moba experience), is a really tall ask just to climb a little higher.

It might be hard to estimate if the rate of change increases with length of win streak. 75% is an arbitrary threshold, but it should be pretty hard to get 75% or 25% if the matchmaker is working well.

Those things you name do introduce noise, but I suspect most (especially in solo queue) also affect the enemy team almost the same as your team.

For example, maybe you're losing a lot of games due to leavers, but since you're not a leaver, there's 4 people on your team who could be leavers and 5 on the enemy team, so it's slightly more likely that the leaver will be on the enemy team (non-leavers would end up with some bonus to rank as a result). Smurfs and hackers would work in the opposite direction, but they would affect other individual players as well so it's probably just that everyone else's rank is slightly lower than it would be without those people, and the skill estimates have more noise.

I do not understand why that is, but I would love to hit 75% win rate.

Unfortunately, without adding bots to the game your long term win rate will be about 50% if the matchmaker is working well. Fortnite added bots though, so that's a way for people to get a higher win rate than 50%.