forum

Cumulative rank

posted
Total Posts
153
This is a feature request. Feature requests can be voted up by supporters.
Current Priority: +1,191
show more
Luvdic

AncuL wrote:

Xanandra wrote:

This means that my friend should be ranked higher than me in the cumulative rank, which makes sense as having 2 modes ranked at ~2000 is more impressive in my opinion.
the more mode you play, the better you are
My point is, if we were to use the cumulative rank based on pp, in my example given, I would have a higher rank than my friend because I have higher pp even though I only excel in one mode (whereas my hypothetical friend excels in two mode)

Bara- wrote:

Xanandra, that would favor #1-10 who play only one mode much more over those ranked #100-1000 in all modes
I am only giving out an idea, it can be fine tuned so that higher ranks wouldn't be rewarded so handsomely, and we would have to define starting at what levels should a player excelling in two modes have a higher rank than a player excelling a single mode, etc etc etc. To do all of these, tests and fine tuning will be required.

My point of my post is that using pp for a cumulative rank is NOT the way to go, unless a method to normalize the pp values between different modes on which I find it difficult.
Topic Starter
Bara-
Ohh
I get your point now
Catgirl
rank is also based on how many people play a mode. being #1000 in standard is much more impressive than being #1000 in taiko for example just because so many more people play standard. pp was used in my example earlier as a "skill metric" because it doesn't depend on what anybody else does with regards to ranking up, even though we know it's not the best system for skill.

if you use an extreme rank example, let's say you have 2 players. one of them only plays standard and mania and is #20k in both games. one plays only taiko and ctb and is #20k in both games. which player should be higher ranked? obviously the first player because standard and mania have larger playerbases. it's possible to get to those ranks in taiko and ctb with a double digit playcount, while reaching #20k in standard requires well over 3000pp. but if we just average ranks then both players are seen as equal.

using a pp system would make up for the difference in the number of players that play each gamemode, but it would need some normalizing if we wanted a perfect ranking system. however, that's not what pp is. all of the other rankings use pp so the best we can do for now is use pp in some way to balance all of the modes.
XEPCOH
Totally support this request!
yaaaay 20k pp club~
Luvdic

Catgirl wrote:

rank is also based on how many people play a mode. being #1000 in standard is much more impressive than being #1000 in taiko for example just because so many more people play standard. pp was used in my example earlier as a "skill metric" because it doesn't depend on what anybody else does with regards to ranking up, even though we know it's not the best system for skill.

if you use an extreme rank example, let's say you have 2 players. one of them only plays standard and mania and is #20k in both games. one plays only taiko and ctb and is #20k in both games. which player should be higher ranked? obviously the first player because standard and mania have larger playerbases. it's possible to get to those ranks in taiko and ctb with a double digit playcount, while reaching #20k in standard requires well over 3000pp. but if we just average ranks then both players are seen as equal.

using a pp system would make up for the difference in the number of players that play each gamemode, but it would need some normalizing if we wanted a perfect ranking system. however, that's not what pp is. all of the other rankings use pp so the best we can do for now is use pp in some way to balance all of the modes.
Yes, it is true that some modes has a higher player base than others, but at least when you're comparing by ranks, you're comparing them by something consistent which is by how far you are from the #1 in the specified game mode.

In any case, I did a little research, here are the amount of pp that players at rank #10000 has in each mode:

Standard: 4287
Taiko: 1816
CTB: 992
Mania: 2784

And here's the amount of pp of players in rank #1000

Standard: 6548
Taiko: 4744
CTB: 3608
Mania: 7132

See how erratic it is? If we were to use pp, the player that is #1000 in taiko and mania will be better than the player that is #1000 in standard and CTB (hey! I thought you said that standard weighted more because of higher player base! guess not). On the other hand, if you look at them as being 1k players away from being the top of two modes, then they are ranked the same, which makes more sense to me.

Anyways, I didn't say it in my earlier post, but I do agree with this feature too! It would be interesting to see
Catgirl

Xanandra wrote:

And here's the amount of pp of players in rank #1000
Standard: 6548
Taiko: 4744
CTB: 3608
Mania: 7132
See how erratic it is? If we were to use pp, the player that is #1000 in taiko and mania will be better than the player that is #1000 in standard and CTB (hey! I thought you said that standard weighted more because of higher player base! guess not).

I think you need to check your numbers again on that. Never mind, you kinda went backwards with what I originally said so I was confused. The more accurate statement would be Standard has a larger playerbase, but mania has more "good players" (based on pp). You have a point, but it's more because mania's pp system is just scaled really poorly at all levels. There are a bunch of players outside the top 2000 in mania who can do 400pp scores. Show me that in standard. It just doesn't happen. You'd have to scale pp somehow to make it more consistent across all modes.

Xanandra wrote:

On the other hand, if you look at them as being 1k players away from being the top of two modes, then they are ranked the same, which makes more sense to me.

Slightly exaggerated, but saying you're rank #5000 in a game with probably 200,000 active players (osu!) is a lot more impressive than saying you're rank #5000 in a game with 2,000 active players (taiko/ctb). Probably bad estimates but hopefully you get my point.

I don't think there's a perfect way to compare across modes but even an imperfect system is better than no system, so I'm still all for this no matter how it gets implemented, whether by rank, pp, or a mix of the two.
Luvdic
I think you need to check your numbers again on that. Never mind, you kinda went backwards with what I originally said so I was confused. The more accurate statement would be Standard has a larger playerbase, but mania has more "good players" (based on pp). You have a point, but it's more because mania's pp system is just scaled really poorly at all levels. There are a bunch of players outside the top 2000 in mania who can do 400pp scores. Show me that in standard. It just doesn't happen. You'd have to scale pp somehow to make it more consistent across all modes.
[/quote]

Hahaha sorry, my bad for not putting it in the best way possible.

I think the source of our disagreement is precisely that, what I am trying to say is that you cannot compare 400 pp obtained in standard and 400pp obtained in mania. Although yes, a rescaling could work too, by using something like... how many people have over XXXXpp in each mode and use that as the scale. It would be interesting to see the different results they yield and see which one seems to be more reasonable haha.
Edgar_Figaro
+1 this idea as I was thinking of a similar idea myself but saw this. I love all the game modes and have always found it a pity so many people play 1 and never try the others. Hopefully a feature like this would help to expand the other modes coommunities
Swerro
+25 votes, really like this idea!

Makes me think of this website http://osu.dawnglare.com/?p=totalpp

I'd love to see countryranking in dutchland for this
Topic Starter
Bara-
That site is cool
I just wish it would go further than top 1000, I can't find myself >__>
Remyria

Xanandra wrote:

Hello.

So, I haven't read the whole posts and discussion so far, only the last page and the first post, and from what I can see at glance is that the usage of pp is flawed if we are to calculate the cumulative ranks through all game modes. The main reason being that pp gained through the four modes are NOT normalized.

I just wanted to point out the the possibility of using the ranks in each mode as the measure instead of pp at each mode, meaning this:

CP (cumulative points) = 1/(rank in standard) + 1/(rank in taiko) + 1/(rank in ctb) + 1/(rank in mania)
If any term results in 0 because of no rank achieved in that mode, it should default to 0
And then rank players based on their CP.

I think this way it will help normalize a bit the parameters as well as rewarding more to those players that has a high rank in a specific mode.
Ex:

I have the following ranks:
standard: 2000 (~5800pp)
ctb: 10000 (~1000pp)
taiko: 30000 (~800pp)
mania: 0
CP = 0.0006333
Total pp = 7600

My friend has the following ranks:
standard: 0
ctb: 2000 (~2300pp)
taiko: 2500 (~3500pp)
mania: 0
CP = 0.0009
Total pp = 5800

This means that my friend should be ranked higher than me in the cumulative rank, which makes sense as having 2 modes ranked at ~2000 is more impressive in my opinion.

Anyways, it should be polished a bit more as people in the top 10 will be weighted too good, etc etc etc. But my point is that using pp as the parameter is not the way to go for this. (Maybe setting it so that it is 1/(100+rank in mode) can help offset a bit, but idk, will require real testings to fine tune it.
In my opinion, that doesn't work, since the reason one is ranked 2000 in CTB with like 2300 pp and you're 2000 in standard with 5800pp is because there is way less players, I am WAY better at standard than CTB, but I'm 2400 in CTB and 7000 in standard, that woul make more sense to give me more credit for my standard than CTB, since I'm better at it.
Alpgh367
Really like this idea! I hope it gets implemented soon :)
LastExceed
This could actually be done by anyone with the osu!api and wouldn't even take that long, I guess I'll give it a try myself later this day
Frozen_Rhythm

LastExceed wrote:

This could actually be done by anyone with the osu!api and wouldn't even take that long, I guess I'll give it a try myself later this day

Cool!, I'll wait for it.
Edgar_Figaro
I gave this idea an additional 10 stars 😀. Tbh can't think of a feature I'd rather have more. I love playing multiple modes and would love to see as an all-rounder how I stack up. Just to put here currently standard: 2735, Taiko 4504, CTB 2000, & Mania 3384 = average 3175.55
Edgar_Figaro

Catgirl wrote:

using a pp system would make up for the difference in the number of players that play each gamemode, but it would need some normalizing if we wanted a perfect ranking system. however, that's not what pp is. all of the other rankings use pp so the best we can do for now is use pp in some way to balance all of the modes.
Yeah I completely agree that using rank wouldn't work. As unbalanced as PP is across different game modes (i.e. Taiko giving way too much at low difficulty maps and mania becoming crazy at high difficulty 7k maps) Rank would be altogether broken. Also if you just averaged ranks no one would ever be #1 as no one is #1 in all 4 Gamesmodes. That and getting the first 10PP in any game mode gains you over a million ranks while at high levels gaining 10PP gains you a couple
Luddy
I really do like the idea of this, it would get players encouraged to play other modes also, so they can look and be better all round in the game. It would also be nice along with this for there to be a new leader board for this rank. But I feel like there could be a few issues with creating a cumulative rank as people could begin to see it as the only rank that matters mostly, plus PP is unbalanced between game modes when compared, and I don't think anyone would like having PP pay out being modified for certain game modes, talk about getting some people mad.
Adder
Would be fun, more thing to farm for the bored ones.
vsgv3
I thought something similar, you beat me to it :D
Ryuo
Nice idea. It might encourage people to try out the other modes. Who knows, they might even be good at them.
Edgar_Figaro
I thought of a slight tweak that might make this cumulative rank a bit more accurate to player skill.

Very early on in each game mode PP is practically free. Especially with the high return at low playcount from the 416PP bonus as well as just filling out your top 100 plays grants alot of easy PP. Now I'd say it's a different threshold for the "Easy PP" in each game mode but I think someone who gets 4000PP in 1 game mode should be higher in a cumulative rank than someone who has 1000 x 4 game modes as it takes more effort and skill. By the same token someone who is 3000 x 4 should definitely be higher than 4000 x 1.

So my proposal would be that in each game mode (Can be different per game mode) below a certain amount of PP it gives dimenishing returns towards your cumulative average. So lets say at 1000PP it's only worth 500PP towards cumulative (0.5x multiplier) at 2000PP it's worth 1500PP (0.75x multiplier) & 3000PP+ would give full value from then on out. These are just example numbers and the multiplier would scale linear. Granted this wouldn't solve the issue of some game modes giving ridiculous PP at the highest ranks. But it would prevent people getting high cumulative rank for very little effort in other game modes.

Let me know your thoughts on this and also if people have any ideas for what the dimenishing factor for each game mode should be ect.
Swerro
This would also (maybe) encourage people to play other gamemodes, as most only play 1 gamemode. Especially for CtB and Taiko, and those could really use that.
I used all my votes again on this feature

I thought of a slight tweak that might make this cumulative rank a bit more accurate to player skill.
This is a really good point you make.

However what I would suggest is a scaling that is similar to the topranking systems in osu!

Your total pp for your best gamemode weighs 100%*
Your total pp for your second best gamemode weighs 75%*
Your total pp for your third best gamemode weighs 50%*
Your total pp for your fourth best gamemode weighs 25%*

*I just took random numbers

However, this still doesn't really fix the problem that for example it would be easier to get 300pp on ctb than on mania, but....,

On the other hand: This would also (maybe) encourage people to play other gamemodes, as most only play 1 gamemode. Especially for CtB and Taiko, and those could really use that. I also support this fact alot.
You'd be surprised on how many people would play more CtB and Taiko after they see how easy it is to get pp on that and raise their cumulative rank (ranking up is fun and addicting after all).
Edgar_Figaro

Swerro wrote:

However what I would suggest is a scaling that is similar to the topranking systems in osu!

Your total pp for your best gamemode weighs 100%*
Your total pp for your second best gamemode weighs 75%*
Your total pp for your third best gamemode weighs 50%*
Your total pp for your fourth best gamemode weighs 25%*
This kinda goes exactly against what I was suggesting in my post as a scaling system like this would favor the player who was skilled in one mode vs multiple. For instance a player who has 10K + 1K +500 + 500 with my system would be given less than a player who had 3000 x 4 compared to a straight add system where they'd be equal. In the system you proposed player 1 would get 11125PP while player 2 would get 7500PP.

My proposal was aimed at Lessening the odds of someone who is top ranked In a single gamemode see cumulative rank get added to the game. Quickly farm 1k in each other mode and jumping over a ton of players.
Swerro
This post is somewhat a summarization of imbalances in cumulative rank

It seems that many people would want to see their cumulative ranking and be a part of it! This is an example of how a cumulative rank would look like on a userpage, I didn't make it as outstanding as the singular gamemode tabs as I read that most people see this as an 'interesting' or 'fun' rank to see, not as a needed feature. So there's no reason to make this a big thing/tab.



Here I made a list containing statistics about the rest of the userpage

Performance: Big problem as Edgar_Figaro mentioned, needs alot of discussion and balancing, I myself don't have enough knowledge to go into this too deep

Performance graph: Has no problem as far as I see. It just shows the summed up ranks you gain and lose every day

Ranked score: Scoring is way different in every gamemode.

Hit accuracy: Taking the average acc of all gamemodes wouldn't work, as it's easier to SS in CtB than in other gamemodes. Also OD <-> PP curve relation plays a role/problem in this considering taiko/mania/std acc differences

Play Count: I see no problem in adding up all playcounts in each gamemode

Play time: I see no problem in adding up all amounts of hours played in each gamemode

Total score: Scoring is different in each gamemode

Current level: Leveling is different in each gamemode

Total hits/fruitscatched: Mania has way more hit objects than std for example. Also problems with CtB etc.

Maximum combo: No huge problem but this would be imbalanced (consider converts, and non-converts on maps)

Kudosu: I see no problem in adding this up. I think a mod on each gamemode takes about equal effort - nvm I don't know I have no experience in modding

Replays watched by other: "there are less ctb players than std players so you'll have less replays watched on that" . ok yeh that's true I guess

Ranks: It's easier to get an S on mania and ctb as you can still miss and get an S rank. It's easier to SS on CtB. Some imbalances

To be honest I myself don't really care about some of the small imbalances as most people see this as an 'interesting' or 'fun' rank to see, I just wanted to point this out because.. idk
mulraf
liking this idea and agreed to the points of the last comment. except for:

Replays watched by other: "there are less ctb players than std players so you'll have less replays watched on that" . ok yeh that's true I guess
but less people are playing it, so you are more likely to get a higher rank on maps. and higher in ranking -> more likely to be seen by people. i myself mainly am a osu!std player. like 5x as much as all other modes alltogether. and still i have nearly as much replays seen in the other modes because in std i get like place 50 in a map. nobody's interested. no replays watched. in ctb there's less people -> easy to get into top 10 of less played maps -> seen by others.

i think the little differences won't be to much of an issue too :) the main thing people will be interested will be pp anyways. and i like the idea of having a small tab on the side. not to big for the people who don't really care that much, yet an extra tab for the ones who do.
Swerro
A Quick obvious example look for osu!next website
You could argue that you it should say 'osu!all' or just a '*' or 'combined' . I say we leave that decision for the osu!website designer

I can't see this feature hurting people for the following reasons:

About balances: The individual gamemodes, and so also this cumulative ranking, will never be perfectly balanced, however, everyone has access to all gamemodes, all beatmaps, so the phrase: "I don't play ctb and it's much easier to get pp in ctb than in std" is very invalid, as >"I don't play ctb" and >"caring about their cumulative ranking" is contradicting. Which is also the reason why I now think that pp scalers are unecessary and too subjective/difficult to implement regarding the 'early pp is free' discussion. (then take the easier early pp if you find your allmode rank matters!)

About having another tab on your userpage: People that 'don't like seeing a bad rank on their userpage' wouldn't mind a cumulative ranking in the same way as std players don't mind their ctb ranking. Besides, cumulative ranking will be not competetive (as in tournaments), and not be a huge important thing on userpage, yet very interesting for a huge group of players who play multiple modes. Respect for other people's interest ;p. Note that this feature almost has 1000 votes.

And, I know I mentioned this many times, the most important effect that this will have: More people playing different gamemodes. That's something that does very good to osu!

I'm all in for this feature+

Ignore spelling/grammar mistakes
Edgar_Figaro

Swerro wrote:

About balances: The individual gamemodes, and so also this cumulative ranking, will never be perfectly balanced, however, everyone has access to all gamemodes, all beatmaps, so the phrase: "I don't play ctb and it's much easier to get pp in ctb than in std" is very invalid, as >"I don't play ctb" and >"caring about their cumulative ranking" is contradicting. Which is also the reason why I now think that pp scalers are unecessary and too subjective/difficult to implement regarding the 'early pp is free' discussion. (then take the easier early pp if you find your allmode rank matters!)

Ignore spelling/grammar mistakes
I still think balancing should be done in some sort of regard so that it recognizes actual all-mode players over 1 mode players. With the most recent change to mania PP, Jhlee currently has more PP in mania than the combined PP of the #1 player in any other 2 modes. If you look at the dawn glare Osu website he is currently in 4th place on cumulative despite having 0PP in the other 3 modes. (seriously hoping they balance mania PP as it's currently over double of the other modes which makes cumulative ranking extremely skewed to heavily favor the mania players)
Swerro
[quote=Edgar_Figaro]

I still think balancing should be done in some sort of regard so that it recognizes actual all-mode players over 1 mode players. With the most recent change to mania PP, Jhlee currently has more PP in mania than the combined PP of the #1 player in any other 2 modes. If you look at the dawn glare Osu website he is currently in 4th place on cumulative despite having 0PP in the other 3 modes. (seriously hoping they balance mania PP as it's currently over double of the other modes which makes cumulative ranking extremely skewed to heavily favor the mania players)[/quote]

Yes I definitely agree with you on that. If the osumania pp get rescaled (this is already on peppy's to-do-list), if all 4 gamemodes should give about the same amounts of pp (13k for top players). That'd be perfect. Balanced enough for me.
I do want to point out to people that std ranking, taiko ranking, ctb ranking, mania ranking are and will always be at least somewhat imbalanced. The same thing for cumulative ranking.

I dislike the thought: "well it's not completely balanced so we won't implement this yet".
I'm strongly convinced (and you too probably) that this feature will bring way more good than bad
Edgar_Figaro

Swerro wrote:

I dislike the thought: "well it's not completely balanced so we won't implement this yet".
I'm strongly convinced (and you too probably) that this feature will bring way more good than bad
Oh definitely I am a major proponent for this feature being implemented. I have put many stars into this request. I think out of all the feature requests this is the one I want implemented the most.

I am more adding my thoughts here to kinda point out issues to possibly be discussed and worked on if this feature does get implemented. I am kinda curious if there is any way of knowing how likely/unlikely a feature is to get added. Like I know this request has over 900 star priority which is good but not by any stretch the highest. But I am not sure how much star priority a feature would need before it would be seriously considered being added. Obviously some requests will never be added no matter how much people want them due to technical, legal, or against the purpose of the game.
Tanomoshii Nekojou
I think this will be a bad idea for 1-2 mode players... But still a good idea for showing who's the best for all of the gamemodes... :D
Scarlet Evans

Tanomoshii Nekojou wrote:

I think this will be a bad idea for 1-2 mode players... But still a good idea for showing who's the best for all of the gamemodes... :D
I think it's kind of the opposite - this will be motivating for players to play other mods too! :) Main division into ranking systems depending on the game mode will remain anyway, so I think that there are no real upsides from having it and it would definitely benefit community and players!

But after you said that, I think that Ranking boards could be even more improved, to please 2-3 mode players too, i.e. aside of the Total (Cumultative) Rank, there should be an option to simply "filter" Ranking Boards with some criteria, where the most basic ones we can think of are: game modes.

What I mean here is allowing to choose which modes exactly we want to see, so someone could check top ranks from mania+CTB, taiko+CTB+mania, std+mania etc. Of course, you would be able to check your own rank too.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

When I get some stars in future, I need to put them here :3
bsf
Good idea!
Tanomoshii Nekojou

Scarlet Evans wrote:

But after you said that, I think that Ranking boards could be even more improved, to please 2-3 mode players too, i.e. aside of the Total (Cumultative) Rank, there should be an option to simply "filter" Ranking Boards with some criteria, where the most basic ones we can think of are: game modes.

What I mean here is allowing to choose which modes exactly we want to see, so someone could check top ranks from mania+CTB, taiko+CTB+mania, std+mania etc. Of course, you would be able to check your own rank too.
Wow. This one is also a good idea... :D
Edgar_Figaro
So I've been thinking about this whole cumulative rank a lot and also the complaints of people that don't want it showing (don't know why it would bother them but OK) Rather than just doing a straight up addition (which puts all Mania players on top) or doing it by rank in gamemode (give huge advantage to CTB & Taiko which have less players) I propose the following calculation formula.

(Standard PP + Taiko PP + CTB PP + Mania PP) - (standard deviation * 2) = Cumulative PP ranking

Basically by doing it this way it helps to mitigate some of the imbalance in each of the modes PP calculations from one another. It also has the benefit of making so 1 mode players never get ranked as if you have 100% PP in a single mode the 2* Deviation brings you exactly to 0PP. Moment you get PP in a 2nd mode you'd get added to the cumulative rank leaderboard.

This would encourage players to play all modes to increase their cumulative rank (best way to improve it is by playing the mode you have the least PP in)
While still allowing 2 & 3 modes players to be ranked and increase in cumulative ranking but just having a penalty for not playing all the modes.

Let me know what everyone thinks of this suggestion.
Megumj
It sounds interresting
Edgar_Figaro
Just to give Some Sample Numbers for Comparison with my suggestion for calculation.

Player A: Standard 1000PP, Taiko 1000PP, CTB 1000PP, Mania 1000PP = 4000PP Total - 2(0 PP Deviation) = Cumulative 4000PP
Player B: Standard 2000PP, Taiko 0PP, CTB 0PP, Mania 2000PP = 4000PP Total - 2(1154.7 PP Deviation) = Cumulative 1690.6PP
Player C: Standard 3000PP, Taiko 500PP, CTB 500PP, Mania 0PP = 4000PP Total - 2(1354 PP Deviation) = Cumulative 1292PP
Player D: Standard 1500PP, Taiko 1000PP, CTB 1500PP, Mania 0PP = 4000PP Total - 2(707.1 PP Deviation) = Cumulative 2585.8PP
Player E: Standard 3000PP, Taiko 3000PP, CTB 0PP, Mania 2000PP = 8000PP Total - 2(1414.2 PP Deviation) = Cumulative 5171.6PP
Player F: Standard 10000PP, Taiko 0PP, CTB 0PP, Mania, 0PP = 10000PP Total - 2(5000 PP Deviation) = Cumulative 0 PP
Emerold
being a little bit late a have to support this idea ^^
This might give "allround" players a little bit of recognition, however there might be a problem with balancing of pp numbers. With modes like mania going crazy there might be a unfair advantage for mode veterans of some kind. However, still supporting this idea. A average rank would be really cool to see.
Tanomoshii Nekojou

Edgar_Figaro wrote:

(Standard PP + Taiko PP + CTB PP + Mania PP) - (standard deviation * 2) = Cumulative PP ranking
Nice. :) :) :)
Edgar_Figaro
btw if anyone wants to see how this formula plays out as an example, I made a page t/566349 that implements this formula to figure out the top 50 all-modes players. Obviously I can't do a Cumulative rank for the entirety of Osu but would definitely be possible for Cumulative rank to be implemented with this being the way of deciding cumulative rank.
Adri
You can find a leaderboard of best users with all pp combined, doesn't mean much but it's always that.

https://osudaily.net/ranking.php
show more
Please sign in to reply.

New reply