Guest
home
beatmaps
rankings
community
store
help
forum
  1. Forums
  2. osu!
  3. Gameplay & Rankings

Performance Points feedback and suggestions (Standard)

posted
Total Posts
2,750
show more
RaneFire
1,181 posts
Joined May 2011
RaneFire

Novixion wrote:

Is the new star system out? I noticed the star rating on some maps have already changed...
...
And I thought 5.00+ are hard to find...
I saw everything become [N], for normals, for a couple of minutes only 2 days ago. I was about to say "I can't even pass normals now" but then it was fixed to represent E/N/H properly with lower star values. Changelog says this was changed on the 27th Feb though, I guess there was some delay.

Yeah very few maps are above 4 stars now, it's practically only the maps on the first page of "sort by difficulty."

IamNotgod wrote:

Literally 7/10 of his top performances were non full combo DTs before the most recent update. I'm not complaining that the updates made him lose rank. In fact, he barely lost any. All that I was saying is that his new top performances don't seem to show what he is good at. Now it shows some easy-medium difficulty FCs with only average accuracy.
I just made an assumption based on the zero info you gave. I don't know who he is, or what maps he's been playing. All I meant by "unless you're jesus" was that you would need some crazy speed just for the pp system to recognise your play if it is a low-acc, non-fc. General rule of thumb, as Tom has explained many times, is that it rates good performances first and foremost, not bad ones.

Hypothetically I could only play DT on insanes and get 60-70% all the time, and then play an easier map and FC it and go WTF. I think it's a matter of what you think you can do, but the system doesn't see it that way.
Keeby
50 posts
Joined July 2012
Keeby
It would be nice to implement the star system in-game, if possible.
Tom94
osu!dev
750 posts
Joined August 2012
Topic Starter
Tom94

Keeby wrote:

It would be nice to implement the star system in-game, if possible.
All planned.
george111cz
7 posts
Joined June 2013
george111cz
I have a question: Did something change in the past few days about calculating pp? I have noticed, that i'm not getting any pp for songs that i finish without full comboing them. So basically after few hours of trying i finally beat a song that is difficult for me, only to see +0 pp. But then i full combo some relatively easy song on Nightcore (+HardRock/Hidden, depends), where the main problem is not to fall asleep during playing and i'm rewarded with at least some pp.

I remember, that finishing difficult songs (for your current level of skill/pp) was always highly rewarded. Was this changed?

The second explanation is, that my performance points simply caught up with me and now i'm where i'm supposed to be (which would, honestly, be quite sad :-))

Well, either way, thanks for any responses in advance and have a nice day everyone :-)

EDIT: Just tried it, played http://osu.ppy.sh/b/46004 on NC, first try, not even knowing how the beatmap actually looks: +1 pp Not even S rank, 93% accuracy. For something like this i'd expect to get +0, seems weird to me.
Commy
6 posts
Joined August 2011
Commy
rank is not showing correctly for a long time, for example, it shows rank 1000 ingame but on site (rankings) it is 980, whats wrong? o.o
Yano
191 posts
Joined November 2012
Yano

Commy wrote:

rank is not showing correctly for a long time, for example, it shows rank 1000 ingame but on site (rankings) it is 980, whats wrong? o.o
Rank isn't updating realtime now (Web)
Commy
6 posts
Joined August 2011
Commy

Ultrayano wrote:

Commy wrote:

rank is not showing correctly for a long time, for example, it shows rank 1000 ingame but on site (rankings) it is 980, whats wrong? o.o
Rank isn't updating realtime now (Web)
well, I dont think it is caused by updating because every single player (rrtyui, www and the others in top5 (?) no) has wrong rank written ingame, on site (profile), you can be #50 but it will still show #53 even if you never had this position
JappyBabes
666 posts
Joined February 2011
JappyBabes

Commy wrote:

well, I dont think it is caused by updating because every single player (rrtyui, www and the others in top5 (?) no) has wrong rank written ingame, on site (profile), you can be #50 but it will still show #53 even if you never had this position
hakurei/shizuru/others that have temporary (i think) bans still have their ranks calculated etc
GoldenWolf
2,983 posts
Joined May 2012
GoldenWolf
some players are hidden




Soinou
747 posts
Joined June 2012
Soinou
I know I must be annoying or everything, but ...



It's OD 10, there is lots of jumps, and it's not really the easiest map possible, but ...

It's not DT, so no pp.

Why ? ...

Edit :

Nah, OD 10 means nothing for ppv2 ...
GoldenWolf
2,983 posts
Joined May 2012
GoldenWolf
Because those maps aren't hard, OD10 isn't hard in itself when you only have 1/2 through the maps

Example: http://osutp.net/scores?pid=475021&s=3
All those maps are OD9.8/10 and pretty much the same accuracy (there are even SSes on the bottom), the acc value can change a lot depending on how hard the map actually is
Soinou
747 posts
Joined June 2012
Soinou
Well, i know that, but it's still hard for me, and pp was before something you gain when you improve.

I'm improving, because i'm doing scores i couldn't do before, and i'm not gaining anything.

It's like the game is saying i suck and i should stop playing, or go farm DT because now only DT is rewarded.
Novixion
92 posts
Joined July 2013
Novixion

Soinou wrote:

Well, i know that, but it's still hard for me, and pp was before something you gain when you improve.

I'm improving, because i'm doing scores i couldn't do before, and i'm not gaining anything.

It's like the game is saying i suck and i should stop playing, or go farm DT because now only DT is rewarded.
Before pp was farming all day.

With the amount of pp you have and the accuracy of those scores, you wouldn't get much pp. To get accuracy pp you need >99%, which is why DT, which grants aim and speed, gives more points.
GoldenWolf
2,983 posts
Joined May 2012
GoldenWolf

I'll assume they're mostly DT/speed based scores

The thing is, those HR scores you've set are far easier than all those DT scores, since the main gain of HR is accuracy, and those accuracies you've got being not so good, you won't gain much/anything from them
Soinou
747 posts
Joined June 2012
Soinou
Yeah, so in fact, it's all about DT, OR having 99% HR, which is kinda impossible for me right now.

So, in the end, i must DT, or do nothing at all ?

Pretty good ranking system, yeah
GoldenWolf
2,983 posts
Joined May 2012
GoldenWolf
Okay let's be honest, 97% OD10 is trash because it like OD8 SS, which is trivial (especially on those maps full of 1/2s / no stream)

This system rewards good performance, not crappy ones

If you can't get high accuracy, don't blame the system for not rewarding you; instead improve your accuracy and get some decent scores
Aqo
2,710 posts
Joined April 2010
Aqo
DT players saying it's too easy to get points with DT and too hard with HR
HR players saying it's too easy with HR and too hard with DT


hmmmmmmmmmm
Ziggo
121 posts
Joined March 2012
Ziggo

GoldenWolf wrote:

Okay let's be honest, 97% OD10 is trash because it like OD8 SS, which is trivial (especially on those maps full of 1/2s / no stream)
The difference is, that SS'ing OD8 gives a lot of accuracy points, while OD10 97% gives practically nothing.
RaneFire
1,181 posts
Joined May 2011
RaneFire

Novixion wrote:

With the amount of pp you have and the accuracy of those scores, you wouldn't get much pp. To get accuracy pp you need >99%, which is why DT, which grants aim and speed, gives more points.
This is a pretty good point.

In fact DT increases all the attributes, even accuracy, but not as much as HR does. HR will mostly increase accuracy, so with the way the system works at present, DT is pretty much the way to go, unless you can SS OD10 like thelewa.
GhostFrog
416 posts
Joined March 2013
GhostFrog

Ziggo wrote:

GoldenWolf wrote:

Okay let's be honest, 97% OD10 is trash because it like OD8 SS, which is trivial (especially on those maps full of 1/2s / no stream)
The difference is, that SS'ing OD8 gives a lot of accuracy points, while OD10 97% gives practically nothing.
I don't know exactly how pp handles OD or how it treats SS, but if you can consistently get 97% on OD10, then, assuming your hits are normally distributed and centered on the correct timing, you can consistently get > 99.9% on OD8. If 97% OD10 isn't giving comparable pp to OD8 SS (I don't know if it does or not), that's unfair to HR players.
Ziggo
121 posts
Joined March 2012
Ziggo

GhostFrog wrote:

I don't know exactly how pp handles OD or how it treats SS, but if you can consistently get 97% on OD10, then, assuming your hits are normally distributed and centered on the correct timing, you can consistently get > 99.9% on OD8. If 97% OD10 isn't giving comparable pp to OD8 SS (I don't know if it does or not), that's unfair to HR players.
Example of an OD8 map
RaneFire
1,181 posts
Joined May 2011
RaneFire

GhostFrog wrote:

If 97% OD10 isn't giving comparable pp to OD8 SS (I don't know if it does or not), that's unfair to HR players.
Problem with this... is that it's only a margin for error, and thus scoring.

People playing OD8 maps with high accuracy could very well be hitting as accurately as OD10 players, but it doesn't matter because it's within that margin anyway. HR increases other aspects as well, like circle size and AR, adding some stress to the situation, which could affect your accuracy unless you're completely comfortable with it. Not many players can read AR10 absolutely comfortably, even if they can perform well at AR10.

You can't look at it as though they only have a OD8 level of accuracy, even if that's what the scoreboard says. Even though it is kinda unfair at the moment, "how much" is debatable because any modification would be arbitrary and based completely on opinion. Silmarilen already mentioned that SS'ing YAMAGEN'S DEVILELIET - EYES OF DEVILELIET is much more difficult than any of his other scores, even though it's OD is lower. The HR players are rewarded in this case though. So the problem is obviously the opinion of "base accuracy skill."
GhostFrog
416 posts
Joined March 2013
GhostFrog

RaneFire wrote:

People playing OD8 maps with high accuracy could very well be hitting as accurately as OD10 players, but it doesn't matter because it's within that margin anyway.
People playing 180bpm stream maps could very well have the speed to play 240bpm stream maps. People playing a jump map nomod (or hey, with HR) could very well be hitting the exact center of every hitcircle and would still FC with CS13. It doesn't matter - a 180bpm stream map can't prove your ability to stream 240bpm and playing a CS4 map can't prove your ability to play CS7. While it's entirely possible some random person SSing OD8 would be getting high accuracy OD10 scores, playing OD8 doesn't test that and they deserve no extra reward for the *possibility* that they're better than their play demonstrates. If they want to be rewarded for good OD10 accuracy, they should find OD8 maps they can play with HR. It's unfortunate that HR will almost certainly make the map AR10 if the player can't read it, but the job of the ranking system isn't to determine whether or not the rules of the game make sense - it's to rate the performance of a player in a particular play.
RaneFire
1,181 posts
Joined May 2011
RaneFire
That's a bit extreme and doesn't really bear extra weight to the argument because they're unrealistic situations.

HR players are rewarded properly for their accuracy in certain cases.

Circle size barely factors into difficulty, unless they get really small. Speed is already considered fully and it doesn't change except with DT, which calculates a completely new difficulty.

Accuracy on the other hand, is not easy to measure. Ideally, everyone should be measured by the OD10 margin of error, and then you can extrapolate more accurate results. But it still won't be perfect, because mapping style can heavily affect your accuracy. Pattern complexity is coming... I hope.
GhostFrog
416 posts
Joined March 2013
GhostFrog

RaneFire wrote:

That's a bit extreme and doesn't really bear extra weight to the argument because they're unrealistic situations.
Then remove the extreme part of it. If you FC a CS4 map, don't you think it's reasonably likely you could have FCed it on CS4.2 (CS3 + HR)? If so, should you be given extra aim points for that possibility?

HR players are rewarded properly for their accuracy in certain cases.
Are you referring to the EYES OF DEVILELIET scores you posted? That map is OD7. OD9.8 is harder compared to OD7 than OD10 is compared to OD8. I ran the same calculation as I did earlier - getting 97% consistently on OD9.8 (assuming normal distribution, hits centered at perfect timing) means getting > 99.98% consistently on OD7.
RaneFire
1,181 posts
Joined May 2011
RaneFire

GhostFrog wrote:

Then remove the extreme part of it. If you FC a CS4 map, don't you think it's reasonably likely you could have FCed it on CS4.2 (CS3 + HR)? If so, should you be given extra aim points for that possibility?
Maybe 1 more point? You should get a better understanding of how human beings perceive their environment. The approach circle scales with circle size as well, and in some cases I find it harder to read a map with very large circles as opposed to smaller ones. The map is also mapped according to the circle size, otherwise you will have weird spacing. So all in all, it's about what is appropriate, and CS plays a very small role in difficulty so long as it's appropriate. HR should increase the difficulty immensely in cases where the CS becomes obfuscated and patterns can no longer be hit in the nomod fashion.

GhostFrog wrote:

Are you referring to the EYES OF DEVILELIET scores you posted? That map is OD7. OD9.8 is harder compared to OD7 than OD10 is compared to OD8. I ran the same calculation as I did earlier - getting 97% consistently on OD9.8 (assuming normal distribution, hits centered at perfect timing) means getting > 99.98% consistently on OD7.
Yes, but getting an SS on that map with OD7, or even just 99%, is many times more difficult than getting 97% on other OD10 maps because of the way it's made. If a map is not difficult to play in aim or speed, why should you get a full accuracy reward from HR?

Again... base accuracy skill... opinions.
Ziggo
121 posts
Joined March 2012
Ziggo

RaneFire wrote:

Yes, but getting an SS on that map with OD7, or even just 99%, is many times more difficult than getting 97% on other OD10 maps because of the way it's made.
Why do you want to compare this map with other maps? The point was comparing no mod and HR on the very same map. And the map you mentioned is really difficult with HR as well, so getting 97% with HR should easily be on a level with 100% no mod.
KaosFR
87 posts
Joined January 2011
KaosFR
How does the algorithm work for accuracy ? Does it compute an estimated unstable rate (or a range) ?
Full Tablet
2,542 posts
Joined September 2011
Full Tablet

GhostFrog wrote:

I ran the same calculation as I did earlier - getting 97% consistently on OD9.8 (assuming normal distribution, hits centered at perfect timing) means getting > 99.98% consistently on OD7.
What calculation method did you use?

If you interpolated this table http://www.mediafire.com/download/pgb55 ... te_v2.xlsx for the "99%confidence" value (guessing since it gives similar results to what you give), take into consideration that, for example, an unstable rate of 100 in a map of 1000 circles should hardly ever result in less than 93.75%accuracy on OD10. For comparing purposes I think low confidence percentages make more sense (since most records are from maps the players have retried several times).

KaosFR wrote:

How does the algorithm work for accuracy ? Does it compute an estimated unstable rate (or a range) ?
It uses a formula made by Tom94 http://pastebin.com/XDDgKEvw#_=_ (this pastebin is outdated though, it is probably somewhat different currently).
Last edited by Full Tablet , edited 1 time in total.
GhostFrog
416 posts
Joined March 2013
GhostFrog

Full Tablet wrote:

GhostFrog wrote:

I ran the same calculation as I did earlier - getting 97% consistently on OD9.8 (assuming normal distribution, hits centered at perfect timing) means getting > 99.98% consistently on OD7.
What calculation method did you use?
Something I assume is much less sophisticated than what the table uses. I found the standard deviation of hit timings that would give 97% accuracy on OD9.8 and found what percent of 300s that would give on OD7 (I basically counted 100s as misses instead to be on the safe side). If you're consistently getting that standard deviation, you should consistently be getting the corresponding OD7 accuracy, unless I did something wrong.
RaneFire
1,181 posts
Joined May 2011
RaneFire
Computers everywhere. Not one who plays the game.
silmarilen
5,579 posts
Joined February 2010
silmarilen
i personally find getting 97% on hr many times more difficult that getting (almost) SS on od8, but that's also because of the increased ar and circle size. usually if i can SS a map on nomod first try i get like 90-95% first try on hr (assuming i can play the map with hr, which i usually cant)
Soinou
747 posts
Joined June 2012
Soinou

Aqo wrote:

DT players saying it's too easy to get points with DT and too hard with HR
HR players saying it's too easy with HR and too hard with DT


hmmmmmmmmmm
The thing is that i'm not a DT player.

Well, sorry for my totally useless posts, it was just me trying to blame something for being bad, but well, I'll try to be a bit more clear about what I think even if nobody cares.

I think the problem with the current system is that it gives pp based on the difficulty of the maps.

However, what is difficulty ? How is it possible to accurately evaluate difficulty for every players ?

For example, like you all point out, my top scores are DT, but ... I find them really easy. I almost never practiced for DT, I just play DT from times to times and that's where I got these scores, which gave me a lot of pp because the system evaluates them as "hard", while they're kinda easy.

But, I got some HD scores only after a long long time practicing HD, and it took me several weeks training with this mod to have some scores like my 96% FC on Torikago, which is considered as bad by the system, because the map is not considered difficult, while this is very difficult for me.

The point I'm trying to make is that, with this system, you get rewarded doing something easy to achieve, and you get no reward doing something that took you a lot of training, and really hard to achieve.

I find this kinda sad.

But well, you'll probably say that it's just me trying to blame the system for being bad or something, or that I just suck, and i should stop/farm DT or anything else, so feel free to ignore me, delete my post or anything, I don't care.
Tom94
osu!dev
750 posts
Joined August 2012
Topic Starter
Tom94

Ziggo wrote:

GhostFrog wrote:

I don't know exactly how pp handles OD or how it treats SS, but if you can consistently get 97% on OD10, then, assuming your hits are normally distributed and centered on the correct timing, you can consistently get > 99.9% on OD8. If 97% OD10 isn't giving comparable pp to OD8 SS (I don't know if it does or not), that's unfair to HR players.
Example of an OD8 map
Hmm, that's not exactly intended though. I'll try to make it more appropriate in that direction.
nooblet
630 posts
Joined December 2011
nooblet

Soinou wrote:

SPOILER

Aqo wrote:

DT players saying it's too easy to get points with DT and too hard with HR
HR players saying it's too easy with HR and too hard with DT


hmmmmmmmmmm
The thing is that i'm not a DT player.

Well, sorry for my totally useless posts, it was just me trying to blame something for being bad, but well, I'll try to be a bit more clear about what I think even if nobody cares.

I think the problem with the current system is that it gives pp based on the difficulty of the maps.

However, what is difficulty ? How is it possible to accurately evaluate difficulty for every players ?

For example, like you all point out, my top scores are DT, but ... I find them really easy. I almost never practiced for DT, I just play DT from times to times and that's where I got these scores, which gave me a lot of pp because the system evaluates them as "hard", while they're kinda easy.

But, I got some HD scores only after a long long time practicing HD, and it took me several weeks training with this mod to have some scores like my 96% FC on Torikago, which is considered as bad by the system, because the map is not considered difficult, while this is very difficult for me.

The point I'm trying to make is that, with this system, you get rewarded doing something easy to achieve, and you get no reward doing something that took you a lot of training, and really hard to achieve.

I find this kinda sad.

But well, you'll probably say that it's just me trying to blame the system for being bad or something, or that I just suck, and i should stop/farm DT or anything else, so feel free to ignore me, delete my post or anything, I don't care.
You find the beatmaps with DT easy because you already have the skills and experience required to play it. Just because you didn't have to "work" for a certain score, doesn't mean it was an easy feat. Half of my top scores were done in less than 5 plays (complete fluke plays of course), while the other half I probably had to play up to probably 50 times over the course of a few weeks (Which is ridiculously high for me, I've never retried past 10 in one go).
The scores may have felt easy, but you wouldn't have been able to pull it off when you just started, right? Everyone has their own strengths and weaknesses, the system can't compensate for everyone individually. In other words, what's easy for you may not be easy for others.
Soinou
747 posts
Joined June 2012
Soinou

nooblet wrote:

You find the beatmaps with DT easy because you already have the skills and experience required to play it. Just because you didn't have to "work" for a certain score, doesn't mean it was an easy feat. Half of my top scores were done in less than 5 plays (complete fluke plays of course), while the other half I probably had to play up to probably 50 times over the course of a few weeks (Which is ridiculously high for me, I've never retried past 10 in one go).
The scores may have felt easy, but you wouldn't have been able to pull it off when you just started, right? Everyone has their own strengths and weaknesses, the system can't compensate for everyone individually. In other words, what's easy for you may not be easy for others.
Well, I know that the system can't compensate for everyone individually, but it feels kinda sad that the system rewards you for choosing simplicity over hard work.
GoldenWolf
2,983 posts
Joined May 2012
GoldenWolf
@soinou
You're missing something though; it's not because you find something easy that it is easy, same when you find something hard it doesn't mean it is hard.

Example: I find these slow ar8 maps hard since I can't SS them, mostly because I suck at reading stuff, but it doesn't mean they're any hard.
So,

Soinou wrote:

The point I'm trying to make is that, with this system, you get rewarded doing something easy to achieve, and you get no reward doing something that took you a lot of training, and really hard to achieve.
you may not always get rewarded for something you find hard, but you'll get rewarded for something that is hard, depends on what you're good/bad at.
Soinou
747 posts
Joined June 2012
Soinou

GoldenWolf wrote:

you may not always get rewarded for something you find hard, but you'll get rewarded for something that is hard, depends on what you're good/bad at.
Yeah, but being "hard" or "easy" depends on the player, so how can you judge the difficulty of a map for every players, and say a map "is" hard ?
Nyxa
848 posts
Joined October 2013
Nyxa

nooblet wrote:

You find the beatmaps with DT easy because you already have the skills and experience required to play it. Just because you didn't have to "work" for a certain score, doesn't mean it was an easy feat. Half of my top scores were done in less than 5 plays (complete fluke plays of course), while the other half I probably had to play up to probably 50 times over the course of a few weeks (Which is ridiculously high for me, I've never retried past 10 in one go).
The scores may have felt easy, but you wouldn't have been able to pull it off when you just started, right? Everyone has their own strengths and weaknesses, the system can't compensate for everyone individually. In other words, what's easy for you may not be easy for others.
This is true. I've managed to do good performances on some maps easily that some of my friends couldn't do at all, even if they can easily clear maps I have lots of trouble with (For example, I FC'd LeaF - MEPHISTO, something none of my other friends managed to do, but I'm nearly at the bottom of my friends list on Nico Nico Chorus - Leia, even though the map isn't that hard). Everyone is good at different things, some people are better with higher AR's, others with lower ones, some are HR players, others HD players, and others DT players. You can't really say "X mod is easier than Y mod", since it varies for everyone. I find HR easier than HD, because I have trouble reading HD and perform better on AR10, even though I'm sure many people think HD is easier than HR.

It's best to just calculate based on what the mod changes on the map (for example, in my opinion, HD should give a similar bonus to HR for aim, since you have to aim for disappearing notes, while HR should give a significantly larger bonus than HD for accuracy, since HD doesn't alter the map's accuracy settings at all). But I think this has already been done quite well, and I don't think anyone should be trying to get the algorithms to change based on what they find easy or hard. It's different for everyone.
GoldenWolf
2,983 posts
Joined May 2012
GoldenWolf

Soinou wrote:

Yeah, but being "hard" or "easy" depends on the player, so how can you judge the difficulty of a map for every players, and say a map "is" hard ?
Some things are objectively hard; like a 129 circles stream at 222bpm
Other things are subjectively hard; like low AR, old maps, squares, ...
+

-Scylla- wrote:

It's best to just calculate based on what the mod changes on the map (for example, in my opinion, HD should give a similar bonus to HR for aim, since you have to aim for disappearing notes, while HR should give a significantly larger bonus than HD for accuracy, since HD doesn't alter the map's accuracy settings at all). But I think this has already been done quite well, and I don't think anyone should be trying to get the algorithms to change based on what they find easy or hard. It's different for everyone.
show more
Please sign in to reply.

New reply

0
770 / 2,750