1. osu! forums
  2. osu!
  3. Feature Requests
This is a feature request. Feature requests can be voted up by supporters.
Current Priority: +25
posted
When scrolling casually through my massive list of beatmaps to arbitrarily determine which song I want to play, I often look for songs where I know I can improve the most on.

Say there are two easy songs, both of them with #1 score 4.4 mil. Say I have 4 mil points on one of them, and 1 mil points on the other. I would obviously choose to play the 1 mil one in order to earn an easy gain of 3 mil points.

The natural way to look for this "Most Improvable" song is to "Sort by Rank" which is by letter grade, but unfortunately, this distinguishes not from an S where I slider-broke somewhere in the middle to get 199 300s and 1 100 but only got half the maximum possible score, and other Ss that are FCs close to perfect.

The remedy then, is to "Sort by Online Rank." However, this has been denied multiple times in the past due to stress on the server.

Fortunately, this is not what I'm looking for. Rather, I'm interested in a sorting by a new statistic that would be stored locally, one that calculates for each song your best relative score. This "relative score" would be a percentage of your score to the score that would have been achieved if you had SSed the song with no mods. (Note that this would mean your "relative score" percentage could be higher than 100%.)

So if your best ranked score on a song where a no-mod SS gets you 4 mil is 3 mil, then this relative score would be 75%. If I used a mod with a 2x score multiplier and still got an SS, I would have 8 mil points, and my relative score would then be 200%.

Note that this "no-mod SS" score is a static number, and would not need online database access to use and would never need to be updated based on the scoreboard. Unfortunately, songs may not currently have this statistic available so some method of calculation would have to be implemented in the beginning. However, it should be a one-time thing and all future usage should be straightforward.

Thus all the FCed songs would be relatively high, in the 95%s, even if one had poor accuracy, while the songs where one can still improve greatly would be in the low percentages, and would still provide an accurate standard of measure to the top ranks in the song.

While this is easy to do manually one at a time by clicking on each song individually, I feel that being able to sort the entire list automatically would definitely be handy. Additionally, the greater reason for this suggestion lies in multiplayer. Always playing in a group with my friends, we tend to choose to avoid songs where one of us already has a score too high to improve upon, and having this number available rather than merely S or A would greatly help determine which songs were available for play.
posted
why can't u just sort by rank?
posted
I'm not very sure about this one. Either it'd take a lot of disk space, or require lots of processing each time you wanna sort, since it has to calculate the max score of each map, and take your highest score from that amount.

Also, for sort by rank, all S's are the same, and all SS's are the same...... So they still sort alphabetically if they have the same rank.
posted
Sounds useful, I would use this a lot X_X

+1 support
posted
+1 support, but I don't think it'll happen
posted
Im with sapphire btw +1 x-x
posted
Hm, I'm with you...looks like it would actually be able to be used by quite a few people at the very least (me as well :D ), and it doesn't sound to demanding an equation...as for concerns about processing power, the equation to calculate actual rank score and accuracy sounds more complicated (to me?) than this so, locally, I doubt it would pose an issue for just filtering...only actual thing you didn't address was spinners, though it'd be easier just to take total rank score over max score so that it would also fluctuate percentages relative to the "maximum" score as well? (though only above)
posted
Not to hijack the request (and it has my full support), this could also be handy for accuracy, sorting from 0 to 100% (sort by rank does not accomplish this btw, an S rank 94% and A rank 99% are quite common).

The beatmap's max score is already computed in the editor without the server intervening, so I suspect that this could be implemented without taxing the server at all.
posted
this would be sooo usefull
posted
i don't think this is that feasible since there is theoretically no maximum score for beatmaps with spinners.

i think a better idea that accomplishes roughly the same thing is to sort by the difference between max combo and the combo you achieved, then accuracy percentage.

these are the main factors that determine your score, so your list would almost always be sorted with the maps you can gain the most score on at the top. the only exceptions are if your missed combo is relatively the same on two maps of different lengths. the longer song would technically gain you more points if your accuracy rate is the same because your combo would be higher.

example:

you have 4 beatmaps with the following maximum combos: 100, 200, 300, 400

you've obtained the following high scores: 98x / 94.99%, 198x / 95.00%, 250x / 90.00%, 325x / 90.00%

you sort by this new option and get the following:

400 - 75x combo missed, 90.00% accuracy
300 - 50x combo missed, 90.00% accuracy
100 - 2x combo missed, 94.99% accuracy
200 - 2x combo missed, 95.00% accuracy (technically has a higher theoretical score gain than the 100x map above, but no system is perfect)
Please sign in to reply.