How is skill calculated???

1235

Comments

  • Might be alphabetical? I usually end up last (or around last) on any shared scores on the song leaderboards, so I figure it might be the same for the skills leaderboard...
  • Probably the same as leaderboards. Every profile has a random number, and those are used for ties. You'll always beat the same people in a tie, you'll always lose to the same people in a tie.
  • Thanks for the precision. I do check my leaderboard ranking on the RB4 record store, and just found out it's different in game. Double wrong for me, lol.
  • HMXOwl said:

    Probably the same as leaderboards. Every profile has a random number, and those are used for ties. You'll always beat the same people in a tie, you'll always lose to the same people in a tie.

    Interesting. Thank for replying Owl!
    I wish same number skill ranking were more skill or accomplishment based.
  • Use total score for that instrument to break skill ties.
  • Hmm... if you'll never be able to beat those due to a random number and will always win against others due to a constant RNG, then should there be tiebreakers in the first place? Can't there be several #1's and such? It's frustrating to have an optimal skill or score and always end up in something other than 1st when someone several spots above you has the same number.
  • gaiacatgaiacat Unsigned
    edited February 2017

    Hmm... if you'll never be able to beat those due to a random number and will always win against others due to a constant RNG, then should there be tiebreakers in the first place? Can't there be several #1's and such? It's frustrating to have an optimal skill or score and always end up in something other than 1st when someone several spots above you has the same number.

    @HMXOwl please make the suggestion to use something like on disc total instrument score or something fair for the tie breaker. In the rockband competitive leaderboards, (skill ranking) it is disheartening that the #1 vocalist may really be number 30 and I have no way to improve my ranking. Nothing to work towards.
    In fact, as new accounts are created, I will probably go lower. Thank you!!!!
  • thatmarkguythatmarkguy Road Warrior
    Yup, I'm all for this. There are so many sensible ways to break ties - total stars earned, total score earned... that sticking with this 'invisible account-associated tiebreak value' is an awful default.

    If the goal is to make sure that non-DLC buyers aren't punished, Disc Songs Score would be a good arbiter that allows non-DLC-owners equal shot as DLC buyers.
  • If the goal is to make sure that non-DLC buyers aren't punished, Disc Songs Score would be a good arbiter that allows non-DLC-owners equal shot as DLC buyers.

    I mean, I've got recorded scores on two thirds of the disc because of leaderboard wipes and stuff that I haven't revisited. That's fewer scores than the launch window. Do I magically reshuffle in the tie rankings? That's kinda weird, isn't it?
  • HMXOwl said:

    If the goal is to make sure that non-DLC buyers aren't punished, Disc Songs Score would be a good arbiter that allows non-DLC-owners equal shot as DLC buyers.

    I mean, I've got recorded scores on two thirds of the disc because of leaderboard wipes and stuff that I haven't revisited. That's fewer scores than the launch window. Do I magically reshuffle in the tie rankings? That's kinda weird, isn't it?
    It's not weird at all. In fact, it would be useful as a large jump down the leaderboard would clue me in that I need to re-do a song that has been wiped.
  • RJUCPURJUCPU Unsigned
    edited February 2017
    Even more obvious, why can't there be ties? If two people (or more) can achieve an optimal score, why must only one be "first"? This has been a frustration on the leaderboards for a long time. If there are 10 people with the same score they should all be 1st and the next score in line should be 11th. Other sites do it this way and it's obviously the most fair and most convenient. It helps immensely when there are 50-100 people tied with the same score (yes this happens) and you can't tell if you where you really are unless you scroll back to see.

    (Edited as I was merging my battles, apologies for the mixed message)
  • Pretty sure I've figured out the basics of Skill scoring. It all comes down to your overall star rating on songs. You want to gold star every single song you've played. That's it. It's not about overall score or number of FCs or stars. It's an average calculation based on the number of stars you get across all of your songs.

    I don't know exact weights for each difficulty tier or how to figure out exactly how many skill points you'll get for each star rating, but that wasn't really of concern since I was going for 1000 Skill. I also only play on Expert, but I would imagine that difficulty rating ties into it somehow. Can't give someone 1000 skill rating for Gold starring 10 songs on Easy.

    Short answer... To get the perfect rating of 1000, you have to Gold Star every song you have played thus far. The easiest way to do this and to generally boost your Skill score is to sort your song library by Stars and find all the songs you've played that you didn't Gold Star... and practice them until you can Gold Star them.

    Source: My skill rating is 1000.
  • thatmarkguythatmarkguy Road Warrior
    edited February 2017
    Way ahead of ya.

    https://www.reddit.com/r/Rockband/comments/5a3vqx/instrument_skill_formula/



    You can get 1000 skill on as few as seven songs played (if that profile has played no other songs). One GS at each of the 7 tiers.
  • thatmarkguythatmarkguy Road Warrior
    RJUCPU said:

    As someone who was recently affected by this, I don't agree with using any on-disc total score, or similar method, for a tiebreaker. That's just not fair to a lot of people who may have entered the game late or choose to do their songs helter skelter.

    I don't follow. How is using the on-disc songs as tiebreaker unfair to people who entered the game late? Everyone has access to the on-disc songs. Using total score is unfair to people who entered late because early adopters have access to some songs that late adopters do not (in addition to favoring those with deeper pockets). Using disc score is using a metric everyone has equal access to improve.


    I do find it hilarious that you complain about unfairness to late enterers... then proceed to propose first-to-achieve-tied-score as an arbiter.
  • gaiacatgaiacat Unsigned
    edited February 2017
    @rjucpu - on disc would be the most fair since everyone has and CAN play those songs.

    @JoeSlapBass - I think you missed the point of the latest discussion - I am also 1000 skill. I am "ranked" #18 in 1000 skill and this friend is "ranked" #3 . We are asking that people who care have a way to improve their rank within a skill level. Some people would also like it applied to leaderboard song scores as well.

  • thatmarkguythatmarkguy Road Warrior
    edited February 2017
    I can't think of a good way to order equal leaderboard song scores. I mean, if they're equal down to the point, the performances were equivalent. It's unreasonable to prefer most-recent or least-recent because both are equally arbitrarily 'punishing' people who need not be downgraded for recentness of play. If the boards could do a "Show everyone with the equal score as being equal rank, and show the currently-viewing user at the top among that rank his/her own view", that would suffice.

    There are lots of fair ways to break the tie on skill, because someone at 1000 skill on 1000 songs has done more than someone at 1000 skill on 7 songs. But if 6 players are tied on a song with the same score? No one should be ahead of any other one (and it harms nobody to have each player shown themselves at front of the pack.)
  • gaiacatgaiacat Unsigned
    edited February 2017

    I can't think of a good way to order equal leaderboard song scores. I mean, if they're equal down to the point, the performances were equivalent. It's unreasonable to prefer most-recent or least-recent because both are equally arbitrarily 'punishing' people who need not be downgraded for recentness of play. If the boards could do a "Show everyone with the equal score as being equal rank, and show the currently-viewing user at the top among that rank his/her own view", that would suffice.

    There are lots of fair ways to break the tie on skill, because someone at 1000 skill on 1000 songs has done more than someone at 1000 skill on 7 songs. But if 6 players are tied on a sing with the same score? No one should be ahead of any other one.

    A simple answer would be as you said and what has been said earlier in this thread - Every 1000 skill is "ranked" #1 - and the 999 skill people would all be ranked #54 (the next number after all 1000 players) and all 998 skill be ranked #345 (the next number after all 1000 and 999 people) , etc, etc, etc,

    Same thing with song scores. All highest score is #1 - skip labeling #2 through however many people till it get to next highest score. they are ranked #34 (or whatever next number after all highest score.)
  • Ahhh... I see. So we've gotten down to the point where we want to decide tie-breakers.

    I agree with the suggestion for equivalent, tied rank on the song scores. If the scores are identical, then you've tied... plain and simple.

    But skill has a lot more involved in the equation and there are a lot of ways to do this. The most interesting suggestion I've heard on this lately is to take a player's rank per song, weigh this across each difficulty tier, and use this average rank as a tie breaker. The problem then becomes, which songs do you include to give an apples to apples comparison on rank? As it stands now, it's a fair comparison between skill scores since it's an average across the song's each person has played, not a pre-defined set.
  • As it stands now, it's a fair comparison between skill scores since it's an average across the song's each person has played, not a pre-defined set.

    Actually, as it stands now is random rank placement.

  • JoeSlapBassJoeSlapBass Unsigned
    edited February 2017
    Yes, yes... in terms of how people are ranked when tied at the same Skill Rank, this is true.

    I was referring more to the problem of which songs to include in the set. I suppose it still wouldn't be a problem to continue to go with using whatever songs the player has completed and registered a score for. This is the way it works now, and the most "fair" way of doing it. The one outlying issue with this method is the guy with 7 songs played and 1000 Skill rank. That seems wrong...
  • edited March 2017
    Ranking via skill seems to be ranking furiously for no reason. :disappointed: but that's just my personally hurt self.

    What is very tragic is that I'm being punished for singing (before skill rating even existed) 100% hard full combos and getting less skill rating than inferior singers got at 4* expert! So sad!!! :disappointed:

    Guess I gotta start my RB vox life over again with my alter ego, who has a skill rating of 890 after cruising through about 100 of the songs I do the best at (and for the most part, completely incompetent compared to my real self).

    So sad, so sad. :disappointed:

    *cries in the corner*
    *okay not really because there are worse things to cry about*
    *Better: Goes and sings 'Happy Song' in the corner*

    ETA: With the following substituted lyrics:

    E - X - P - E - R - T - Expert, sing in it

    [skip some more of the depressing not parodyable lyrics here]

    But if you sing on hard
    And full combo it all well
    just to prove your skill
    You'll just get screwed 'cause

    Everybody else just went and sang in expert
    even though they couldn't get past 4 stars
    and their skill ratings are nine hundred plus
    but you'd be lucky to just get seven hundred
    and you're grinding on in expert
    but can't beat your own high scores....


    :blush:
  • gaiacatgaiacat Unsigned
    edited March 2017





    E - X - P - E - R - T - Expert, sing in it

    [skip some more of the depressing not parodyable lyrics here]

    But if you sing on hard
    And full combo it all well
    just to prove your skill
    You'll just get screwed 'cause

    Everybody else just went and sang in expert
    even though they couldn't get past 4 stars
    and their skill ratings are nine hundred plus
    but you'd be lucky to just get seven hundred
    and you're grinding on in expert
    but can't beat your own high scores....


    :blush:

    Dude, sing on expert.
    If you cant get better than those who get four stars on expert then they ARE better than you. Expert vox requires more control and precision.

    And I feel you on hard scores. I wish individual scores could be wiped manually. I have the same problem on guitar/bass where I played well on hard instead of expert on the first play through.
  • edited March 2017
    gaiacat said:


    Dude, sing on expert.
    If you cant get better than those who get four stars on expert then they ARE better than you. Expert vox requires more control and precision.
    And I feel you on hard scores. I wish individual scores could be wiped manually. I have the same problem on guitar/bass where I played well on hard instead of expert on the first play through.


    Eh? Fact check... I can get a lot better than those who get four stars on expert.

    Example:
    • My high score on "Throne" (by that same band, coincidentally) is 169k - 98% hard
    • My highest expert score on that song is 167k - And that's 5* Expert.
    After I found out how skill rating was calculated, I went through about 50 more songs, replay on expert... got 5*X on most of them, beat my hard score on like ... a whopping 5 of them.

    So that'd be strange logic, that I can't do as well as someone who just got 4*X on those songs, given how I just got 5*X...

    There are even songs where the gold star threshold is slightly lower than the 5* hard well-squeezed full combo (one of the charts for "Metropolis" I believe).

    So that load of scores where I scored in the top 1% of all players... was on the wrong level.

    But actually... I totally agree with you on the "wipe my hard scores" idea. I can go back and play almost all of them at 5* Expert on the first play.

    (New Tip idea: Doing a full combo on a hard track is going to cause your skill rating to take it on the 'Otherside'. It's actually easier to just do a 5* expert and have a rating of 860 instead of 730.)

    (Newer Tip idea: Crappy skill rating make you sad? At least you can rejoice in your global top 300 disc song score and top 100 stars earned score.)


  • gaiacatgaiacat Unsigned
    edited March 2017
    AND I improved almost every category skill column and dropped from #17 to #20 in rank.
  • gaiacatgaiacat Unsigned

  • gaiacatgaiacat Unsigned
    @AndyTheIndigoer - I understand what u are saying and at your skill level you do better on hard than on expert. You have the same situation I have on guitar. I'm good enough to get 100% on hard, but not good enough to keep the streak going on expert to beat my hard score.

    If skill was based on total score on a song then we would both do better on these instruments. Harmonix decided to base skill level on , well, skill. How much of a song can you play on the hardest level without dropping a note or phrase?

    Hard vocals score from start oof song:
    800-2400-4800-8000- then 3200 every phrase while keeping streak

    Expert - 1000-3000-6000-10000- then 4000 points every phrase while keeping streak.

    The most points come from keeping the error free streak, overdrive placement, and playing on the highest difficulty possible. I agree with How Harmonix decides skill level. I just don't agree on the random placement of rank within the skill level.
  • edited March 2017
    @gaiacat -- From your message I understand better now, that you're upset that the top ranks are random. Thanks. I think it might be the raw number of gold stars or something -- maybe you're being punished for having a smaller library or something? It doesn't appear to be alphabetical... but I agree, that's wrong.

    I'm assuming that you don't have any 5*H songs in there by accident. ;-)

    And I understand how difficult expert is compared to hard. In fact I think it's not very nice to only make it 1,000 points per 4x phrase, because of the effort required.

    However, we are still going to end up disagreeing on how HMX decides skill level because of that one detail, which is a lot more glaring if you've accidentally played anything on hard.

    The only reason I can see why they're doing this is to save on data transactions -- This way if a high score is recorded, the score and the stars can be sent in one push and replaced all at once, even if half of the score is replaced by something lower (5*X by 5*H).
  • gaiacatgaiacat Unsigned
    edited March 2017
    @AndyTheIndigoer

    @HMXOwl already confirmed that ranks within a specific rank level is not calculated on skill. Its more of a if u get the same as someone else it was already predetermined who would take better ranking and that person will always get better ranking than you as you will always get better ranking than some other random person regardless of stats other than skill level

  • gaiacat said:

    @AndyTheIndigoer

    @HMXOwl already confirmed that ranks within a specific rank level is not calculated on skill. Its more of a if u get the same as someone else it was already predetermined who would take better ranking and that person will always get better ranking than you as you will always get better ranking than some other random person regardless of stats other than skill level

    Actually that does stink :disappointed:

    They should have considered one of the above factors mentioned by everybody else.
  • Everyone had an invisible static number assigned to their profile in RB3 for tiebreakers, pretty sure RB4 works the same but haven't asked.
1235
Sign In or Register to comment.