Signature Performance - An Explanation

[BohemianMatt]

Mine hasn't changed in forever. I'm glad mine is Could You Be Loved by Bob Marley, but I'd like to see what else has been deemed worthy since then.

[Xebarsis]

Ironically my signature performance is Charlene.

http://www.rockband.com/users/Xebarsis/activity

[KMO_2000]

Your specific rank # on a leaderboard doesn't affect signature performance because it's not normalized across songs. A rank of #1,000 on a leaderboard with 1,000 scores means something very different than #1,000 on a leaderboard with 100,000 scores. So it's not useful for comparing performance across different songs.

(Statistician hat on)

That's true, but you shouldn't be comparing rank #, you should be comparing rank percentile (rank/number of scores):

#1,000/1,000 = Top 100% = rubbish
#500/1,000 = Top 50% = average
#1,000/100,000 = Top 1% = pretty good
#1/1,000 = Top 0.1% = very good
#10/100,000 = Top 0.01% = fantastic
#1/100,000 = Top 0.001% = Rock Legend.


The only issue remaining then is easy songs where you can hit the top fairly easily = eg a song where 100 people are "first equal". To deal with that, the calculation should treat all tied ranks as equivalent to the middle rank, not the highest, so treat all those people as being rank "50.5" in this case, not 1.

[Edit 14-Apr-11: maths correction - to get this precisely right, you need to always take a notional midpoint, even for solo scores. The numerator needs to be (your_rank+next_rank)/2-1. The calculations above should be:

#100,000/100,000 = 99999.5/100000 = 0.999995 = abysmal
#1,000/1,000 = 999.5/1000 = 0.9995 = rubbish
#500/1,000 = 499.5/1000 = 0.4995 = average
#1,000/100,000 = 999.5/100000 = 9.995E-3 = pretty good
#1= among 100/100,000 = 50.5/100000 = 5.05E-3 = pretty good
#1/1,000 = 0.5/1000 = 5E-4 = very good
#10/100,000 = 9.5/100000 = 9.5E-5 = fantastic
#1/100,000 = 0.5/100000 = 5E-6 = Rock Legend.
]

If you wish to select from the player's overall best instrument - sounds reasonable - that should be standardised on a percentile too, rather than absolute total score.

I think this calculation will do a better job of what you're trying to attempt, given your stated intent in the original post.

In my case, I'm clearly best at Pro Keys compared to other players, based on "total score ranking divided by number of people ranked". And I'm #1 on a number of Pro Keys leaderboards, which would then mean this algorithm would probably select whichever song I'm untied in the lead of has the most ranked players, giving the lowest (rank/players) number. And the result of that would make more sense to me as a Signature Performance than my fairly unremarkable gold star Guitar run of "Killing Loneliness".

[KMO_2000]

Actually, my signature performance by that algorithm would probably be Pro Keys on Roundabout or one of the other on-disc songs. My #1's are on DLC, and there are going to be more than 10 times as many people playing the on-disc songs, so my #9 for Roundabout is a better percentile than my DLC #1s. Still fine by me.

[hmanji]

I think this calculation will do a better job of what you're trying to attempt, given your stated intent in the original post.

I still don't have the foggiest idea what HMX is trying to achieve or what "signature performance" means.

[Daijobu]

So, funny story. Or not. A friend of my husband's was over and decided that he'd play a song on my tag while I was out of the room. As a result, my signature performance is something that I didn't play, and don't think I can bury any time soon.

Is there any hope, or do I just get to swear under my breath every time I see his score as my signature performance? ...jerk

[TheHundredDollarHeadache]

If you wish to select from the player's overall best instrument - sounds reasonable - that should be standardised on a percentile too, rather than absolute total score.

I agree with this. Unless the player plays practically nothing but Pro Guitar/Bass/Keys, one of the Basic "Instruments" or Pro Drums will be the Scores Highest On instrument.

Using this system, my Scores Highest On instrument will switch to Pro Guitar (17th out of 22,900) from Pro Drums, and my Signature Performance will probably be my 5-star of Rainbow in the Dark.


The "Most Played Genre" also needs standardization. Most players have plain old Rock as their most played genre, because of the sheer number of songs dumped into this ghetto of genericness. It should be done by the following formula:

Adjusted Genre Play Count = Raw Genre Play Count / Number of Songs Released In That Genre.

So, funny story. Or not. A friend of my husband's was over and decided that he'd play a song on my tag while I was out of the room. As a result, my signature performance is something that I didn't play, and don't think I can bury any time soon.

Is there any hope, or do I just get to swear under my breath every time I see his score as my signature performance? ...jerk

Play a different instrument enough to beat your guitar total score. I played enough Pro Drums to get rid of my joint first place on Llama as my Vocals signature performance.

[SheSaidSheSaid]

I just don't understand why it changes so much. Mine used to go back and forth among "PDA," "The Killing Moon," and "Killing Loneliness" all the time. Now it's "Real Wild Child." I like all four songs (naturally), and, also naturally, all four are gold-star FC's, but it changes sometimes to a song I haven't even played in a few weeks (noticed that with "PDA" a little while ago). I'm #2 in the world on "PDA," so I thought that made a perfectly logical signature performance, but if I'm even ranked on the two RB3 on-disc songs it's in the hundreds.

[Bront20]

I think they keep tweaking it a bit. I still wish they did a top 5. They'd swap out a little more often, and they wouldn't have to limit all of them to your top instrument.

[BohemianMatt]

I just don't understand why it changes so much.

I have the opposite problem. Mine's been stuck on Could You Be Loved by Bob Marley for a month or two (since they first tweaked it). I'm waiting to see if it will ever change again.