View previous topic :: View next topic |
Author |
Message |
JCirri
Joined: 04 Feb 2006 Posts: 4576
|
Posted: Thu Feb 16, 2006 6:30 am Post subject: 5-star cutoff confirmed |
|
|
In one of my previous posts I had indicated that I thought 76,500 was the cutoff for 5-stars on I Love Rock & Roll - Expert and some of the data I posted was from my memory, and not entirely accurate.
Now after playing this song about 50-100 more times and documenting the results, I finally found the exact cutoff for 5-stars. And as you can see my percent is higher on the 4-star so ratings are not based on percent. In all of my trials, never did I have a 4-star higher than a 5, so this proves that raw score determines ratings.
I will be making a new forum and posting challenges for you to help me find the rest of the cutoffs. I'll post a range for each song and if you get a score in the requested range and post the screenshot, we'll update the range accordingly until we have them confirmed exacly (like the one above). _________________
|
|
Back to top |
|
|
Jed
Joined: 14 Feb 2006 Posts: 656 Location: Texas
|
Posted: Thu Feb 16, 2006 6:33 am Post subject: |
|
|
I'll give it a shot, on some of the easier songs anyway. >_> That's kinda weird though that the exact cutoff doesn't end in zero. I figured to make things easier they would've done it that way. Nice find. |
|
Back to top |
|
|
JCirri
Joined: 04 Feb 2006 Posts: 4576
|
Posted: Thu Feb 16, 2006 6:37 am Post subject: |
|
|
Yeah, the fact that it doesn't end in zero makes me think it's based on a formula. Possibly based on some percentage of the total number of notes / chords and total length of freezes are my guesses. As we get more confirmed cutoffs we can analyze the data more precisely. _________________
|
|
Back to top |
|
|
Jed
Joined: 14 Feb 2006 Posts: 656 Location: Texas
|
Posted: Thu Feb 16, 2006 6:42 am Post subject: |
|
|
Awesome. I'm a statwhore/datawhore. I'm always interested in finding out how things work and patterns and stuff. Do you plan on finding the cutoffs for every difficulty? And also, are you gonna get the cutoff from 3 to 4 stars as well? |
|
Back to top |
|
|
Jaksiel
Joined: 05 Feb 2006 Posts: 778 Location: Troy, NY or Hooksett, NH
|
Posted: Thu Feb 16, 2006 6:44 am Post subject: |
|
|
Ha! Interesting. _________________
|
|
Back to top |
|
|
JCirri
Joined: 04 Feb 2006 Posts: 4576
|
Posted: Thu Feb 16, 2006 6:51 am Post subject: |
|
|
Jed wrote: | Awesome. I'm a statwhore/datawhore. |
Me too.
I'd like to confirm as much as possible. I've already made a dynamic page for myself to be able to see all the current possible ranges and their widths for all 4-star and 5-star cutoffs based on the scores in the system.
I haven't really formatted the page yet, and several of the ranges have negative widths due to false scores that get entered. I've been doing my best to remove them.
When I get it sorted out I will post all the ranges for 4 and 5 star cutoffs and we can start to narrow them down. _________________
|
|
Back to top |
|
|
cinder
Joined: 05 Feb 2006 Posts: 26
|
Posted: Thu Feb 16, 2006 5:48 pm Post subject: |
|
|
okay, I'll say it here as well. Score and percent have nothing to do with stars, it's ALL ABOUT STREAKS.
I'm not joking here, everyone's trying to find cutoffs that don't exist. The game analyzes how much time you stay at x4 or x8. That's it.
Thanks. |
|
Back to top |
|
|
RiskyChris
Joined: 05 Feb 2006 Posts: 21
|
Posted: Thu Feb 16, 2006 7:22 pm Post subject: |
|
|
cinder wrote: | okay, I'll say it here as well. Score and percent have nothing to do with stars, it's ALL ABOUT STREAKS.
I'm not joking here, everyone's trying to find cutoffs that don't exist. The game analyzes how much time you stay at x4 or x8. That's it.
Thanks. |
Your data is interesting, and I would like to subscribe to your newsletter. _________________
|
|
Back to top |
|
|
JCirri
Joined: 04 Feb 2006 Posts: 4576
|
Posted: Thu Feb 16, 2006 9:30 pm Post subject: |
|
|
OK, here's my proof that rating is not based on streaks or how much time is spent at 4x/8x.
In the first score, the 252 note streak accounts for every note that I hit in the song (just one big streak), such that 222 notes were played with 4x.
The second score I consistently broke combo after 40 notes every time (such that only 10 notes were played at 4x for each streak).
The difference between these 2 scores is 504 points. It takes at least 3,000 points (10*50 + 10*100 + 10*150) just to bring the multiplier up to 4x one time, and in the 2nd score I had to do this many times (7 to be exact) which means at least 21,000 points were gained with a multiplier less than 4x, vs. at most 6,000 (assuming all chords) in the first score gained under 4x. So this score differernce is easily covered by time spent under 4x (since 15,000 >> 504) which implies that more time was spend at 4x on the 4-star rating than the 5-star rating.
Put another way, in the first score 222 notes where played at 4x. In the second, no more than 70 were played at 4x.
And even assuming that all 70 of these were chords and all 222 were single notes (which is clearly not the case), 140 is still considerablly less than 222. _________________
|
|
Back to top |
|
|
OtherBill
Joined: 12 Feb 2006 Posts: 65
|
Posted: Thu Feb 16, 2006 9:52 pm Post subject: |
|
|
-applause-
Y'know, I was going to offer to run similar experiments this weekend, but...well, now I don't have to.
So far, nobody has been able to provide smoking-gun screenshots of two summary screens--same song, same difficulty--where the higher score gets the lower rating. Until they do, I'm personally sticking with the "score threshold" theory.
(Don't get me wrong, I'll happily entertain other theories, but the "score threshold" theory continues to be the most compelling.) _________________
|
|
Back to top |
|
|
Matt
Joined: 04 Feb 2006 Posts: 3780 Location: Bethel, Vermont
|
Posted: Thu Feb 16, 2006 11:42 pm Post subject: |
|
|
cinder wrote: | okay, I'll say it here as well. Score and percent have nothing to do with stars, it's ALL ABOUT STREAKS.
I'm not joking here, everyone's trying to find cutoffs that don't exist. The game analyzes how much time you stay at x4 or x8. That's it.
Thanks. |
Curiously, where did you get this information? |
|
Back to top |
|
|
cinder
Joined: 05 Feb 2006 Posts: 26
|
Posted: Fri Feb 17, 2006 3:17 am Post subject: |
|
|
Matt wrote: | cinder wrote: | okay, I'll say it here as well. Score and percent have nothing to do with stars, it's ALL ABOUT STREAKS.
I'm not joking here, everyone's trying to find cutoffs that don't exist. The game analyzes how much time you stay at x4 or x8. That's it.
Thanks. |
Curiously, where did you get this information? |
Multiple Harmonix employees. I want to get one of them to write something explaining it, maybe I'll try to push on that. |
|
Back to top |
|
|
cinder
Joined: 05 Feb 2006 Posts: 26
|
Posted: Fri Feb 17, 2006 3:25 am Post subject: |
|
|
RiskyChris wrote: | cinder wrote: | okay, I'll say it here as well. Score and percent have nothing to do with stars, it's ALL ABOUT STREAKS.
I'm not joking here, everyone's trying to find cutoffs that don't exist. The game analyzes how much time you stay at x4 or x8. That's it.
Thanks. |
Your data is interesting, and I would like to subscribe to your newsletter. |
You can feel free to look for a black cat in a dark room that isn't there, that's fine.
http://home.comcast.net/~kagovoni/log011606.html
About JCirri's second set of pictures, I'm guessing he broke his score on purpose, and went right back to x4. And like I have said before, it's not the highest streak, it analyzes how long you many consecutive notes you had throughout the song. |
|
Back to top |
|
|
Jed
Joined: 14 Feb 2006 Posts: 656 Location: Texas
|
Posted: Fri Feb 17, 2006 3:51 am Post subject: |
|
|
I can easily fake an AIM conversation too. Doesn't prove anything. The screenshots do. And until I see CONCRETE EVIDENCE proving otherwise, I'm with JCirri and the score theory. |
|
Back to top |
|
|
OtherBill
Joined: 12 Feb 2006 Posts: 65
|
Posted: Fri Feb 17, 2006 4:54 am Post subject: |
|
|
Taken from cinder's log:
Quote: | (13:29:55) HMX Employee: it will look at your score and say "hmmm you must have had a x4 multiplier for this much of the song, so you get this star rating" |
Ummm...I don't know about you, but if a Harmonix employee says "it will look at your score", then that sort of supports the "score threshold" theory... _________________
|
|
Back to top |
|
|
|