[AAA Equivalency] I think your magic formula is broken

**Sanjixcon** · 10-26-2016, 11:51 PM

Re: [AAA Equivalency] I think your magic formula is broken

poor toba

**Soundwave-** · 10-27-2016, 12:55 AM

Don't say "I'm doing X with a song" and not tell me what song it is if you're at all ashamed of your performance.

I will find out.

Every.

Time.

P.S. It was Policy In The Sky!

P.P.S. Not to imply there's something to be ashamed about. We all start somewhere.

P.P.P.S. Ugh. It's called dramatic effect.

**inDheart** · 10-27-2016, 09:46 AM

Re: [AAA Equivalency] I think your magic formula is broken

i feel like it used to get clipped properly, so this makes me wonder if the recent changes made -9.97 the new clipping threshold :S

**YoshL** · 10-27-2016, 09:52 AM

Re: [AAA Equivalency] I think your magic formula is broken

tbh if you're not able to AAA a 1, then your AAA equivalency should be far below a 1.

**devonin** · 10-27-2016, 10:47 AM

Re: [AAA Equivalency] I think your magic formula is broken

1) The AAA equivalency formula is actually really nice, allowing up to thousands of goods before completely dropping to 0.

I reeeealy don't think that's true (not should it be)

**Trumpet63** · 10-27-2016, 12:27 PM

Re: [AAA Equivalency] I think your magic formula is broken

Originally posted by Soundwave-

Now you can calculate these by hand and find that they should be positive, but furthermore:

1) The AAA equivalency formula is actually really nice, allowing up to thousands of goods before completely dropping to 0.

2) Even if you do manage to do this horrendously bad, the radical isn't real for values less than 0, and thus should probably be clamped to 0 AAA equivalency.

Thus, I conclude, it is borked.

You are correct that in general, the formula becomes very flat at a large number of goods, especially at high difficulties. If you calculate the AAA equivalents by "hand" based on what is posted here, you will find that they should be positive; however, that formula is no longer up-to-date as of the release of the new difficulties. I still need to update that, sorry. In all actuality, I believe those negative numbers are right.

But I can say that regardless of whether the formula calculates a negative AAA equivalent, it should never be shown as negative, hence what you posted is a bug. You can expect I'll be deep-frying Prawnskunk about it.

**rushyrulz** · 10-27-2016, 12:31 PM

Re: [AAA Equivalency] I think your magic formula is broken

Originally posted by Trumpet63

You can expect I'll be deep-frying Prawnskunk about it.

**Soundwave-** · 10-27-2016, 12:37 PM

Re: [AAA Equivalency] I think your magic formula is broken

Originally posted by devonin

I reeeealy don't think that's true (not should it be)

Above: The overly complicated delta polynomial, normalized by lambda (the weight of 1 good).

It is. The formula is least lenient at difficulty 5, allowing up to 675 goods, and most lenient at difficulty 110 (well, 120, but there's no song over 110 currently), allowing 24882 goods.

Above: Percentage of "good leniency" at given difficulty vs AAA equivalency / difficulty. Accurate to 1% for difficulty 1 to 0.0091% for difficulty 110

Most of the heavy lifting of penalizing the AAA equivalency comes from the exponents. At just 4% of the "good leniency", you only get 10% of the AAA equivalency, which is pretty rough.

And yes, I did clip that graph to only 10% of the good leniency because your AAA equivalency pretty much disappears after that.

Originally posted by inDheart

i feel like it used to get clipped properly, so this makes me wonder if the recent changes made -9.97 the new clipping threshold :S

I would assume it's related to the difficulty changes as well, but as you can see, there's no oddities expected even at higher difficulties than previously. It must be an edge case or something.

EDIT:

Originally posted by Trumpet63

However, that formula is no longer up-to-date as of the release of the new difficulties. I still need to update that, sorry. In all actuality, I believe those negative numbers are right.

Wow I worked very hard on this post and now you're telling me the formula is wrong I mean I could have guessed that

**Trumpet63** · 10-27-2016, 12:50 PM

Re: [AAA Equivalency] I think your magic formula is broken

That's a very nice post Soundwave. You made me happy regardless.

**Soundwave-** · 10-27-2016, 01:00 PM

Re: [AAA Equivalency] I think your magic formula is broken

Originally posted by Trumpet63

That's a very nice post Soundwave. You made me happy regardless.

I appreciate you bolstering my artificial metric of self worth.

**Trumpet63** · 10-27-2016, 02:17 PM

Re: [AAA Equivalency] I think your magic formula is broken

Prawns were fried, but it was Velocity who did the fix. All negatives should just be zeroes now.

**XelNya** · 10-27-2016, 03:11 PM

Re: [AAA Equivalency] I think your magic formula is broken

Originally posted by Soundwave-

I appreciate you bolstering my artificial metric of self worth.

Dammmmmmn

**Soundwave-** · 10-27-2016, 06:34 PM

Re: [AAA Equivalency] I think your magic formula is broken

Originally posted by Trumpet63

Prawns were fried, but it was Velocity who did the fix. All negatives should just be zeroes now.

Huzzah! You could say that was a speedy fix.

No?

**Velocity** · 10-28-2016, 12:19 AM

Re: [AAA Equivalency] I think your magic formula is broken

Originally posted by Soundwave-

Huzzah! You could say that was a speedy fix.

No?

ugh.

[AAA Equivalency] I think your magic formula is broken

[AAA Equivalency] I think your magic formula is broken

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment