Billboard Hot 100: 08/18/2012

[mateuszr]

Hi everyone!
I'm new on PulseMusic but I read the forum every day. I think I can help you a little bit. (Sorry I don't speak english so sometimes you can't understand me :| )

I was shocked when I saw 'Lights'on #2 Hot100. I was more shocked when Billboard didn't show on-demand numbers. I think Billboard doesn't want someone to uncover Hot100 formula.
IMO it should look like that:

Points = Sales[/k] x A + Radio Audience[/m] x B + On-Demans Streams[/k] x C

where:
0% < A, B, C < 100% and A + B + C = 100%

I wrote algoritm to find A, B, C from last week chart. Unfortunately, 14 option works in this case:

1. Sales: 34%; Radio Audience: 65%; On-Demand Streams: 1%
2. Sales: 35%; Radio Audience: 63%; On-Demand Streams: 2%
3. Sales: 35%; Radio Audience: 64%; On-Demand Streams: 1%
4. Sales: 36%; Radio Audience: 62%; On-Demand Streams: 2%
5. Sales: 36%; Radio Audience: 63%; On-Demand Streams: 1%
6. Sales: 37%; Radio Audience: 61%; On-Demand Streams: 2%
7. Sales: 37%; Radio Audience: 62%; On-Demand Streams: 1%
8. Sales: 38%; Radio Audience: 59%; On-Demand Streams: 3%
9. Sales: 38%; Radio Audience: 60%; On-Demand Streams: 2%
10. Sales: 38%; Radio Audience: 61%; On-Demand Streams: 1%
11. Sales: 39%; Radio Audience: 58%; On-Demand Streams: 3%
12. Sales: 39%; Radio Audience: 59%; On-Demand Streams: 2%
13. Sales: 39%; Radio Audience: 60%; On-Demand Streams: 1%
14. Sales: 40%; Radio Audience: 57%; On-Demand Streams: 3%

I need more data to find one formula. I hope Billboard will show more information.

[Deleted]

Thanks for the contribution.

[Lozzy]

mateuszr, I remember a month or two ago they gave us a couple of 'change in total points' figures, which I haven't seen since. Did you use those in your calculations? If not, I think they may be worthwhile to try and narrow it down or something.

Also, those options don't incorporate the possibility of A, B and C being whole numbers, do they? If Billboard was to choose a number like 37% or 59% that has no significance, then they may as well have chosen some decimal to make it harder for us.

[mateuszr]

Aug 14, 2012 1:55:11 GMT -5 Lozzy said:

mateuszr, I remember a month or two ago they gave us a couple of 'change in total points' figures, which I haven't seen since. Did you use those in your calculations? If not, I think they may be worthwhile to try and narrow it down or something.

I remember that. I'll try to use that information.

Aug 14, 2012 1:55:11 GMT -5 Lozzy said:

Also, those options don't incorporate the possibility of A, B and C being whole numbers, do they? If Billboard was to choose a number like 37% or 59% that has no significance, then they may as well have chosen some decimal to make it harder for us.

A, B, C can be any real number(A, B, C ∈ ℝ) but ratios must be constant(A:B:C=const.). If they use other numbers we will have other points but the same positions(I hope you understand me ).

[mateuszr]

6/9/12:
"With "Maybe" up by 7% in overall Hot 100 points and "Somebody" down by 4%, Jepsen continues to narrow the gap between the chart's top two titles, likely setting up a closer battle for next week's No. 1 song."

www.billboard.com/news#/news/phillip-phillips-storms-hot-100-s-top-10-1007189352.story

Only one formula works:

[Lozzy]

Hopefully we get some useful figures this week to test that out. :)

[Rodze]

Don't forget we never get info in the old streaming, which is a small component, but might be difference between the formula not working for a song.

[dilek85]

Aug 14, 2012 9:51:52 GMT -5 mateuszr said:

6/9/12:
"With "Maybe" up by 7% in overall Hot 100 points and "Somebody" down by 4%, Jepsen continues to narrow the gap between the chart's top two titles, likely setting up a closer battle for next week's No. 1 song."

www.billboard.com/news#/news/phillip-phillips-storms-hot-100-s-top-10-1007189352.story

Only one formula works:

good job mateusz, nice to see another polish guy here, pozdrawiam.

[icefire9]

Thanks to that formula, I've been doing some calculations on this week's hot 100.

AI:
Using today's AI figures-
Wide Awake: 164.767x.75(mediabase impressions are larger than the ones Billboard uses)x.65= 80.32 points
Lights: 161.738x.75x.65= 78.85 points
Call Me Maybe: 123.362x.75.65= 60.14 points
Whistle: 84.547x.75x.65= 41.22 points

Streaming:
Pretty much guesses based off of previous figures.
Lights: 910x.01= 9.1
Call Me Maybe 890x.01= 8.9
Whistle: 800 (I'm probably being generous)x.01= 8.0
Wide Awake: 550x.01= 5.5

Sales:
Also roughly estimated.
Whistle: 225 (marginal increase from last week, which could easily be wrong)x.34= 76.5
Call Me Maybe: 156 (6% decrease from last week)x.34= 53.04
Wide Awake: 120 (decrease despite discount, I don't know)x.34= 40.8
Lights: 110 (8% decrease from last week due to falling out of the itunes top 10 halfway though it, could be more or less obviously)x.34= 37.4

Totals:
Wide Awake:126.6
Whistle: 125.7
Lights: 125.4
Call Me Maybe: 122.1

I would consider this a statistical tie given I'm mostly guessing with all these figures (except airplay), though it certainly looks like Call Me Maybe is very vulnerable. It also looks like Wide Awakde does indeed have a good chance at #1 this week, despite my previous doubts.

Also, considering how close these songs are, passive streaming could make the difference.

[dlostfan]

Aug 14, 2012 1:35:06 GMT -5 mateuszr said:

Hi everyone!
I'm new on PulseMusic but I read the forum every day. I think I can help you a little bit. (Sorry I don't speak english so sometimes you can't understand me :| )

I was shocked when I saw 'Lights'on #2 Hot100. I was more shocked when Billboard didn't show on-demand numbers. I think Billboard doesn't want someone to uncover Hot100 formula.
IMO it should look like that:

Points = Sales[/k] x A + Radio Audience[/m] x B + On-Demans Streams[/k] x C

where:
0% < A, B, C < 100% and A + B + C = 100%

I wrote algoritm to find A, B, C from last week chart. Unfortunately, 14 option works in this case:

1. Sales: 34%; Radio Audience: 65%; On-Demand Streams: 1%
2. Sales: 35%; Radio Audience: 63%; On-Demand Streams: 2%
3. Sales: 35%; Radio Audience: 64%; On-Demand Streams: 1%
4. Sales: 36%; Radio Audience: 62%; On-Demand Streams: 2%
5. Sales: 36%; Radio Audience: 63%; On-Demand Streams: 1%
6. Sales: 37%; Radio Audience: 61%; On-Demand Streams: 2%
7. Sales: 37%; Radio Audience: 62%; On-Demand Streams: 1%
8. Sales: 38%; Radio Audience: 59%; On-Demand Streams: 3%
9. Sales: 38%; Radio Audience: 60%; On-Demand Streams: 2%
10. Sales: 38%; Radio Audience: 61%; On-Demand Streams: 1%
11. Sales: 39%; Radio Audience: 58%; On-Demand Streams: 3%
12. Sales: 39%; Radio Audience: 59%; On-Demand Streams: 2%
13. Sales: 39%; Radio Audience: 60%; On-Demand Streams: 1%
14. Sales: 40%; Radio Audience: 57%; On-Demand Streams: 3%

I need more data to find one formula. I hope Billboard will show more information.

If you run all these scenarios for all the past weeks since the formula change, you will find that only this works in all the cases.

Okay, I am giving a simple formula, which should work always.

Multiply sales by 6.
Multiply Airplay by 9.
Divide streaming by 2.

For example, if a song as 200k sales, 100M Airplay and 800k streaming, then the points are:

sales = 200 * 6 = 1200
airplay = 100 * 9 = 900
streaming = 800 / 2 = 400

Total = 2500

Or you can use something proportional to this.

[badrobot]

You're getting close to the formula I'm using:

sales * 1
streaming * 0.1
airplay * 1.5

My formula gives a little more weight to streaming than yours does though.

[HolidayGuy]

My my my at all of the math crunching. Is it really that big a deal to just wait and see who places where? That's how it worked in the ol' days, after all. :)

[badrobot]

^I think people are overcomplicating it, which is why I like my simple formula. :)

BB always just did simple math before (this times 5, that times 10, etc.), the level of calculations some people are doing is not really in line with how they have historically approached the Hot 100. And, anything based around % is going to be fundamentally wrong.

[Au$tin]

Aug 15, 2012 9:38:10 GMT -5 HolidayGuy said:

My my my at all of the math crunching. Is it really that big a deal to just wait and see who places where? That's how it worked in the ol' days, after all. :)

You mean in the ol' days where it's very possible that artists cheated their way to #1 just by paying off someone? We had no way of knowing the stats in the ol' days whereas now we do. If we know the formula, we can challenge Billboard if any trickery is afoot. (I believe someone caught their error a few years ago when "Alejandro" and "Your Love Is My Drug" should have been switched.)

[cesarams]

I'm personally waiting for this #1 for Taylor's song. I love her and loving so much now that her album will be called "Red", my favorite color. And she always releases her albums at octobers (my birthday month). I feel like she's always giving me a gift. Go Taylor!

[HolidayGuy]

I hardly think any act "cheated" their way to No. 1. The system that was in place was not as accurate as the one that came, but I wasn't even talking about that era. Billboard did not report airplay and sales numbers as regularly as it did later on when it first started using BDS and SoundScan for the Hot 100. Actual sales figures were few and far between until later in the 90s. And I don't recall seeing airplay figures reported until, like 1996 or 1997, in the Singles Spotlight column.

Hence, those were the "good ol' days" I was speaking of. :)

With other info added to the Hot 100 mix, that's going to make it less easy to figure out rankings- and I like it.

[Verisimilitude]

Ellie's not even top ten on sales.

[HolidayGuy]

Wrong thread, cesa. I did post that info in the 8/25 one, though.

[dlostfan]

Sales+Airplay points.

Wide Awake: 211
Whistle: 205
Lights: 191
CMM: 190

Difference of 1 points can be countered by 18k difference in streaming.

So if Whistle has 110k more streaming than WA, it wins....

I think every song has a chance depending on streaming.

[cesarams]

Aug 15, 2012 10:27:15 GMT -5 HolidayGuy said:

Wrong thread, cesa. I did post that info in the 8/25 one, though.

Sorry. I'm a little lost today. Gonna delete it.

[mateuszr]

Aug 15, 2012 7:05:18 GMT -5 dlostfan said:

If you run all these scenarios for all the past weeks since the formula change, you will find that only this works in all the cases.

Today Billboard gave us some information about last week:

#1 "Call Me Maybe":
Sales: 166K
Radio Audience: 99M
ODS: 824K

#2 "Lights":
Sales: 120K
Radio Audience: 114M
ODS: 882K

#3 "Whistle":
Sales: 222K
Radio Audience: 59M
ODS: 687K

#4 "Wide Awake":
Sales: 130K
Radio Audience: 118M
ODS: 594K

It's very useful. Now we have 4 options:

1. Sales: 36%; Radio Audience: 61%; On-Demand Streams: 3%
2. Sales: 37%; Radio Audience: 60%; On-Demand Streams: 3%
3. Sales: 37%; Radio Audience: 61%; On-Demand Streams: 2%
4. Sales: 38%; Radio Audience: 59%; On-Demand Streams: 3%

----
I added this week data and I changed algorithm(it includes rounding of numbers) and percentage step.

I think it is the best method:
1. Sales: 38.5%; Radio Audience: 58%; On-Demand Streams: 3.5%