Get This Free Cheat Sheet Guaranteed To Help Your Next Mix

Thursday, December 22, 2011

Yes, Math Can Predict A Hit

Everyone has heard the phrase "If it were so easy everyone would be doing it." Having a hit record might be within "everyone's" grasp, thanks to a new site set up by an artificial intelligence group at the University of Bristol, who believe they have "The Hit Equation."Yes, they have delineated the hit record down to the equation below:

OK, that didn't mean much to me either, but the site does have a few interesting tidbits. After analyzing hits on the UK charts over the last 50 years, they have determined:

1) In early decades, hits tended to be harmonically simpler than non-hits. However, nowadays the opposite is true.

2) Around 1980 seems to be a creative period in pop music.

3) Dance songs were more popular in the 1980s.

4) Music of all qualities are getting louder.

That last one seems to be stating the obvious. I'm not sure that I agree with first three.

The site, called, has a section on Expected Hits (Gnarls Barkley's "Crazy"), Unexpected Hits (Alicia Key's "Empire State Of Mind"), Hidden Gems (Sheryl Crow's "First Cut Is The Deepest"), and they even go as far as to make predictions on current songs.

If you ask me, this is fools gold. Even if it were possible to predict a hit, it still is working with last week's (month's, year's, decade's) data. By the time you determine what works, the public (fickle as we are) has moved on. The only thing different between using math to predict a hit and the record labels following jumping on a trend is the time lag. You can only follow a trend, you can't make one this way.

Check out and let me know what you think. Have a great holiday.!

You should follow me on Twitter for daily news and updates on production and the music business.

Don't forget to check out my Music 3.0 blog for tips and tricks on navigating social media and the new music business.

1 comment:

Faza (TCM) said...

I'm with you on this one - it's not like this is the first time this has been tried.

Moreover, I cannot see this as being particularly helpful for two reasons:

1. Analysing an already written (and fully produced) song in order to predict its hit potential still leaves you with the problem having done the work only to find out the equation thinks it sucks. I cannot see anything there that would help you actually use this equation in the creative process.

2. The performance index is around 0.6 at best. Now, if we were to view hit/flop as a binary choice (a simplified view, I admit), then statistically our chance of predicting the outcome by random guessing would be around 50%. The improvement from applying the formula isn't huge and it would be interesting to compare it with human performance based on informed guesswork by competent people (whoever those may be). Of course, in real life we would be concerned with a specturm of success measured by number of sales, or even profitability, which could possibly explode the entire idea of mathematical prediction in a cloud of complexity (there's insufficient available historical data, for a start).

It's nice headline-grabbing stuff, but I cannot see any sensible business application for it.


Related Posts Plugin for WordPress, Blogger...