Predictive vs Prescriptive Analytics

A lot of people read the article I had in RIS news last year. If not, HERE:

(Link updated, 2018!)

I wanted to talk a little further about the idea of prescriptive analytics. As a few people know, I like to bet sports. REALLY like to. (Yeah, I live in Nevada, so its legal. Don’t judge. )

If you’re really good at what you do, then you are willing to put your money where your math is. So this football season, I built a NFL betting modeling system. Without telling too much details, as that may be a whole different story line later, it was a fun experiment.

First, gather the data: Build the history. Data without enough enough data points is too inconclusive. In week 6, I added in the predictive engine, and started picking bets. All told, against the spread for the year, I was +7%. Many lessons learned, many theories tested, but suffice to say I was pretty happy by the end of the season as I picked more winners than losers. Many things came clearly into the light – visualization. How often to use teasers, how often to pick Over/Under. But my best lesson was that the people (Customers? Guests? Shoppers? Apply your own business here) are willing to overpay for a favorite. The Panthers were an exception, and very predictable. But most other ‘favorite’ picks were overpriced. Hmmm. As I believe I understand the concept of para-mutual, and how sports betting is a balance of money, so I cant blame the house for rigging the system: Blame the consumers.

Point? Predictive analytics was effective. However, where I didn’t end up with a new car, new house, lovely beach condo on Maui was in the prescriptive piece. I need to refine this to tell me what actions to take based on the prediction. Yeah, the Pats covered a lot – but whats the bet amount? Planning the actions is what separates a moral win from a real win.
Analytics drives business.

PS- Dead wrong on Superbowl. What makes analyics awesome? Its not perfect. If it was, this would be boring. The best model wins the most, but not all. Betamax was great technology. Creative Lab’s MP3 player was awesome in 2000. And we all know how that ended up.

-THAT Planning Guy