In game two of this year's World Series, the old-school Fox announcers were indignant that Dodger's pitcher Rich Hill was pulled from the game based on 'the analytics guys'. Meaning: the number crunchers, who know nothing about baseball, are taking over the game and they are wrong.
While the number crunchers were perhaps wrong in this particular case (the Dodgers did go on to lose the game), the announcers missed the bigger point about making data-driven decisions. Those of us in this business know that the analytics will be wrong - sometimes frequently - but it will be less wrong than 'seat of the pants' decisioning.
An example I saw at the beginning of my career was when credit scoring models replaced criteria-based lending. Old-time lenders were very resistant to an algorithm making decisions. And those of us involved in creating said algorithms has to defend this new approach: it made the wrong decision with some regularity. What took some time was showing the old guard that this new technology, while imperfect, represented progress.
It is encouraging to see the sports franchises who have embraced analytics doing well (Dodgers, Chicago Cubs). It was also interesting that the true early adopter of sports analytics (the Oakland A's under Billy Beane) has been among the leaders in 'cost per win' - perhaps the ultimate long-term measure of success in sports.
Friday, October 27, 2017
Thursday, October 26, 2017
Future of Analytics?
I have been immersed in the analytics business since 1991. From this first internship when I was in grad school, I have occasionally spent some time thinking about where the business of analytics is headed.
Looking back 25 years, I would have sworn that the following would have happened by now:
Looking back 25 years, I would have sworn that the following would have happened by now:
- Artificial intelligence (AI) would be in place for most analytic applications
- Analytics would be so prevalent that there would be no competitive advantage for using it (kinda like using computers in general)
- Data would be so available that it would drive most business decisions
- The Atlanta Braves would have won five World Series
Well, the future of 2017 is here - and none of the stuff above is really in place. In fact, the methodologies in place have scarcely evolved since 1991 (although computing power has increased dramatically).
So, what do things look like in another 25 years in the year 2042 (gulp):
- AI is more prevalent, although humans play a large role in defining problems to be solved
- Analytics in general is so pervasive that there is an extreme disadvantage for not using it
- Your basic iPhone 27 (titanium edition) advises you on all basic and major decisions, based on analytics applied to cloud-based data
It will be fun to see what happens!
Subscribe to:
Posts (Atom)