My notes from an interview with Nate Silver, founder of FiveThirtyEight.com, at Adobe Summit:
Question: When did you say you could have a profession in number and stats?
I don’t know when really but I took a boring consulting job right out of college. In my spare time I started working on a program to forecast how baseball players would do. I think KPMG technically owns everything I do still. (audience laugh after pregnant pause…..)
Question: (roll Charles Barclay video on his anti-analytics riff) What do say to that riff?
Charles is a really smart guy actually but it’s pretty much total BS what he said. Houston Rockets are run by an MIT grad. The spurs are analytics driven. The best teams in the NBA use analytics. We can now measure how good or lazy people are on defense. (How far someone runs during a game is the stat)
Question: Let’s talk baseball. Is there a difference between how teams use data or is it the new norm?
There is an aspect of adverse selection for the teams that don’t use analytics. It’s gone from an open source kind of thing to being behind closed doors. (kind of a shame really) All but two or three teams list an analyst in the front office.
Question: Is the NBA harder or is the data set different than baseball?
Baseball is a weird sport because people take their turn. When modeling interactions between players like in the NBA things become more challenging. There is pretty good data though. The NBA is at the peak of growth for this now. In the sports analytics world basketball is the most exciting?
Question: What about soccer?
There is huge money in soccer. Munich is doing analytics. Others are throwing millions at players. Soccer is seeing a lot of growth. They used to only keep track of goals and red cards. But now they are looking at a lot more in crosses, touches, tackles, etc.
Question: Another area where you came to the fore is politics. What is the difference with politics?
I think politics is different for a lot of reasons. You don’t have the competition you have in sports. Politics is about how the media talks about it. If you actually look at polls, the public is a lot more chill than the media. They focus on pretty basic stuff. The media will accentuate the outliers. An Obama poll showing him winning in Utah would get all the media attention.
Question: What were the polls missing?
The polls were fine. The people couldn’t read the polls correctly. When you have so much data like you have now, it’s hard to read. The majority of polls showed Obama in the lead in the last month. It was a pretty basic thing. Look at polls in swing states, take an average and you saw the result.
Question: How do you see this shaping up?
One of the things about prediction is that you can make a decent prediction in the short term but it becomes a lot more fuzzy the further out you predict. All the indicators show a really close election in 2016. To a first approximation it’s close. Hillary seems reasonably safe in a Democratic primary.
Question: You are also an entrepreneur. One thing you have to do is how to communicate the data. How do you and your colleagues do that?
The first part of it is to understand the question. We are trying to avoid putting three unrelated items together. We are doing basic techniques and then communicate it. If you can’t communicate the method then you missed. He used an example of an academic paper with bad writing that can obscure brilliant insights.
Question: So, if you don’t know your data then you should be a poor writer?
We want our writers to know their subject and data really well. We try to reverse the shouting and screaming when people don’t know their topic.
Question: What can people take away from your approach?
It’s about asking the right questions. We use some Adobe products to understand how our site is performing. We may not be using the right measure. We want to measure according to business objectives to get out of the short term. It used to be we had really great way to measure offense but really poor ways of measuring defense. Billy Bean’s teams started with poor defense. As he measured better, his teams defense got a lot better.
This stuff takes time vs a quick fix.
Follow Us
