- Peter Keating, ESPN Senior Writer
- 0 Shares
PETER KEATING: So I have to ask: What do you think of the new Internet meme Drunk Nate Silver?
NATE SILVER: [Laughs] If only people knew the real drunk Nate Silver. I'm not so dark, necessarily. I just get into stupid arguments about sports with my friends. It's one thing when you have yourself, but it's another thing when you start to symbolize a movement and you don't really have control over it in a certain sense.
KEATING: So you believe you can successfully insulate yourself from this giant, maniacal cult that's growing around you?
SILVER: I hope so.
KEATING: In your book The Signal and the Noise: Why So Many Predictions Fail -- But Some Don't, you say you got your first baseball card when you were 7, read your first baseball analysis when you were 10 and invented your first stat when you were 12. What was that stat?
SILVER: It was some mishmash. It was like batting average times runs divided by five. It was supposed to be some superstat. But I don't think I really thought through whether it was really driving at anything in particular.
KEATING: Did your interest in developing PECOTA, a sabermetric system you created that forecasts MLB player performance, come from the data and what analysts were looking at?
SILVER: I looked at Total Baseball and later Baseball-Reference.com and just saw how different players' careers developed. I thought, Let's create a blueprint for different options that a player might have, how the tree might unfold. To me that seems much more satisfying and intuitive as a baseball fan, to say, Look, you have this third baseman, and his stats up to a certain age are somewhat like Mike Schmidt's but also somewhat like Gary Scott's or Kevin Orie's. It's having a memory, kind of, for all the ways that players can develop.
KEATING: So there are two things there. One is the idea that when you look at the back of a baseball card there's a story there. When you see Carl Yastrzemski go from 40 doubles to 40 home runs, you almost imagine his body filling out and that he learns how to hit at Fenway Park.
SILVER: Oh sure, yeah.
KEATING: So the numbers are always connected to storytelling. But what you're also saying is there's a range of probability. It's interesting because you've been in two fields -- baseball and politics -- that people are drawn to because they want certainty. They demand it. It's almost like you're fighting
the tide by allowing for the fact that there might be a reasonable margin of error in prediction.
SILVER: Oh sure. And when I wrote the book and you talk to people in other fields, the forecasters are not very good. In baseball we do a lot better; the predictive analytics are a lot better in baseball than in almost any other field, I think. But still people say, Well, you can't account for the human element, and, well, we can, right? We can quantify how much we don't know. We can kind of say, Here's what we can tell you; we can tell you this player is a good enough prospect that it's unlikely he's going to totally bust out, but we can't say whether he's going to be a star or a league-average player. Or we know that sometimes pitchers with this profile, say, have a lot of strikeouts, a lot of walks. We know they sometimes develop into stars, but we can tell you the odds are about 1 in 10, and maybe there are things this guy does that make him more likely than the average player to be one of those 1 in 10.
KEATING: Does being a smart fan mean something fundamentally different from what it meant 20 or 30 years ago when the data wasn't as good? To be a smart fan, you have to allow yourself room for that margin of error or for that uncertainty, build that into how you think. You have to think differently.
SILVER: Yeah. Look, baseball is still wonderful in that you do have to play out the seasons and the games. That's another reason I like having these terms of probabilities or odds. We know that the odds of a 2-seed beating a 15-seed in the NCAA tournament are about 5 percent or something. But it doesn't mean it's not fun when that 5 percent comes up.
KEATING: My favorite passage in the book is when you write about how enough time has passed for you to evaluate the PECOTA forecast. The guys you projected to do the best as minor leaguers added, I think, an estimated 546 wins, or generated 546 wins for their teams, and the top picks by Baseball America generated 630. And on the surface that's disarming because it's like, Well, look, the scouts actually did better and were adding
something. But it undermines the whole stats vs. scouts debate because scouts use stats, right? So the whole question of there being a diametrically opposed war is sort of ridiculous.
SILVER: That's right. And I think if there was a time when we were kind of in the Moneyball, Sharks vs. Jets complex, we passed through that phase pretty quickly. And one reason we celebrate baseball in the book is that, Hey, you do have competition, and competition tends to get rid of bulls -- faster than other things, right? If you're losing 90 games every year, then sooner or later you're going to lose your job. And sometimes it's unfair if you're not given the right resources, but there are incentives to improve and to get better. And really the value in baseball comes from young talent, just the way the economics are structured. And so scouts and statheads are only trying to figure out how minor league players will do. I mean, that's an amazingly valuable resource, for a team to have any method at all of doing that. But it's all fusing together too, where now you have some teams using stats to evaluate their scouts.
KEATING: The other interesting thing underlying the PECOTA evaluations is that you can say the scouts have value added -- and you can quantify that value added. And it's around 15 percent.
SILVER: Yeah. For some reason that figure of adding 15 or 20 percent accuracy seems to come up in different contexts in the book. So, for example, human weather forecasters improve on the computer output, which is very good, but they improve enough by about 15 or 20 percent. Subjective forecast or judgmental forecasts of the economy are not very good, but they're still better than purely model-based ones by about 15 or 20 percent. If we have a reasonable grounding in the stats and the probabilities, then we can add some value on top of it. Whereas, if you think you're going to create miracles, then you get yourself in trouble.
KEATING: You did fantastically well in predicting the presidential elections in 2008 and again in 2012. Yet there were those in the Romney camp who
were shell-shocked by the results. Do you think there's an analogy to that in sports?
SILVER: There is always the tendency in sports coverage to [look at] the outliers, the guy who's been on a crazy hot streak. And of course, it's fun when he wins 14 games in a row or something. But sometimes people read too much into it. People forget that you have thousands and thousands of at-bats every season and all these different ways to look at the data. You're going to have some weird things happen. As a general rule, organizations of any type have difficulty with honest self-evaluation. So a major league team thinks, We have an 85-win roster, and we're gonna go and sign this free agent who will give us six more wins and make us a 90-win team and push us over the playoff hump. Well, that could be a correct line of thinking, but if you're not being honest about the 85-win thing ... Because what usually happens is that you say your third baseman was injured half the year and he'll be back full time now, and this rookie pitcher is going to be really good and make 28 starts for you. But you forget that someone else is going to become injured.
SILVER: And that your shortstop, for reasons that had nothing to do with his talent, just had a good, lucky year. People tend to blame all the ways in which they're unlucky without accounting for the fact that we probably got lucky in other areas as well.
KEATING: You made the case for Mike Trout to win AL MVP, but Miguel Cabrera won it. Is that the kind of thing that gets you irritated?
SILVER: No, I don't get irritated. Compared to politics, sports is much more civil and reasonable. Basically you get a lot of people in politics who create and manipulate their own truth. You don't have people who are as delusional in sports, I don't think.
1dMarc Stein and Ramona Shelburne