Baseball Prospectus - Chat with Nate Silver

Chat: Nate Silver

Welcome to Baseball Prospectus' Thursday October 09, 2003 3:00 PM ET chat session with Nate Silver.

Nate Silver is an author of Baseball Prospectus.

Nate Silver: Greetings from Chicago, all, where the only thing higher than the spring-like temperatures is the blood alcohol content of Cubbie fans. Oh, that and Brad Penny's ERA. I'll start taking your questions in a couple of minutes.

Colin (Boston): How did you and Jonah do in Tout Wars? What will you do differently next year?

Nate Silver: Jonah and I (and Will Carroll) finished second. Couldn't get quite enough pitching down the stretch. Nevertheless, we were pleased with the result. The PECOTAs held up well, and Jonah pulled off some masterful trades. In truth, I'm not sure that there's anything we'd do differently; we were facing some tough competition, and there's an element of luck involved in any fantasy game.

Okay, one thing: Jonah should have listened to me about Scott Podsednik.

Greg S. (Pequannock, NJ): Nate, how exactly does PECOTA gather data on players who haven't played a day in the major leagues? Does it only use minor league seasons, and if so, how reliable are those in prediction?

Nate Silver: Good question. PECOTA uses Clay Davenport's minor league translations, which are the best in the business. Are minor league statistics reliable? The simple answer is yes, although some numbers hold up better than others - power stats translate better than walk rates, for example - and PECOTA accounts for this. It's also worth keeping in mind that the forecast variance is generally higher for any young player, whether he's been in the majors or the minors, and PECOTA accounts for that, too.

Robert L Krahn (Phoenix, AZ): Nate, Looking forward to Shandler's Fall League Symposium. Hope to see you there next year. Alot of Games and Great Talking. I like others need to start looking over our teams and come up with keepers. Using your Pecota Cards what should we be looking for to help us with this ? Could you give us a example of both a hitter and a pitcher, if you look at them differently. Talking about your cards when will you new ones be out ? Thank You For Your Time. Bob Krahn

Nate Silver: We will almost certainly have the PECOTAs out earlier this year. We are aware of just how early fantasy drafts can start, so we're going to make every effort to get them ready for Scoresheet drafters and so forth.

The strength of PECOTA is that it can identify particular combinations of attributes that are favorable or unfavorable. To use an example that I've used before - Jeremy Giambi. The combination of very high walk rates, very high strikeout rates, and very poor athleticism (as PECOTA measures by speed and height/weight) doesn't tend to age very well. Conversely, a power/speed guy like Carlos Beltran - those guys often have late peaks.

With pitchers, identifying breakout candidates isn't quite so exact a science, as there aren't as many statistically definable attributes to work with. Pitchers with high strikeout rates and moderately high walk rates - I'm thinking someone like Matt Clement or Kerry Wood - sometimes break out in their mid-20s if their command improves at all. But only sometimes.

Byung-Hyun Kim (Dog House): Am I finished with the Red Sox? What do you see in my future?

Nate Silver: Doesn't it seem like managers get more complexes about players during the post-season than at any other time, just when they should be coming out with all their guns firing?

Seriously - Boston is a brutal media market and a provincial city, and Kim is, by all accounts, a sensitive guy. I think there's going to be a better fit elsewhere: San Diego, Toronto, who knows.

Kevin (Austin): Prior and Zambrano had over 200 IP before the playoffs. How long before they break down? Also, does Juan Cruz have any sort of future as a Cub?

Nate Silver: I'm not sure that the Zambrano hasn't broken down already. He's running on, what, five bad starts in a row now, and I have a working theory that the trickier the pitch (Zambrano's bread and butter is his sinker), the more vulnerable a pitcher is to the effects of fatigue.

Prior I think will be okay for the balance of the playoffs. Longer-term, there all of the usual concerns related to high usage. The degree to which his mechanics insulate him from that is not something I'm ready to take a position on just yet.

I think the Cubs would trade Cruz if they got an attractive offer for him.

Bill Johnson (Los Alamos, NM): "Working Late" was one of the best baseball articles I've ever read, Nate; congratulations, and I wish I'd been there -- and I say that as a Cardinals fan.

On to business. Do you think that this year's post-season antics are going to force baseball to get serious about the problem of players trying to injure players? The Fick episode was flagrant, but is it really any worse than one of the game's great control pitchers throwing two pitches at a guy's head and saying "oops, it slipped"? Isn't it time for some rules changes involving intentional attempts to injure, or at least uniform enforcement of the rules that exist?

Nate Silver: Glad you enjoyed the piece. To be a little bit Machiavellian about it, I think some degree of 'violent' conduct is going to tolerated by baseball. The sport has long had a complex about not being as 'tough' as other sports, and if there are a couple of collisions at home plate, or the hitter gets a little testy after being brushed back - well, to some extent, that's good for ratings.

But the balance that needs to be struck is when guys start getting hurt. The NFL has established that balance by protecting (to some extent) the quarterback, and leaving everyone else (to a large extent) vulnerable. I don't think we're at the point in baseball, yet, where we need to worry about that. Rather, I think there are more of these types of plays in the postseason, because guys are more willing to risk long-term injury for the sake of short-term gain.

The Fick play was just plain silly.

Mike W (Chicago): Nate, what do the ladies respond to more, the U of C pedigree, or the number-crunching job and hobby?

Nate Silver: It's the package, baby. The whole package.

Bill (Houston): Have you considered using Pecota to evaluate the performance of managers? It might have to be done on a multiple year basis, but it would be interesting to see if there is a correlation between the manager and his players' actual performance relative to their Pecota projections. Maybe Dusty really does get more out of his players than another manager would.

Nate Silver: Sure, you could use the PECOTAs – or any set of forecasts, really – to perform those kinds of evaluations, but it might not be as simple as it looks.

There's some pretty good evidence, for example, as detailed in this year's book, that the Giants really did have a disproportionate number of veterans who performed better than expectations during Dusty's tenure. The degree is almost certainly statistically significant.

But the cause and effect can be problematic. Vets did well in an organization that included Dusty – but the organization also included Brian Sabean, one of the league's sharper GMs, and Stan Conte, perhaps its best trainer. It's easier to sort through the "what"s than the "why"s.

Eduardo (Park City UT): Hello Nate. My question: What do the Oakland A's need to do to get out of the first round of the playoffs? Beane's on record, of course, as saying that the postseason's a coin flip, but that's a testable hypothesis. And if you test it, using WP against strength of schedule (.500 for the regular season, .600 for the postseason), you find, using a pooled variance estimate, that there is a statistically significant difference between the A's performance in the regular season and postseason, at the 90% confidence level. Why do you think this is, and what can the A's do about it?

Nate Silver: I think the A's need to work on their baserunning, quit taking so many strikes, and keep Tim Hudson chained to his hotel bed.

Seriously: I think the post-season has brought to light certain of the A's limitations, but I don't know that those limitations are inherent to anything about the post-season itself.

wmcdonal56 (South Bend IN): I cant believe you would participate in the Cub-propaganda machine. May you only buy Tickets through scalpers, and throw 120 pitches each start!!!

Nate Silver: Do I think Wrigley is a unique, beautiful place to watch a ballgame? Yes. Do we support everything that the Cubs have done this year? Uh, wait for the book.

Jim Hendry (Chicago, IL): Nate, Can you believe that my acquisition of Tom Goodwin looks pretty good at this point? Me neither. I swear Dusty's preternaturally gifted.

Anyways, what do you think I should do to improve the team over the offseason. Obviously, we've been successful and Tribune Co. would rather that I stand pat, but we only won the division by one game and had the fewest wins of all the playoff teams. Dusty wants me to bring back Stan Javier. I don't know about that. What do you think?

Nate Silver: The doomsday scenario for the Cubs if Sammy's second half decline proves to be the new reality, and if, content with the playoff run, they don't make improvements elsewhere to hedge against it. So, yeah, I think it would behoove them to acquire another bat. I know that BP has not generally been high on Ivan Rodriguez, but he could be a good fit.

Oh yeah, and give Hee Seop Choi another chance.

Jonathan Adelman (Laramie, WY): Hi Nate. Thanks for taking the time to chat! Why don't more analysts don't get away from picking "magic numbers" and instead project confidences in a range of values? And, where did you get the idea to correlate things like body type and performance?

Nate Silver: Without getting too philosophical about it, people are predisposed to having yes/no answers. They'd rather have you tell them, say, "Jose Cruz is going to have a breakout season", even if that's patently misleading, then something with a bit of nuance – "Cruz is more likely than a lot of other players at his age to have a breakout, but these things are not absolute".

Now, when you're doing your analysis, writing your findings up – then I think at some point having a qualifier on everything is only going to make your writing weak. It's implicit that your response is a judgment call, rooted in the numbers, but not bound by them. But in terms of the forecasts themselves – I think it's "right" to own up to the fact that our predictive powers only go so far.

As far as looking at body type (really just height and weight) – I had the data, I plugged it into the model to see whether it improved the forecasts, and I was surprised by the extent to which it did. It would probably improve the forecasts even more if we had a fully accurate set of heights and weights. I think Mo Vaughn is still listed at a svelte 180 lbs or something.

Dr. C (Mobile, AL): The one thing I see off on Pecota is playing time. How can that be fixed or is that a flaw in the system? Also, how accurate were the breakdown and attrition numbers?

Nate Silver: PECOTA tries to estimate playing time endogenously – that is, based on a player's history of playing time and performance, without any further human input. There are strengths and weaknesses to that approach. It's a purer solution in terms of methodology, and the system picks up on certain things that a human might overlook – for example, once a player starts to lose playing time due to injury (think Ken Griffey Jr.), his problems are likely only to compound in future seasons.

But it won't pick up on the fact, say, that Shawn Estes is going to throw a lot more innings than he should because Dusty likes him too much. Or, to use a morbid example, PECOTA didn't know that Darryl Kile had died. For next year, we are considering running updates in-season that would account for this kind of information, as well as take advantage of Will's knowledge on injuries.

Joe Gargon (CT): Nate, is there any chance we'll be able to see all PECOTA numbers on one page, instead of having to look them up on individual player pages?

Nate Silver: Yes.

Jonathan Adelman (Laramie, WY): Andruw Jones or Carlos Beltran, based on your gut feeling. And, recognizing that (1) it doesn't seamlessly incorporate defense, and (2) even if it did, popular opinion and play-by-play data provide nearly opposite opinions on Andruw's defensive ability, does PECOTA agree with your assessment?

Nate Silver: My gut feeling – Gary, am I allowed a couple of these? - is that Beltran will contend for the MVP next year. Rany and I had the chance to talk with a couple of the Royals when they were last in Chicago, and they had a lot of admiration for Beltran: he can do certain things, particularly with his stride in the outfield and on the bases, that other players just can't, and he seems to have his head screwed on straight enough.

As for Jones, my gut is that we're going to have to be content with him being merely a very good player, and not a great one. This is the second year in a row now that his speed metrics have been way down, and the improvements made after age 25 are usually fairly marginal.

UK (Chicago): Concerning Grudzielanek, even though he surpassed his PECOTA forecast pretty substantially and his career is a long enough to determine this could very well be a fluke (2nd best in 9 years), I'm having troubles creating odds of him repeating his 03' year or regressing, any guess?

Nate Silver: I don't think the system will like Grudz very much, because his advancing age will offset most of the improvements he made. Now, it's possible that Grudz has had an epiphany of sorts, realizes that the only way he's going to stay in big league ball is if he can keep his OBP up, and has changed his approach at the plate as a result. I'm not being sarcastic; I think that's entirely possible. But I don't know that, and PECOTA certainly won't.

Greg Tamer (Purdue University): Will you be evaluating PECOTA's predictions for the 2003 season AND releasing the findings to the Premium readers?

Nate Silver: Based on the preliminary stuff I've looked at the system held up really well. But unfortunately, we're already in the heart of book writing season, and I already have chapter authors bitching at me for PECOTAs, so the process will be a little backward: evaluate the 2003 results, make any necessary improvements for 2004, run the new numbers, and then go back and talk about 2003.

Also, I hope this won't sound like a cop-out, but there's a danger in being too results-driven here. It's the way PECOTA approaches the problem of forecasting that distinguishes it, and not necessarily a few points of MSE here and there.

I don't type this fast, by the way; a couple of these answers were prepped ahead of time.

batpig (San Francisco): What does PECOTA have to say for itself on the HUGE miss on Mark Lorreta? The 90th Percentile prediction was BELOW his career averages!! I know that journeymen second baseman don't have bright futures after 30, but isn't it preposterous to suggest that, in a best case scenario, a player wouldn't even mach his career averages??

Nate Silver: I don't know that his career averages are the relevant thing here, however. All the research I've done - and most all the research that everyone else has done - suggests that it's the most recent three years that count. The value added by including information from years previous to that is marginal, at best. We wouldn't have been criticized for predicting that Rickey Henderson had no real chance to hit his career averages, for example.

But, yeah, PECOTA missed on Loretta, missed big. There are going to be a couple of those every year.

Kate (Grosse Point): Yankees? Next Year?

Nate Silver: Tejada at short? Jeter in center? Bernie in right? Probably not.

xianb (champaign, il): Thanks for your piece on the Cubbies. I had been so sick of hearing "95 years blah blah blah" everytime the announcers couldn't think of anything to say that I was worried that maybe it'd be better if they just lost 90 games again. Will Baseball Prospectus' success encourage other sportsjournalists to follow suit and actually research stories and write well or are we stuck with the same old stories about Veteran Experience, etc. forever?

Nate Silver: There are some really good beat writers out there, like Posnanski in Kansas City, Curry in New York (though he's not really a beat guy anymore), and so forth. And people like Jayson Stark and Peter Gammons got where they are for a reason.

But as with any profession, there are also plenty of burnouts, and plenty of free-riders. I think there's a greater chance that we convert some members of the media into using more of "our" kinds of statistics than that we can revive the passion of someone who has lost interest in the game

Theo Epstein (New York): Which one of my big name free agents after 2004 will I keep? Who should I keep?

Nate Silver: I suspect I'll be in the minority among BP authors, but I'd actually prefer Pedro if forced to choose. I think Theo would prefer Nomar, however.

danielj (Davis): How much value does adding body type factors into the equation give PECOTA over a system that overlooks them?

Nate Silver: It's hard to quantify that sort of thing. Five percent? Seven percent? I dunno. Body type isn't the most important factor by any means. But it does improve the projections enough to be worth including, and a whole bunch of marginal improvements can make the system a lot better in the aggregate.

Mark (San Leandro, CA): Is it now fair to call Billy Beane overrated? His tier two signings of FAs and his own guys have been Bonifayesque. Terrence Long through 2005, Scott Hatteberg to a 2+1 deal, paying above market for Mecir, signing Dye to a too-long and too expensive deal after his career season. If John Mabry hadn't played out of his mind last season, he'd look even worse. Isn't Beane really just another GM with better hair, good looks, and a better press agent?

Nate Silver: I think Beane and the A's still do a superior job of using objective measures to evaluate performance, but that doesn't mean that they're immune from the subjective factors that creep in to compete with them. Billy, yes, builds up attachments to certain of the people that work for him. That's one of the inevitable risks of being human.

Josh (Queens, NY): Glad to see BP in the New York Sun! I have a question about your ALCS Preview. Huckabay's on record at ESPN.com as picking the Sox to win the World Series, but you two picked the Yankees to win the ALCS in six. What's the deal? Did you convince him of the error of his way?

Nate Silver: I picked the Sox to win the WS at the start of the year, too, but the condition of Johnny Damon, the condition of the Boston bullpen, and the particular match-up advantages that the Yankees have in this series (lots of LHP, and lots of guys that work the count) induced Gary and I to think otherwise.

Hee Choi (Miami, FL): Just wondering if the committee that worked on the Free Santana campaign was looking for work. Campaign contributions welcome.

Nate Silver: You might want to contact the Dick Gephardt people, too.

Dr. Googol (Evanston, IL): Hey, SABR has an annual convention, Ron Shandler does his thing in Arizona, why don't you guys put together a gathering in say, Chicago? I've been to the Pizza Feeds in Chicago and Palo Alto, and they're a better time than a typical White Sox game. Do it in March, host fantasy drafts, bring Wolverton and Huckabay out here for talks on front offices, business, gossip, player evaluation, and broadcasting, bring Pappas in for Business of Baseball stuff, Chris Kahrl on transactions and rules. It will rock! Beer, Women, Baseball, Fantasy Drafts, T-Shirts, maybe a golf game, wiffle ball, poker! Set it up, Natester! People would lay out enough cash, and you could get Kenny Williams to come out, probably Billy Beane, and a bunch of players. Do it!

Nate Silver: We have a policy of keeping the cost of Pizza Feeds and other BP-sponsored events as reasonable as possible.

That policy would be irreconcilable with getting a sufficient number of Women to attend such an event.

Nate Silver: That's all, folks. Hope you enjoyed it.