Every year I try to project every team’s record and runs scored and allowed, using as much information as is available to me in the waning days of March. I do it because it’s fun, and because the process of making those predictions is very educational for me in the ramp-up to the season. The process, rather than the end results, is what is important, because the chance of getting many teams’ overall records or run differentials correct is fairly slim. The value of the pieces I write at that time is in the analysis, the words; the numbers are for information purposes only.
However, it’s important to learn from any process, so as we close out 2009, I want to take a look back at my predictions for the season and see what lessons can possibly be learned. I’ve written this piece before, so if you want some deeper explanations of process, you can check out the recaps of previous seasons. One key change, as I sit down to this, is that I’m no longer making a global adjustment based on overall run scoring. In the comments section of last year’s piece, I was convinced that the mistakes I’m making that show up in the overall run totals aren’t so much an error in misprojecting run environment globally, but the adding and subtracting of small errors in specific team projections. So the comparisons this will be between my unadjusted predictions and the actual numbers. For what it’s worth, this year was as close to dead-on as I’ve ever been, projecting 22,906 runs in leagues that scored 22,408, a difference of a tick more than two percent. It’s not Nate Silver on Election Night, but it’ll play.
Let’s look at the top and bottom five by the Err score, which is the difference between a team’s projected and actual run differentials:
Actual Predicted Team RS RA RS RA Err Marlins 772 766 797 792 1 Rays 803 754 766 719 2 Braves 735 641 799 703 2 Astros 643 770 731 843 15 Mariners 640 692 641 671 22 Mets 671 757 812 712 186 D'backs 823 748 720 782 137 Indians 847 822 773 865 117 Nationals 710 874 707 757 114 Angels 883 761 741 726 107
Maybe I shouldn’t be, but I am pretty happy with that first list. To peg three teams essentially at their differential-and to be fairly out of step with the field on two of them-is something I’ll hang my hat on. With that said, look more closely at the numbers, and you find that while I had the differentials close to correct, the path to being so wasn’t all that helpful. The Mariners’ figures are close enough, but in the cases of the Rays and Braves, I missed on both runs scored and runs allowed, but by both in a way that allowed be to nearly nail the differential. I remain unsure if this is the most valid means of evaluation-is it more important get the elements of differential correct, or the differential itself?
I’ll take a pass for the Mets, who whiffed by 141 runs on offense because of all their injuries, the collection of which was largely unforeseeable in March. As far as the Diamondbacks and Indians go, I’m open to the idea that I’m systematically overrating “good” organizations, as I seem to miss on those teams to the high side with some frequency. I’ve certainly been accused of bias regularly, and I think there’s a case to be made that I have to be more careful about falling in love with a GM, a front office or a particular team’s offseason, and take a skeptical eye with teams that, in my mind, have a certain progressive seal of approval. That’s not me saying, “I whiffed on the Diamondbacks and Indians because I’m biased,” so much as me saying I’m willing to think about the charge. In the Tribe’s case, remember that they made another dump trade this season, a deal that pushed them into the AL Central cellar by season’s end. That, and all other midseason deals, are a total non-factor in preseason analysis. You cannot predict who will make what deals at the deadline and how that will affect projections.
Another way to look at the problem is to add up the absolute value of all the misses. If you do this, you get Net Error Score (NES) of 3496, which is low for my history with this task. I’m making 60 predictions and missing by an average of 58 or so runs a pop, which doesn’t seem competent. Net Error Score is also a way in which to evaluate the various team predictions; in Err, guessing 30 runs high on runs scored and 30 runs high on runs allowed is a score of “zero,” or a perfect match, In NES, it’s a 60. Here are the rankings by this method:
Actual Predicted Team RS RA RS RA NES Mariners 640 692 641 671 22 Giants 657 611 657 658 47 Padres 638 769 611 792 50 Marlins 772 766 797 792 51 Phillies 820 709 819 774 66 Rangers 784 740 860 888 224 Yankees 915 753 789 675 204 Cardinals 730 640 772 788 190 Mets 671 757 812 712 186 Angels 883 761 741 726 177
I can’t figure the Phillies, who I’m accused of hating pretty rigorously. Their run prevention was a lot better than expected, which has to be largely their good defense, because the pitchers don’t add up to that at all. Even giving them a half-season of Cliff Lee and some decent work by Pedro Martinez wouldn’t account for 70 runs. The bullpen wasn’t that good, and we all know about the year the closer had. Yet they allowed 65 runs fewer than I expected and ran away with the NL East. The other teams on the good side of the ledger don’t tell us much, if anything. On the bad side, you had the Rangers remaking themselves as a pitching-and-defense squad, with Elvis Andrus at the helm. The Cardinals got a dream season from Joel Pineiro and Chris Carpenter‘s return, which explains much of their run prevention. The Mets show once again, as do the suddenly-cheap Padres and the suddenly-addled Giants.
My goal is to predict runs scored and allowed, and let wins and losses fall where they will. With no team possessing a :clutch” skill, it’s reasonable to figure that wins and losses, with maybe a small adjustment for bullpen quality, will be what the RS and RA say they will be. For completeness, though, here are the W/L predictions, formatted just like the above charts:
Team Actual Predicted Pirates 62-99 62-100 Brewers 80-82 79-83 Blue Jays 75-87 76-86 Rays 84-78 86-76 Dodgers 95-67 92-70 Mets 70-92 92-70 Indians 65-97 84-78 D'backs 70-92 88-74 Nationals 59-103 74-88 Angels 97-65 83-79 Padres 75-87 61-101
This is my last column for Baseball Prospectus. My contract ends today, making me like any number of free agents looking for work. No hard feelings or recriminations, just two entities doing business.
When I left BP-temporarily, as it turned out-in 2002, I wrote a big piece about it. I’m not doing that today. I just want to thank everyone who reads us, who listens to us on the radio, watches us on TV, buys books, writes comments, sends in e-mails… all of the people who have helped us build Prospectus into what it is today. I never in a million years thought I’d get to write about baseball for a living, but I did. I did thanks to Gary Huckabay and Clay Davenport; Christina Kahrl and Rany Jazayerli; Dave Pease and Keith Woolner and Keith Law and Nate Silver and so many other people who built Prospectus in the 1990s. In the 2000s, we benefited from the grace of men such as Rob Neyer, Peter Gammons, Billy Beane, Brian Kenny, Jeff Erickson, Bernie Miklasz, Chris Stone, Michael Epstein, and Louie Belina, who spread the word about what we were doing, who helped us make the leap from niche to mainstream, even as the mainstream was coming our way. I cannot ever thank these people, or the hundreds like them who simply pointed at BP and said “read those guys,” enough.
From where I sit, I have a particularly personal view of the role chance plays in a life. If I hadn’t gone back to USC, if I hadn’t gotten an e-mail account and found rec.sport.baseball, if I hadn’t joined NASA and met Rany, if I hadn’t done any number of things, I have no real idea of what I’d be doing today. I can’t even guess at it. I got blindingly lucky, beyond anything I personally deserve, to have fallen in with such a talented, dedicated, hardworking group of people, and for that, I am forever in their debt, as I am in yours, readers. Thank you, for everything.
“What are you thinking about?”