|
The Prospectus Hit List is back for its sixth season! This year, on the orders of Dr. James Andrews, I’ll be publishing separate AL and NL editions, as well as publishing a combined ranking for those who wish to quibble over interleague superiority. As ever, we at BP are determined to put our best foot forward when it comes to predicting the upcoming season, and the foundation of our predictions is PECOTA. From the basic projections, our staff adjusts for expected playing time, strength of schedule, reliever leverage, and team defense to generate the Projected Standings, which have been updated frequently throughout the spring based upon the latest news and analysis. The Hit List Factors below are the Pythagenpat winning percentages derived from the latest runs scored and runs allowed projections, which are included in parentheses at the end of each team capsule. As you quibble with the rankings-I certainly have-remember that projections are not destiny; they’re shorthand for a wider range of probabilities centered on the stated won-loss records. As proud as we are of our system’s track record, we’re eager to put the theoretical behind us and watch the season unfold. Play ball!
| Rk |
Team |
|
|
Hit List Factor
|
|
Comment |
1
|
Red Sox |
95-67
|
–
|
.592
|
|
Defensive Posturing? New England worrywarts may fret about a lack of offense straight out of some Borgesian nightmare. Indeed, the winter’s key arrivals-John Lackey, Adrian Beltre, Marco Scutaro, Mike Cameron-tilt more towards run prevention, bolstering the rotation both directly and with a renewed commitment to defense borne of last year’s sorry 28th-place ranking in Defensive Efficiency. As for the offense, relax chowdaheads, we’ve got the Sox projected for a True Average of .270 (second-best in the majors), not to mention the top record in all of baseball. (847 RS, 696 RA) |
2
|
Rays |
92-70
|
–
|
.571
|
|
Rays-ed Hopes: The darlings of 2008 got a harsh lesson in come-back-to-earthiness last year, but this team is so stacked it should carry an NSFW tag. The addition of Wade Davis to the rotation, the continued development of David Price, and a bounceback from B.J. Upton all add to the upside achievable by this talented corps, headed by MVP candidate Evan Longoria and the lineup’s Swiss Army knife, Ben Zobrist, and backed by an organizational depth which is simply unrivaled. (820 RS, 705 RA) |
3
|
Yankees |
91-71
|
–
|
.566
|
|
No rest for the World Champions. Despite their efforts to get younger-punting Johnny Damon and Hideki Matusi for Curtis Granderson and Nick Johnson-their success still hinges upon whether Derek Jeter, Jorge Posada, and Mariano Rivera can continue defying the aging process, not to mention whether Alex Rodriguez‘s hip remains intact. PECOTA sees the Yanks having their hands full battling the younger Rays and deeper Red Sox, and that’s without accounting for their efforts to ward off the inevitable distractions surrounding The Jobacalypse. (859 RS, 749 RA) |
4
|
Athletics |
83-79
|
–
|
.517
|
|
Averaging 75 wins since 2007, the A’s make for a surprising choice to break out of the latest AL West scrum. Indeed, “breaking out” is a stretch given how tightly bunched around .500 its four teams are. Unless and until the likes of Chris Carter and Michael Taylor show up, the A’s lineup is merely ordinary, a great distance removed from the walk-and-wallop days, but at least it’s not dependent upon the wishful thinking of healthy seasons from Eric Chavez and Bobby Crosby. Headed by ace-in-the-making Brett Anderson and a roll of duct tape that will be stretched to accommodate both Ben Sheets and Justin Duchscherer, it’s the rotation which will make or break Billy Beane‘s boys. (731 RS, 705 RA) |
5
|
Rangers |
83-79
|
–
|
.512
|
|
Last year’s upstarts won’t sneak up on anyone this time around, and while the holdover talent is plenty good, what’s really interesting here-particularly in light of last year’s Elvis Andrus-led defensive improvement-is the daring remake of the rotation into one that can miss bats. Towards that end, not only have they brought in Rich Harden, but they’ve brought Colby Lewis back from Japan and C.J. Wilson forward from the bullpen. As with the A’s, things could get very interesting if and when this team dips into its own reserves to give Neftali Feliz and Justin Smoak key roles. (800 RS, 780 RA) |
6
|
Mariners |
82-80
|
–
|
.509
|
|
The additions of Cliff Lee, Chone Figgins and even Milton Bradley have turned Jack Z‘s Mariners into the hip pick in the AL West. PECOTA sees the upgraded defense amply supporting the somewhat shaky rotation behind Lee and Felix Hernandez, but it’s got plenty of concern for an offense where only the front four hitters-Ichirio Suzuki, Figgins, Casey Kotchman and Bradley-forecast for True Averages above .260, with Kotchman’s .262 still a significant drag from the first base spot, to say nothing of the three hole. In the Pacific Northwest, 3-2 games may be the new grunge. (727 RS, 713 RA) |
7
|
Twins |
81-81
|
–
|
.501
|
|
The loss of Joe Nathan to the surgeon’s table notwithstanding, there’s plenty of reason for optimism in the Twin Cities. Start with the new ballpark and its pricey-but-possible byproduct, the Joe Mauer contract extension. Move along to the upgrade to a real live middle infield (Orlando Hudson and J.J. Hardy) and the returns to health not only of rotation anchor Kevin Slowey but of potential staff ace Francisco Liriano (take the under on that projected 4.58 ERA) and you’ve got hopes for Central success that should remain undimmed even amid the dreaded closer-by-committee controversy. (815 RS, 813 RA) |
8
|
White Sox |
79-83
|
–
|
.487
|
|
Fronted by Jake Peavy for a full season, the rotation should be a strength, but Ozzie Guillen can emphasize the running game until Juan Pierre‘s cow is waved home, and the White Sox will still live and die by the long ball. Judging by the forecast for a .417 SLG (10th in the league, and in a hitter’s park), that means facing this lineup’s utter mortality. Full years of Carlos Quentin, Gordan Beckham, and Alex Rios will help, but the Sox have too many low-OBP sinkholes, and the Mark Kotsay/Andruw Jones DH plan is a big bowl of wrong-even more so if Oz makes good on his threat to bat Kotsay third. (748 RS, 769 RA) |
9
|
Tigers |
79-83
|
–
|
.486
|
|
Dontrelle Willis‘ startling overtures towards adequacy underscore the Tigers’ biggest challenge: getting more than 1.7 WARP from a quintet on whom they’re spending $65 million this year, almost half their payroll; they’re left banking on Willis and Jeremy Bonderman while sinking the cost of Nate Robertson. Easier to understand is their bet on a pair of former Yankees, Austin Jackson and Johnny Damon, to upgrade the top two spots of their lineup, though the team’s PECOTA forecast of a .259 True Average suggests that makeover may only scratch the surface. (762 RS, 784 RA) |
10
|
Indians |
79-83
|
–
|
.485
|
|
With no Central team set to run break away from the pack, the Indians may wind up making a competitive bid even as they rebuild. Particularly so if Fausto Carmona can deliver upon a spring which has him back in his dazzling 2007 form, helping new manager Manny Acta avoid falling back upon the eminently hittable leftes which have pockmarked recent Cleveland rotations. The return to health of Grady Sizemore, a full season of Matt LaPorta, continued development of the already-impressive Asdrubal Cabrera and Shin-Soo Choo-not to mention the anticipated mid-summer arrival of Carlos Santana-at least lift this team into the realm of sleeper, which is more than one might hope for after consecutive years of trading away Cy Young winners. (767 RS, 792 RA) |
11
|
Orioles |
78-84
|
–
|
.480
|
|
While there’s no guarantee that the Orioles can snap their string of 12 consecutive losing seasons, PECOTA thinks they can at least be within hailing distance of .500, and with Matt Wieters and Brian Matusz joining the growing cast of young studs from day one, they’ll certainly be more interesting than in years past. Which isn’t to say there won’t be growing pains, or that other potential flaws-Brian Roberts‘ back, Garret Atkins’ dying bat, Miguel Tejada‘s adaptation to the hot corner, not to mention a whole lot of average-at-best rotation filler (Kevin Millwood, Jeremy Guthrie, Brad Bergesen)-won’t manifest themselves. (797 RS, 831 RA) |
12
|
Angels |
78-84
|
–
|
.477
|
|
Fallen Angels? Despite winning five division titles in six years, the Halos enter 2010 as AL West underdogs, at least where PECOTA is concerned. Blame the losses of John Lackey, Chone Figgins, and Vlad Guerrero, the latter two to division rivals. Much hinges upon the ability of Brandon Wood to make good on his once-blue chip status, the rebounds of Ervin Santana, Joe Saunders, and Scott Kazmir, and the potential maturation of Jered Weaver into a ground-ball-generating staff leader. Don’t put it past manager Mike Scioscia to squeeze extra wins out of this projection via his skillful bullpen management, but don’t be surprised when this team doesn’t dominate the division as in years past. (796 RS, 835 RA) |
13
|
Royals |
76-86
|
–
|
.467
|
|
Zack Greinke and Billy Butler aside, the Royals have turned into a zombie flick; how else to explain the employment of undead retreads like Jason Kendall, Yuniesky Betancourt, Rick Ankiel, Jose Guillen, and Kyle Farnsworth, to say nothing of the sheer existential horror of the franchise’s past decade and a half? Hope and faith here is best left to individual quests: another Cy Young-caliber season for Greinke, a rebound for Alex Gordon, continued success mining stathead wisdom for Brian Bannister, and so forth. (738 RS, 792 RA) |
14
|
Blue Jays |
72-90
|
–
|
.440
|
|
The Doctor is Out: So the Blue Jays begin the post-Halladay/Ricciardi era facing the distinct possibility of being overtaken in the AL East by another flock of luckless birds, and this writer will have to find a new whipping boy beyond the embattled ex-GM. The lineup projects for plenty of power, particularly if Travis Snider can live up to his impact-player billing. But patience is in shorter supply, unless you’re talking about the organization’s ability to burn through young arms. Shaun Marcum‘s back from Tommy John surgery to be the Opening Day starter, but can the likes of Ricky Romero and Brandon Morrow survive this franchise’s arm-mangling tendency? (732 RS, 830 RA) |
The Prospectus Hit List rankings are derived from Won-Loss records and several measurements pertaining to run differentials, both actual and adjusted, from Baseball Prospectus Adjusted Standings through the close of play on every Thursday.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now
|
Before you -1 this comment out of existence, please consider the A's are consistently picked at or near the top of the AL West every year at BP and have finished, on average 18 games out of 1st place the past 3 years.
I'm well aware that the system has been wrong about the A's - my best guess is that a good chunk of it has to do with optimistic projections (human-influenced playing time estimates included) for injury-vulnerable guys like Eric Chavez, Bobby Crosby, and the Duke. On the other side of the aisle, the link between the Angels' recent penchant for overachievement and Scioscia's skill at running a bullpen is one I've actually explored at length on the site and in the book.
But beyond that, I'd love for you to explain to our readers how our bias is actually BUILT INTO the model. Really, do enlighten us.
But come on man, the A's have been overrated by Pecota every year since they started to suck (2007).
Rather than ask us to explain Pecota -- I don't understand it -- may I suggest you, and maybe Nate and Clay if you can get them, get under the hood and see if you can find that pesky loose valve that's causing that annoying "let's go Oakland" chant.
"What? No - I'm 100% positive it couldn't be injuries, luck and random variance. Why do you ask?"
Maybe the problem is the charged word "bias".
What we have are a bunch of approximation models for things like offensive production, defense and pitching and models try to estimate the "norm" but are rarely completely comprehensive. Close, perhaps, but not perfect.
I know you know all about statistical analyses, so this is nothing new to you.
However, ANY projection system will tend to have less than fully 100% non-skewed results.
Take the case of multiple regression. We apportion the sums of squares of the dependent variable among the independent variables in the model. The remainder is called "error" which is generally assumed to be randomly distributed. However, it hardly ever is.
"Error" truly means variance in the dependent variable that cannot be attributed to the independent variables you have built into the model. I presume PECOTA has many (and all of the most important) variables specified in your model. However, it cannot account for everything.
My best guess is this- the factors not in the model, such as manager, precision around injuries, "intangibles", "chemistry", etc.- all have small, usually inconsistent effects, but not zero effects. This results in "error" that most attribute to luck, randomness or noise, that could more accurately be portrayed as "unexpalined variance". Most of the time, unexplained variance, even in the best models, is not fully random.
I do not accuse BP of intentional bias. However, it is clear that PECOTA tends to highly rate some teams that make decisions in a manner consstent with BPs approach and tends to underrate some teams that take differing approaches. The reason is the "unexplained variance".
I'm terrified of having my heart broken. Again. A's as 4th best in the AL? I reeeeeallly want to believe it's true. I want to have a shot at avenging 2001. Slide, Giambi, slide!
I have yet to find a really good reason to deviate from the same Hit List methodology I've been using for the past five years. A certain element of the readership would like the list to reflect more immediate changes in team fortunes, but the quick and dirty studies I did on shorter increments of team performance (past week, past month, etc.) showed that they're not particularly indicative of team strength, which is to say that they added no value whatsoever to the predictive power of the Pythag-based winner percentages which underly the calculation of the Hit List factor. I may try to do a more formal study towards that end, because the original one was done as a comment to a column.
The Computer can't account for Mike Scosia's managing prowess thus the OVER on Angels wins looks juicy - especially if Santana rebounds from the Verducci effect and Kazmir pitches the way he did the last six weeks in Anaheim.
They belong one rung above the Royals at best.
Projected - Team - 2009 Runs
859 NY 915 (in 2009)
847 BOS 872
820 TB 803
815 MN 817
800 TX 784
797 BALT 741
796 LAA 883
767 CLE 773
762 DET 743
748 CHI 724
738 KC 686
732 TOR 798
731 OAK 759
727 SEA 640
On the projected side, there's a range of 132 runs between top and bottom with 4 teams at 800 and above and no one below 700. But from 2009, there was a range of 275 runs between top and bottom with a more diffused bell curve of 1 team over 900, 4 more over 800 and 2 below 700. I haven't gone back beyond 2009, but it seems unrealistic to expect all the runs to be so evenly scored between 14 teams.
Perhaps it's impossible to predict which teams are going to have outlier seasons (good or bad) for runs scored, but it seems extremely likely that some of the teams above are going to over or underperform by a significant margin to become those outlier teams. PECOTA doesn't seem able to account for this.
Teams don't end up as outliers without a whole lot of things coming up right (or wrong) - say, several guys hitting or pitching to their 70th to 90th percentiles (or 30th, or 10th). And that, by definition, is an unlikely occurrence, which no projection system is going to nail.
Now, one thing I think would be very interesting for us to try to do would be to present team run totals (scored and allowed) using some kind of probablistic distribution the way we do with individual players. If we could tell you, for example, that not only are the Yankees projected to score 859 runs but that their 75th percentile projection is for 930 runs and their 25 percentile projection is for 820 runs. That might do a better job of making the potential outliers more visible.
Does PECOTA take a player like Mauer and take his forecast (his percent chance of injury, percent chance last year was a peak year, percent chance he could improve) and then convert him into a static number (he's worth 60 runs scored) and then just add up all of the static numbers for each player on the team to get the run scored amount? Or does PECOTA give a run score range for Mauer (he's worth 40-80 runs scored) and then use those ranges (and their associated probabilities) to create a single value for the team?
Because with all of the percentages built into the system, you should be able to generate a model which could factor what happens if the bulk of the team plays at their 30% forecast versus their 70% forecast. Then you would have a scoring range for each team.
So if you use the percentiles to drive a range of forecasts, you have to somehow adjust other players as well according to how each player's playing time changes. I'm not saying it can't be done, but I don't know if it can be done well.
Clay Davenport's Playoff Odds Report does account for many of those things and provides a range for teams. I don't know when Clay is going to fire that up, though.
Take the Yankees last year. Why did they score more runs than projected? Well, the new ballpark helped a little. But the key differences were that Jeter had a season that no one his age should expect (while not unprecedented it was highly unusual), Matsui had much more playing time than expected and Swisher played more than expected (because Nady was hurt). Would you want a model that predicted that Jeter was going to have that kind of season at that age? Or one that assumed that, despite what the Yanks were saying, Swisher would play every day and Nady wouldn't? Or that a 34 year old outfielder coming off knee surgery (for the umpteenth time) would have 100 more plate appearances, triple his HR output and double his RBI output? A model has to project based on past information.
If one thinks that Liriano is going to be the pitcher he looked like four years ago -- which is certainly not impossible -- then I agree the Twins are going to be the favorite in the Central. But there is no way for a statistical model to assume that Liriano will perform at 80-90% of his Pecota projection and everyone else will perform at 50%.
These projections are just a start. They say if everyone does what they're expected to do, this is what the teams are putting on the field to start the season. And given that around 35-40 players hit each team's roster every year, it is a reasonable expectation that the folks who perform over their projection will be offset by those who underperform. But as we see every season, some teams just don't regress to that mean. A bunch of guys put together career years and the rest do what they're supposed to, and a pennant is won. It will certainly happen this season. Despite these projections, some team will win at least 87 games in both the West and Central. Your job Mr. Phelps, should you choose to accpet it, is to figure out who they are.
For example, the Red Sox acquired Victor Martinez at last year'd trade deadline, which by itself accounts for a significant chunk of the 25-run swing between their projected and actual runs scored. Then again, the Indians scored more runs than their projection despite trading Martinez away, FWIW.
Yet, if a team scores 30 more runs than PECOTA projected apparently the system must be broken.
Rangers (pitching?), As (a BP tradition to overrate this ball club), Indians (see As), Orioles are overrated.
One thing is clear, the projections are slightly optimisticly biased. They project 1,148 wins and only 1,120 losses and I can predict with virtual certainty that the final league record will be 1,134 and 1,134 (barring unplayed games or a significant change in the rules). I do wonder where that comes from.
Otherwise, yes, these rankings seem fine.
In fact, though I'm splitting the two leagues into separate lists this year, one idea I'm leaning towards is to publish - via the blog, or appended to the second list of the week (I haven't decided upon the schedule) - a combined ranking applying some kind of league adjustment based upon the AL's interleague dominance. It will probably be on the order of 30 points, such that a .500 AL team would rank at .530 in the combined list, while a .500 NL team would rank at .470.
But is their GM ever pounded for his decision? Nope. In fact, he has 2 executive of the year awards for his efforts. It's a joke.
Beyond that, note that I've been quite critical of Mark Shapiro here, here and numerous times elsewhere. You might also note that he was kicked upstairs over the winter for the highly regarded Chris Antonetti to take over. He's got his work cut out, particularly in improving the Indians' sorry draft record of recent years. But at least you won't have to worry about Shapiro getting any undue hardware.
The A's are NOT the 2008 Rays. The quantity of talent is just not there.
The Yankees' projection in runs scored shows a 60 run drop. OK, this is possible, taking into account the potential dropoffs at DH, LF, SS and C. But there'll be some improvement at 3rd (barring suspensions) and CF. OK, I can live with the falloff of 60 runs there, though I think it's more likely to be 30 or 40 runs.
The Yankees' projection is also based on the fact that Yankees' run prevention only improves by 4 runs. I assume this is because the system projects backsliding by Pettitte, Rivera, the bullpen, and Jeter's defensive improvement, because the change from Hughes/Wang/Mitre/Gaudin to Vazquez is worth at least 25 runs even if Vazquez' ERA is about 4.70, plus the defensive improvement from Melky/Damon/Gardner to Granderson/Gardner/Winn/Thames is worth about another 15-20 runs.
In my opinion, the Yanks' run prevention improves by at least 30 runs, and that's worth about 3 games more in the standings.
OK, it's a modest quibble, but that's my story and I'm sticking to it. For now, anyway.
And any improvement in 3B is less than certain. A-Rod only played 14 fewer games last year than 2008, but he had a year better rate-wise than his career averages. He is at the age where declines, albeit minor, are to expected. Overall being flat there seems likely.
You can't include dropping Hughes from the rotation and then count him in the rotation. And you would be hard pressed to find 25 runs from the other 3, as while Wang was awful, but he only pitched very few innings, Gaudin was actually significantly better than Vazquez's median AL history and Chamberlain was about the same as Vazquez will likely be.
Sure, any team can plausibly win, but the Royals are among the least likely to plausibly do so.