While looking toward the future with our comprehensive slate of current content, we'd also like to recognize our rich past by drawing upon our extensive (and mostly free) online archive of work dating back to 1997. In an effort to highlight the best of what's gone before, we'll be bringing you a weekly blast from BP's past, introducing or re-introducing you to some of the most informative and entertaining authors who have passed through our virtual halls. If you have fond recollections of a BP piece that you'd like to nominate for re-exposure to a wider audience, send us your suggestion.
It's been a decade since the first time PECOTA projected team performance at BP in the piece reprinted below, which was originally published on March 19, 2003.
The PECOTA system acknowledges that there is a wide range of variance intrinsic to any set of forecasts. What's more, there's no reason to expect that this variance will be unrelated to the team that a player toils for. On the contrary, there are myriad anecdotal examples of entire teams who routinely fall toward the top or the bottom of their forecast range. Under Dusty Baker, the Giants have consistently gotten more production than would reasonably be expected from a set of thirtysomethings. Under Leo Mazzone, the Braves have consistently turned waiver wire fodder into good or even great bullpen arms.
Indeed, it's possible to conjure up an argument like that for just about every team, and some of the time, you won't even be BSing. Translating player projections into team forecasts is an exercise that caters mostly to the left side of the brain; you're sure to see some more creative solutions in the coming days as we publish the BP author forecasts, crazed opium dreams like the Cubs taking the pennant. I have myself deviated from the HAL 9000 version in quite a number of cases. Nevertheless, we've never had anything quite like PECOTA before, and it's worth seeing what it has to say.
A quick outline of the methodology used here:
- The weighted mean forecasts for hitters, and the median projections for pitchers, were used in order to determine a player's rate of performance. Medians were used instead of means for pitchers because the mean forecasts tend to give disproportionate weight to disaster seasons in which a pitcher posts a very high ERA in a very small number of appearances, something which isn't quite as devastating to a team's fortunes as it appears in the subtotals.
- Playing time was forecast based on a combination of PECOTA's own estimates and information gathered from RotoWire, local newspapers, and consultation with other BP authors. I'll go to my grave committed to the proposition that a completely objective guess at playing time provides better information than a completely subjective one; it accounts for the risk of injury, the effects of aging, and the underlying gravity that tends to push better players into more frequent use. Nevertheless, there's no reason that a combination of the two approaches can't be more effective still, and that is what has been employed here.
I completed depth charts for each team, making sure the plate appearances and innings pitched estimates added up correctly at each position. I did not allow a player to substantially exceed his PECOTA playing time estimate if it looked like the forecast was capped based on injury risk or aging. I did increase his playing time by a reasonable amount if his role has changed (Danny Graves, Erubiel Durazo), or if the forecast was limited by predicted ineffectiveness, but the player's team had no obvious alternatives (Rey Ordonez, the Royals' bullpen). But generally, the forecasts were conservative about divvying out playing time, and I believe they do an adequate job of accounting for injury risk and rewarding team depth, things that are often overlooked in exercises like these.
- Team estimates for BA, OBP and SLG were translated into an estimate of runs scored by means of David Tate's Marginal Lineup Value formula, which is normally used in order to estimate the contribution of an individual player, but is just as effective in forecasting teamwide run scoring. An adjustment to the total was applied to account for forecasted stolen base attempts, successful and not.
- Forecasting runs allowed was more straightforward, as PECOTA predicts this for each pitcher directly.
- As a sanity check, I tallied up the team runs scored and runs allowed estimates to make sure that they were in accord with one another. It turned out that PECOTA did pretty well for itself: the aggregate figures differed by only 0.4%. I increased each team's runs allowed estimate by that amount to keep everything at par. Note that the AL and NL were not treated separately; PECOTA seems to think that the NL is a little bit stronger, and will gain some wins in interleague play. Since that's how things played out last year, and since there's been some emigration of talent to the senior circuit this winter, I'm not prepared to dispute it.
- Finally, the runs scored and runs allowed estimates were translated into won-lost record by means of the Pythagenport formula.
OK, that was about as quick as an Arvydas Sabonis crossover dribble. But never mind; it's time for the fun stuff.
Team W L BA OBP SLG RS ERA RA ------------------------------------------------------------------- Yankees 109 53 .272 .364 .478 986 3.91 688 Red Sox 104 58 .275 .351 .457 914 3.83 673 Blue Jays 80 82 .264 .337 .442 838 4.84 846 Orioles 69 93 .245 .311 .396 674 4.51 788 Devil Rays 57 105 .250 .314 .391 674 5.29 923
New York Yankees
Those of you that have grumbled that the PECOTA forecasts are too conservative should be suitably impressed by this set of results. Yeah, it's boring to predict that the division standings will sort out in the same order as they have for each of the past five seasons, but the stratification within the East is as great as it has ever been.
One-hundred and nine wins is the sort of gaudy number that will cause Bud Selig to wake up drenched in his own sweat. The Yankees get on base so darn well that they'll produce plenty of long rallies, and very possibly approach 1000 runs. Sure, our projection for Hideki Matsui is optimistic (.285 BA/.423 OBP/.572 SLG), but we know a thing or two about how to interpret Japanese stats, and he's filling a slot in the batting order that was a black hole last year (or more precisely, a Rondell White Hole). Moreover, the Yankees lineup was inefficient last year, scoring about 30 fewer runs than you'd expect from the MLV formula. Some of that was due to Joe Torre's stubborn insistence that Alfonso Soriano (.291/.325/.520) is a leadoff hitter–a mistake that he might well replicate this year–but even so the Yankees are likely to get more bang for their buck.
PECOTA expects the Yankees' pitching to be about where it was last year. There's plenty of depth to go around, and Mike Mussina (200 IP, 3.51 ERA) and Roger Clemens (170, 3.54) have the peripheral numbers to suggest that they'll be better than a year ago. Jose Contreras is the one player in the league that we don't have a reliable projection for; I've penciled him in at a 4.00 ERA in 120 innings, which looked conservative a month ago, but much less so after a shaky spring training. Truth is, the Yankees might not need him.
Boston Red Sox
The Red Sox's Pythagenport put them at 100 wins last year, and that was with nearly 1000 plate appearances of ineptitude between Tony Clark, Jose Offerman, and Rey Sanchez, all of whom have now been exiled. I've already drawn the ire of Red Sox Nation by suggesting that Jeremy Giambi (.253/.378/.448) isn't likely to maintain a performance anywhere near as good that of his brother. Manny Ramirez (.300/.408/.596) is expected to show some erosion in his batting average, which is often the first sign of age. Indeed, a lot of the forecasts for unmatched Sock hitters are pessimistic. But put 'em together, and the Red Sox are the only team apart from the Yankees with productive hitters in all nine slots in the batting order, as well as a deep bench and money in their pockets in case anything goes wrong. Their bats will be very good.
The pitching projects to be the strongest in the league by a toenail. Derek Lowe (170, 3.54) isn't likely to match his production from a year ago, buoyed as it was by an absurdly low hit rate on balls in play that won't be there with Todd Walker replacing Sanchez up the middle. But Pedro Martinez (200, 1.97) is the best pitcher in the league by an astounding margin, and there's adequate depth both at the bottom of the rotation and in the bullpen. If the league plays out as PECOTA thinks it will, Theo and company will have a fascinating decision come September: rest Pedro for the post-season with the wild card firmly in hand, or go for broke and gun for a division title that would end five long years of Yankee hegemony?
Toronto Blue Jays
What are the Blue Jays to do? They've got a nifty little offense, led by Carlos Delgado (.279/.410/.562), Eric Hinske (.269/.362/.497), and a bunch of players on the right side of 30. Although the acquisition of a couple of mid-priced free agents might have made them division favorites in the AL Central, it still might leave a pretty good team 15 games out of the running in its own division.
That said, there are some pessimistic projections in the bunch. Frankie Catalonotto (.285/.356/.413) isn't expected to get his power groove back, and the catching position doesn't have any obvious solution. Roy Halladay (190, 4.33) has about the nastiest projection for him that you're likely to see, the product of a home run rate that PECOTA expects to rise, as well as the last remains of his 2000 campaign that it really doesn't know what to do with. I'd be inclined to take the over on the wins total, but I expect that J.P. Ricciardi will have begun to unload the team's remaining veteran talent by the All-Star break, bringing in young talent for a future season in which the old guard looks a little more vulnerable.
In a perfect world, Jay Gibbons (.257/.326/.483) and Jack Cust (.224/.344/.426) give the Orioles some punch in the middle of the order, and the starting rotation is solidly average. That's still not going to be good for more than 75 wins, not when you're counting on retreads like Marty Cordova (.241/.305/.402) and Jeff Conine (.265/.320/.400) on the right side of the defensive spectrum. Gibbons and Cust excepted, the Orioles have very few players on either side of the diamond whom it would be fair to label breakout candidates, and there's not much help from the farm system on the way.
Tampa Bay Devil Rays
Lou Piniella has not had his last screaming fit. PECOTA doesn't think that Rocco Baldelli (.234/.278/.373) is anywhere near ready for prime time, nor does it care for Toby Hall (.260/.303/.398), who has been cited as a breakout candidate in other circles. Shoring up the infield defense to boost the confidence of a young pitching staff might be a good approach, but except for Joe Kennedy (190, 4.48), what pitchers do the D-Rays have who will benefit more from Rey Ordonez' (.238/.285/.305) defense than they'll be harmed by being within spitting distance of Mt. Piniella? The silver lining is that the AL will have first dibs in the 2004 amateur draft; even then, the prospect of indentured servitude in the black, green and purple might leave likely first pick Stephen Drew thinking long and hard about pulling a stunt like his brother did.
Team W L BA OBP SLG RS ERA RA ------------------------------------------------------------------- White Sox 86 76 .265 .335 .452 850 4.56 797 Twins 84 78 .268 .330 .429 799 4.39 768 Indians 67 95 .252 .325 .407 732 5.04 879 Royals 64 98 .262 .327 .416 762 5.47 951 Tigers 55 107 .244 .313 .381 647 5.21 913
Chicago White Sox
The White Sox finished 14 games behind the Twins last season, but their Pythagenport records were virtually identical, and odds are the teams should converge this season. PECOTA expects that the effective replacement of Royce Clayton with Joe Crede (.268/.321/.425)–Jose Valentin (.251/.323/.467) will migrate to shortstop–will be just enough to offset the loss of three-fifths a year of Ray Durham. Similarly, the addition of Bartolo Colon (200, 3.74)–beer gut, declining strikeout rate and all–should help to counteract an almost inevitable decline in performance by Mark Buehrle (200, 4.20), who it thinks is too much of a finesse pitcher at too young an age.
The Sox offense should be good for 200+ home runs again, setting off plenty of fireworks over the Dan Ryan Expressway. If there's a guy that pushes the offense from good to great, it's more likely to be Carlos Lee (.278/.353/.485), he of the great second half, than Frank Thomas (.255/.371/.481). The upside of the pitching staff is limited, as the bottom three slots in the rotation will be filled by youngish hurlers who look more like adequate organizational players than top prospects; but in the weakest division in baseball, that might be enough.
Even if Torii Hunter (.282/.329/.509) and Jacque Jones (.273/.323/.455) give back some of what they've gained, the Twins' offense stands a good chance to improve. There's a bevy of young talent at the corner positions, and while it seems like they've been around nearly as long as long as Trammell and Whitaker, Cristian Guzman (.274/.307/.405) and Luis Rivas (.270/.321/.387) are young enough that they should improve a little bit, and just maybe a lot.
It's the pitching staff that is the greater concern. All of the pitchers in the starting rotation carry some baggage with them–Rick Reed's (150, 4.65) age, Brad Radke's (140, 4.15) hammy, and so on. Although he didn't pitch up to his ERA last year, the acquisition of Kenny Rogers (150, 4.66) makes for a good hedge against those risks, even if it did cause Scoresheet owners everywhere to run screaming over Johan Santana's (130, 3.78) temporary removal from the rotation. The bullpen is good, but not likely to receive the same inspired 2002 performances from the likes of J.C. Romero (70, 3.92) and Tony Fiore (70, 4.36). Add it all up and PECOTA expects the Twins to give up 50 or 60 more runs this year, and the division title in the process.
Squint hard enough, and you begin to see something. By the middle of June, Brandon Phillips (.251/.309/.408) should be entrenched at second base and Victor Martinez (.259/.342/.421) behind the plate. Travis Hafner (.269/.370/.482) won't replace Jim Thome's production, but he's a safe bet for better-than-average production. A last hurrah for Ellis Burks (.280/.356/.519)? Another run as Pedro Cerrano by Karim Garcia (.255/.310/.472)? While Indians fans might have lost their faith, at least there's hope, which you can't say in Detroit or Kansas City.
Still, one of the advantages of an exercise like this is that it provides a check against unrealistic expectations, and the Indians have more than their share of problems. PECOTA is pretty confident that Milton Bradley (.248/.320/.378) doesn't have the power to be more than a fringe player, and the pitching staff behind C.C. Sabathia (190, 4.38) is patched together with strings and duct tape. There's a lot to like here, but it would be easy to expect too much too soon from a smart organization, as a lot of us did from the Padres last year.
Kansas City Royals
Take away Paul Byrd from a pitching staff that gave up nearly 900 runs last year, and what do you get? Well, Carlos Beltran (.284/.358/.505) traded long before the deadline, and Mike Sweeney (.319/.396/.535) thinking hard about exercising his out clause. Part of the high team ERA is the result of park effects; Kauffman increased run scoring by an astounding 30% last season. But most of the innings on the pitching staff will be eaten up by pitchers who would be better off somewhere else in Middle America–like Omaha, or Wichita, or taking in a Yakov Smirnoff show in Branson.
The Tigers deserve more credit than we've given them. They picked up three outstanding prospects in the Jeff Weaver deal, some live arms in the Mark Redman trade, and their drafts have received favorable reviews the past two times around. But it's not clear that they've hit bottom just yet, and the Tigers' lineup will feature everyday players like Dean Palmer (.187/.285/.331) and Damion Easley (.234/.312/.350) wouldn't even be on the roster if not for their misguided contracts. Even if they exceed their forecasts, Carlos Pena (.248/.341/.446) and Eric Munson (.236/.327/.417) won't be able to carry dead weight like that.
Nor is the Tiger pitching staff likely to provide much relief. These were the bottom five pitching staffs in team strikeouts last year:
Team SO ------------------- Detroit 794 Kansas City 909 Colorado 920 Pittsburgh 920 Tampa Bay 925
It wasn't even close. The Tigers struck out nearly 100 fewer batters than the next team from the bottom, which doesn't bode well for soft-tossers like Mike Maroth (150, 5.30) and Andy Van Hekken (150, 5.48). Like a 15th seed from the Sunbelt Conference, the Tigers' biggest advantage is that they have nothing much to lose, whether it means trading Maroth or Van Hekken if and when he gets hot, giving Munson his best shot at third base, or staying patient with Franklyn German (50, 4.96) as his command flutters in and out.
Team W L BA OBP SLG RS ERA RA ------------------------------------------------------------------- Mariners 90 72 .265 .341 .410 788 4.02 705 Athletics 89 73 .258 .333 .426 788 4.05 710 Rangers 83 79 .264 .335 .456 858 4.78 833 Angels 82 80 .266 .328 .421 781 4.41 772
A lot of people are ready to write the Mariners off; of the 30 or so people who filled out prediction forms at our Chicago Pizza Feed last weekend, just a couple picked the Mariners to win the division, while a couple more expected them to sneak in as the wild card. Many of their vulnerabilities were exposed last season: Freddy Garcia (190, 3.81), worn down by heavy usage, pitched horribly in the second half, Edgar Martinez (.274/.394/.483) suffered through an injury-riddled campaign that looks like the beginning of the end, and Jeff Cirillo (.255/.306/.348) did everything he could to counteract Andres Galarraga's example that road stats for Rockies shouldn't count. None of those problems are likely to correct themselves this season.
But the Mariners are still a good team. I'd rather have seen them add a power hitter, but the addition of Randy Winn (.273/.341/.412) should give them one of the best outfield defenses in recent history. Bret Boone (.270/.329/.460) wasn't quite as good as in 2001, but he's no fluke, and PECOTA thinks he's Jeff Kent Lite. Jamie Moyer (200, 3.48) takes advantage of his context as well as any pitcher possibly could, and many of the comparables on his list aged with surprising grace. Arthur Rhodes (60, 2.21) might be the best reliever in the league. Greg Colbrunn (.280/.344/.471) might be the most overlooked acquisition of the winter, and looks like a smart hedge against a decline in performance by Edgar. Because the starting lineup is old, there aren't many breakout candidates among the bunch, but if the A's and Angels both fall back to earth, slow and steady could win the race.
The A's have spawned as many copycats as the West Coast Offense, but they still do some things better than any other organization, whether it's creating a supportive environment for innovators like Rick Peterson to work in, or filling their 40-man roster with overlooked players like Michael Neu (40, 4.63) and Jason Grabowski (.252/.343/.422) who will be ready to step in and contribute if the headline names falter. They'll beat that projection. Maybe.
Still, there are several reasons to think they'll finish closer to 90 wins than 100. The A's finished seven games ahead of their Pythagenport record last season. Miguel Tejada (.281/.335/.487) isn't likely to have quite as good a season, especially if he's older than his listed age. Billy Beane is counting on Chris Singleton (.262/.298/.396) to improve the outfield flycatching–something that PECOTA's simple team defense adjustment hasn't accounted for–but he's going to create a drag at the bottom of the batting order.
Most important of all are the Three Aces, each of whom is forecast for a drop in performance this season. Barry Zito (200, 3.67) is the best of the bunch, but he was hit-lucky last year. PECOTA expects Mark Mulder (190, 4.21) to be more vulnerable to the long ball. Tim Hudson's (210, 3.87) strikeout rate has declined in each of the last two seasons, and was no better than league average last year. They're fine pitchers, all of them, but counting on all three to pitch as well as they have been is risky business, even before getting into the question of injuries.
Even without Ivan Rodriguez, the Rangers are going to score some runs. Apart from The Obvious (.302/.407/.621), PECOTA regards the Rangers' abundance of options at the corner positions as a key strength. Kevin Mench (.261/.332/.485) isn't having the breakout season that we're forecasting for him? No problem; maybe Ryan Ludwick (.256/.326/.470) is. Hank Blalock (.267/.335/.430) still going through his prospect puberty? Mark Teixeira (.265/.338/.465) will be just up I-35.
The arguably neurotic Buck Schowalter will have plenty of scabs to pick at. His biggest decisions revolve around offense vs. defense up the middle. I'm inclined to agree with Joe Sheehan that a good center fielder is a must in Arlington; with the exceptions of Ugueth Urbina (60, 2.91) and possibly Chan Ho Park (160, 4.68), the Rangers don't have many pitchers who will thrive without it. Replacing Mike Young (.251/.305/.376) with Blalock at second, on the other hand, might be worth a couple of wins, as could riding Todd Greene (.249/.286/.448) for all that he's worth.
We haven't written very much about the Angels this spring because they haven't done very much; it's conceivable they'll go into Opening Day with as many as 23 players from last year's World Series roster. As such, I will bear the burden of being the first among us to suggest that all those stathead cliches that were disproved so spectacularly last October ought still to be taken into account. You know, stuff like:
- Batting average is less consistent from season to season than other components of offensive performance, and the Angels were highly dependant on their high batting average last season.
- Few teams in recent history have been as efficient as the Angels were last season in turning balls in play into outs. No team since the Mark Belanger Orioles have sustained such an elite performance for more than one season.
- The Angels hit exceptionally well last season with runners in scoring position. Clutch hitting might exist, but clutch hitters don't, and clutch hitting teams sure as hell don't.
Each of those factors are taken into account in one way or another in the projections; both the offense and the pitching are expected to regress, and PECOTA thinks the Angels will play .500 ball.
I don't mean to dismiss the Angels' success from a year ago; in a league that has become increasingly dependant on working pitchers deep into the count, the Angels may gain a profound comparative advantage from swinging early. In a league that has become increasingly dependent on acquiring expensive free agents, the Angels have won with their own talent and cost-effective acquisitions. Their approach led to a championship last season, something that will never be taken away from them. It wouldn't surprise me if they continue to defy expectations. But the weight of statistical evidence is against them.