In a September that appears slated for a dearth of late-season drama, the American League West contains not only the last vestige of a real post-season race-the only one where the underdog has even a 10 percent shot according to our Playoff Odds-but also a bit of potential history, at least from a sabermetric standpoint. Once again, it’s the Angels, those anti-sabermetric darlings, making that history.
According to our Adjusted Standings page, through Sunday the Angels were 11.5 games above their third-order Pythagenpat projection, a fancy way of saying that they’ve won over 11 games more than the combination of events on the field-their hits, walks, total bases, stolen bases, and outs of all kinds, as well as those of their opponents, all adjusted for park, league, quality of competition and temperature of porridge-would suggest. That’s by far the top mark in the majors this year, and while it’s not enough to break the single-season record of 16.0, set by last year’s Anaheim team, it does crack the all-time top 10, and place them in select company:
Rank Year Team W-L Pct R RA AEQR AEQRA D3 1 2008 Angels 100-62 .617 765 697 754 725 16.0 2 2004 Yankees 101-61 .623 897 808 911 831 12.7 3 1970 Reds 102-60 .630 775 681 757 676 12.6 4 2007 D'backs 90-72 .556 712 732 708 739 12.2 5t 1954 Dodgers 92-62 .597 778 740 782 749 12.1 5t 2005 White Sox 99-63 .611 741 645 740 684 12.1 7 1905 Tigers 79-74 .516 512 604 524 601 11.9 8t 1924 Dodgers 92-62 .597 717 679 717 684 11.7 8t 2002 Twins 94-67 .584 768 712 759 741 11.7 10 2009 Angels 86-56 .606 786 679 777 739 11.5 11 1954 Indians 111-43 .721 746 504 717 511 11.4 12t 1907 Cubs 108-44 .711 574 390 552 394 11.2 12t 1961 Reds 93-61 .604 710 653 705 658 11.2 14t 1972 Mets 83-73 .532 528 578 533 583 11.0 14t 1931 A's 107-45 .704 858 626 841 639 11.0 16 1984 Mets 90-72 .556 652 676 657 671 10.7 17t 1977 Orioles 97-64 .602 719 653 719 662 10.2 17t 2008 Astros 86-75 .534 712 743 683 727 10.2 17t 2006 A's 93-69 .574 771 727 791 772 10.2 17t 1936 Cardinals 87-67 .565 795 794 808 809 10.2 21t 2007 Mariners 88-74 .543 794 813 792 824 10.0 21t 1997 Giants 90-72 .556 784 793 780 789 10.0
Projected across a 162-game schedule, the Angels’ current performance is the equivalent of outdoing their third-order projection by 13.1 games, which would rank second on this list. However, it’s a misnomer to say they’re actually “on pace” for such a finish, since teams that are outperforming their Pythagorean records by wide margins in either direction tend to regress to the mean. Case in point, they lost on Monday night to reduce their D3 (the difference between their third-order wins and actual wins) to 10.9.
Still, making the list is remarkable enough; from among a field of over 2,200 team-seasons dating back to 1901, just one percent of them have turned in a season at least 10 wins above expectation. What’s even more remarkable is that this marks the second year in a row that the Angels have exceeded expectations by at least 10 games, and the third year in a row they’ve done so by at least eight games, both of which are firsts. Only five teams have even managed the latter feat in back-to-back years:
Year Team W-L Pct R RA AEQR AEQRA D3 1908 Pirates 98-56 .636 585 468 583 483 8.5 1909 Pirates 111-42 .725 701 448 695 468 8.7 1930 A's 102-52 .662 951 751 947 772 9.3 1931 A's 107-45 .704 858 626 841 639 11.0 1960 Yankees 97-57 .630 746 627 741 641 9.6 1961 Yankees 109-53 .673 827 612 816 623 8.2 2002 Twins 94-67 .584 768 712 759 741 11.7 2003 Twins 90-72 .556 801 758 783 777 8.5 2007 Angels 94-68 .580 822 731 799 751 8.1 2008 Angels 100-62 .617 765 697 754 725 16.0
And to get ahead of ourselves a year, only two other teams have been as many as six games above projection for three years in a row:
Year Team W-L Pct R RA AEQR AEQRA D3 1961 Dodgers 89-65 .578 735 697 737 707 9.0 1962 Dodgers 102-63 .618 842 697 848 715 6.3 1963 Dodgers 99-63 .611 640 550 633 554 8.5 2002 Twins 94-67 .584 768 712 759 741 11.7 2003 Twins 90-72 .556 801 758 783 777 8.5 2004 Twins 92-70 .568 780 715 761 724 7.2
Quite simply, we’re in Almost Neverland.
We often talk of teams that over- or underperform their projected records as “lucky” or “unlucky,” but it’s a misnomer to chalk up the entirety of such discrepancies to luck. They generally stem from an irregular distribution of runs, so “randomness” may be a better term. Overachieving teams tend to win most of the close games but get blown out a few times. The 16 teams who exceeded their third-order projections by 10 wins or more while playing in the Retrosheet era (1954 onward)-call them the “Plus Tens”-went a combined 469-281 (.625) in one-run games and 324-170 (.656) in two-run games, but in games decided by six or more runs, they were just 223-259 (.462), a mark that includes the 1954 Indians’ 20-5 record in such blowouts. All told, those teams went 1016-710 (.586), right in line with their overall .589 winning percentage, while outscoring their opponents by 125 runs in such extreme games. That’s the equivalent of a 95-67 team outscoring their opponents by just 11 runs over the course of a season, about nine percent of the run differential such teams have historically posted.
These Angels are fairly typical of the trend, as they’re a big-league best 26-14 in one-run games, but just 1-5 in games decided by at least 10 runs. Their run differential in those one-run games is +12, while in the blowouts, it’s -46. At the two extremes, that flattens to a 27-19 record for a team that’s nevertheless been outscored by 34 runs, the kind of thing which can really distort a team’s projected record. Similarly, last year’s Angels were 31-21 in one-run games and an even more astounding 30-7 in two-run games (whereas this year’s Halos are just 6-17 after Monday night’s loss to the Yankees); they were also 1-5 in games decided by more than eight runs.
A major factor in outperforming one’s projected record is having relatively more success in higher-leverage situations, such as hitting well with runners in scoring position, or being especially stingy in late-game relief. As I noted last year, a strong bullpen is a consistent means of such overachievement; the correlation between a team’s cumulative WXRL and its D3 is .42, whereas it’s just .20 for SNLVAR. Of the 16 Retrosheet-era teams above, the 1977 Orioles were the only ones who failed to finish in their league’s top three in WXRL.
Which brings us to another mark of distinction for the 2009 Angels, as they rank just sixth in the league in WXRL, some 2.3 wins out of the top three. Strong bullpens have been the hallmark of the Halos under Mike Scioscia; they’ve never finished in the bottom half of the league on his watch, thanks in large part to stalwarts like Francisco Rodriguez and Scot Shields, who were key components of the team’s relief corps from 2002 through 2008:
Year WXRL AL Rank
2000 9.9 5
2001 7.2 7
2002 12.6 2
2003 10.7 3
2004 13.0 4
2005 13.4 1
2006 14.2 2
2007 10.5 6
2008 13.3 3
2009 8.0 6
This year’s bullpen, of course, lacks K-Rod, who departed for the Mets via free agency after breaking the single-season saves record last year. It didn’t get much help from Shields (6.62 ERA, -0.6 WXRL) before he underwent season-ending knee surgery in mid-June, nor has it gotten much from Jose Arredondo (5.98 ERA, 0.3 WXRL), who joined Rodriguez and Shields among the league’s top 20 in WXRL last year as a rookie. Still, the pen has overcome a gruesome first month to reach the middle of the pack. Since I first peeked into their house of horrors back in May, they’ve actually put up the league’s second-best WXRL (6.6, behind only Cleveland’s 7.0) and Reliever Fair Run Average (4.17, trailing only Texas’ 4.13).
Which isn’t to say it’s been all roses. Replacement closer Brian Fuentes actually leads the league in saves with 41, but that ranking belies his 4.10 ERA and his 2.4 WXRL, which ranks only 19th in the league, 10th among the 15 pitchers with at least 10 saves, and second on the staff behind Darren Oliver‘s 2.6. Fuentes’ mediocre performance has even created some controversy in the Happiest Place on Earth lately. On Sunday, Scioscia began a ninth-inning save situation with righty Kevin Jepsen on the mound, playing matchups to account for the fact that his lefty closer has had a difficult time getting righty hitters out now that he’s in the AL instead of mile-high in the senior circuit; they’re batting .269/.364/.462 against him this year, compared to .212/.300/.335 from 2006-2008.
So no, these aren’t the Angels of old, as evidenced by the fact that their 4.57 ERA not only ranks an unappealing 10th in the league, which is the team’s highest mark since 2000, and by more than a quarter of a run to boot. The offense has made up for it by scoring 5.5 runs per game, second in the league and a dead heat for the franchise record, while providing a fair bit of clutch hitting. Their .299/.377/.465 performance with runners in scoring position is the league’s best by about 25 points of OPS, and they’ve also got the top marks with men on base, with men on third and two outs, and so on.
So, clutch pitching and clutch hitting have both turned up again for the Angels, and while it hasn’t been in their usual proportions, it’s still been enough to push them past the Rangers, who actually lead them in the third-order standings by 2.3 games. Barring a late collapse, it will be pushing the Halos into the history books as well.
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
How well does defense correlate with third order wins and WXRL?
As for the question about defense, last year when I wrote up one of the overachiever/underachiever articles, I think I ran a correlation of DE and/or PADE against D3 and/or WXRL, but I'll be damned if I can find it. Might need to do it again when I get a chance.
PADE correlates much better with WXRL than regular DE does, but even that's only about .27.
I also don't like the luck argument... sometimes luck is a factor, but there's a saying I have regarding luck/probability games. Luck is always a factor but good teams/players are better able to capitalize on luck for their favor or for the disadvantage of their opponent.
And if by implementing such an adjustment, would it result in actual W-L records correlating more strongly with expected W-L records?
Someone somewhere has to have taken a stab at this.
Alas, for reasons unclear, it currently resides in the dusty basement of Baseball Prospectus, right next to the big Xtapolapocetl statue. I'm investigating the possibility of dragging it back upstairs.
This might be a hallmark article here at BP. I'm just not sure many of your fellow writers are on board.
The main reason to pick the A's for the division this year was the unaviodable regression to the mean by the Angels. Given the Angels ability to repeat their Pythagorean overperformance (and the fact the A's are just terrible)make that prediction look ridiculous.
Most of the sabermetric community have laughed at some of the elements (clutch hitting, importance of closer, small ball, productive outs, managerial input, team chemistry) that have been associated with the Angels teams over the past few years.
The sabermatricians are the ones looking for answers in light of a second year of Pythagorean overperfomance that can't be passsed off as mere randomness.
Kudos to Jay for putting the subject on the table for discussion.
Clutch performances exist, and it's certainly not unheard of for individual hitters to string together multiple clutch seasons in a row. But predicting *which* hitters will do so — in other words, identifying it as a persistent skill from year to year — so is another matter entirely. Take these Angels, for example: while they've led the league in hitting with RISP, the correlation between the individual players' OPSes in such situations between this year and last year is just 0.08 (20 PA minimum in both samples) - pretty close to none. Up that to 50 PA and it's actually strongly negative (-0.74). In other words, identifying which Angels own some persistent clutch skill with RISP is pretty much impossible given that dearth of information. Now, there's certainly more to clutch performance than just that one split, but it's just an example. The scope of team-level clutch hitting performance is one that, to my knowledge at least, has yet to be investigated in depth.
The take-home lesson about the Angels in the grand scheme of things is that they've had a tremendous amount of success over the years with an organizational philosophy that places a greater emphasis on things like team speed, baserunning, and contact skills — areas often undervalued in the stathead realm, and less easily quantified than the more basic building blocks of offense — than most teams. If one understands that the true meaning of Moneyball isn't "OBP Uber Alles and Death to Scouting" but that a team should attempt to exploit the inefficiencies in the marketplace, then the Angels are a model team in that respect, and they certainly deserve to be hailed as such.
We've known since the peak years of Darryl Strawberry that an unusual mix of situational matchups can look like clutch (or choke) behavior. Is it possible that Mike Scioscia's great anti-sabermetric skill is really the most sabermetric skill of all: Strat-o-matic matchup management? Statheads have been saying since the days of Earl Weaver and Sparky Anderson that good managers put players in situations where they are more likely to succeed, and keep them out of the ones where they are more likely to fail.
I'd be curious to see whether (say) the Angels have faced same-handed pitching (or hitting) less often as a team, over the past few years, than have other teams. Or whether they have smaller platoon splits on average. Etc.
1. There's a negative correlation between relievers used and actual winning percentage (-.39) and a similar one for Pythag (-.36)
2. There's a negative correlation between positional sac hits (-.15) and winning pct or Pythag.
3. There's virtually no correlation between blown quality starts and actual or projected record (.05)
4. There's a surprisingly solid correlation (0.5) between runners moving with the pitch and actual winning percentage. It's slightly less (.43) with projected winning percentage, and at least somewhat visible (.20) between runners moving and pythagorean differential, though the latter nearly disappears if you remove the Angels from the set.
Turning to the 2009 Bill James Handbook, which I have within easy reach, and looking over Scioscia's record, he's never lead the league in lineup platoon advantage, but his teams have tended to be above average in that regard, with 63% having the advantage at the start of the game over the course of his career. His teams have never led the league in sac hits or reliever usage, either.
Where he's consistently led his peers is in two categories that mesh with what I was discussing above: stolen base attempts and runners moving with the pitch. Given that and #4 above, I'd say it's worth earmarking the topic of baserunning success and its link with overperformance for further study.
But it does make sense that in close games, runners are moving more often than in a blowout, for example... and teams that are successful at moving runners over repeatdly, without giving up outs (via a hit), would win more games than expected.
I realize this has rambled around a bit. Do you think you will end up doing an article that revisits and reorganizes all this, particularly diving more into the "runners moving" angle?
I'll almost certainly revisit this at some point in the future. Perhaps not until I've done the 2009 underachievers column but this is clearly an area fertile for further research.