March 25, 2008
Doctoring The Numbers
Royals and Phillies
One of my favorite old-time columns from this humble website is this one, written by Michael Wolverton back in 2004, which completely opened my eyes to not just the absurdity of the unearned run, but how evaluating pitchers by ERA instead of by RA systematically underrated good pitchers. That's because, with few exceptions, pitchers who excel at preventing earned runs also excel at preventing unearned ones. (For more on the subject, here's Michael's follow-up piece.)
The biggest of those exceptions is the knuckleball pitcher, who by his very nature will allow large numbers of passed balls (or rather, will force his catcher to allow them). As passed balls count as errors, they lead to unearned runs. By contrast, if you were looking for a pitcher who did not surrender a lot of unearned runs, you'd want to find one that:
To illustrate the point, let's compare two pitchers. Hoyt Wilhelm has the lowest career ERA of any pitcher in the live-ball era with 1000 or more innings; his 2.52 ERA is 23 points lower than the next-best pitcher, Whitey Ford, at 2.75. (Pedro Martinez had a career 2.58 ERA as recently as four years ago; it's at 2.80 now.) By contrast, Curt Schilling has a 3.46 ERA for his career, an excellent figure in today's game, but a full 37.3 percent higher than Wilhelm's career mark. But factor in unearned runs, and the comparison changes dramatically:
Pitcher IP ER ERA UER Pct. RA Hoyt Wilhelm 2254 632 2.52 141 18.2% 3.09 Curt Schilling 3261 1253 3.46 65 4.9% 3.64Pct. refers to the percentage of overall runs allowed which were unearned.
Wilhelm has a 94-point advantage in ERA, but judge them by runs allowed, and that margin is cut by over 40 percent, to 55 points. No question, Wilhelm has a higher ratio of unearned runs in part, because there were more unearned runs in general in the 1950s and 1960s than in today's era. But among pitchers who debuted since 1920, Wilhelm ranks 12th overall in his portion of runs allowed that were unearned. Every other pitcher in the top 40 had retired by 1953, Wilhelm's sophomore season. The amazing movement on his knuckleball, which made him so effective in the first place, also was his undoing, far more often than the average hurler.
Schilling, on the other hand, holds the major league record for the lowest career UERA--unearned runs per nine innings. Here's a list of the top five, with a minimum of 600 runs allowed:
Pitcher IP ER ERA UER UERA RA Curt Schilling 3261 1253 3.46 65 0.179 3.64 Mike Mussina 3362 1383 3.70 91 0.244 3.95 Jim Deshaies 1525 702 4.14 41 0.242 4.38 Sid Fernandez 1867 696 3.36 53 0.256 3.61 Woody Williams 2216 1031 4.19 65 0.264 4.45
These five pitchers share the same characteristics we talked about earlier: they all had good strikeout rates, they all had good-to-outstanding control (OK, maybe not El Sid), and they were all fly-ball pitchers--with the exception of Mussina, they were extreme fly-ball pitchers, and while this led to a lot of home runs, it also led to relatively few grounders in the hole that the shortstop could bobble or throw into right field. The matter of right-handed vs. left-handed pitcher appears to be a minor consideration; Fernandez and Deshaies were both lefties.
(Incidentally, if you lower the threshold to 500 runs allowed, an almost incomprehensible name appears on the list: Todd Van Poppel. I have no explanation for how a pitcher with a 5.58 ERA could have a UERA of 0.24. Only 4.3 percent of Van Poppel's runs allowed in his career were unearned, the lowest total ever among pitchers with as many runs allowed as he has.)
When the Hall of Fame case for Curt Schilling gets made…this argument won't be listed, because it won't need to be. Schilling's 216-146 lifetime record, 127 ERA+, and his well-deserved reputation as one of the greatest post-season starters ever should make him a lock. But when Mike Mussina has his case made, it will definitely be worth pointing out that his ERA underrates him because he gave up so few unearned runs. The American League last year had a UERA of 0.42. Mussina's ability (and it is an ability, in the sense that it's not random) to limit unearned runs is the equivalent of docking his ERA by 18 points, from 3.70 to 3.52. That's the difference between a pitcher in the Jack Morris class (and not quite good enough) to a pitcher who deserves serious consideration for admission.
And all this ties to the Royals…how? Because the Royals have a young pitcher very much cut from the same cloth as Schilling and Mussina--a fly-ball pitcher with a terrific strikeout-to-walk ratio, better than three to one. And for all the discussion about Zack Greinke the person, it's easy to lose sight of Zack Greinke the pitcher--a pitcher who has, to this point in his career, been stingier with the unearned run than any pitcher before him.
In his rookie season of 2004, Greinke had a respectable 3.97 ERA--but went the entire year without surrendering an unearned run. That's not unheard of (Schilling was a perfect 90-for-90 in 2006), but it was an auspicious start. In 2005, as his career crashed around him, Greinke still only surrendered seven unearned runs. Last year, he allowed just two.
For his career, Greinke has a 4.63 ERA in 456 innings. That's not good, but just nine of the 244 runs he's allowed, or 3.7 percent, have been unearned. His UERA is 0.178, which is even lower than Schilling's. Moreover, he has exactly the kind of skill set to keep this going--he has terrific control, a good strikeout rate, he is a significant fly-ball pitcher, and he's a terrific defensive player (he's made just one error in his entire pro career). Somehow, Greinke has kept unearned runs off the board better than almost any pitcher in history, even though in his three full seasons in the majors, the Royals have ranked second, first, and fourth in the AL in errors.
Undoubtedly, when evaluating the future of the Royals or even the future of Zack Greinke, Greinke's ability to limit unearned runs ranks near the bottom of the list. It's worth mentioning that, for all the drama and uncertainty that have been the hallmarks of his career, he has done a significantly better job of keeping runs off the board than his traditional statistics would suggest.
As if being a Royals fan hasn't been hard enough the last dozen years or so, last year the team even had its name stricken from the record book. The only significant single-season major league record held by the team was broken last year, to remarkably little fanfare.
OK, so that record was for at-bats in a season. It was still a record, and with 705 AB in 1980, Willie Wilson had held that record for 27 years, fending off the likes of Juan Samuel (701 in 1984), Ichiro Suzuki (704 in 2004), Jose Reyes (696 in 2005), and Juan Pierre (699 in 2006). But a record that had come under assault for years was finally breached by Jimmy Rollins, who bested Wilson's mark with two games to spare and finished with 716 at-bats. The 11 at-bat gap between first and second place now is greater than the difference between second place and 11th.
Along with the at-bats, Rollins set a number of related records. Unlike the other names above, Rollins didn't earn his high at-bat total by eschewing walks at every opportunity; he had 49 of them, along with seven HBPs and six sacrifice flies. He therefore wrested an even more meaningful record, the one for plate appearances in a season, away from another Phillie:
Year Player PA 2007 Jimmy Rollins 778 1993 Lenny Dykstra 773 1974 Pete Rose 770 1975 Dave Cash 766 2007 Jose Reyes 765
While not to discount Rollins' performance last year, it's undeniable that his raw totals were heavily inflated by being the beneficiary of so much playing time. As has been heavily reported, Rollins and Curtis Granderson became only the third and fourth players ever to go 20-20-20-20 in doubles, triples, homers, and stolen bases. (Frank "Wildfire" Schulte in 1911 and Willie Mays in 1957 were the first two.) Rollins is in his own club as the only player in major league history with 30 doubles, 30 homers, 20 triples, and 20 steals in a season. Only one other shortstop has finished with more extra-base hits in a season than Rollins:
Year Player XBH 1996 Alex Rodriguez 91 2007 Jimmy Rollins 88 1982 Robin Yount 87 2001 Alex Rodriguez 87 2002 Alex Rodriguez 86
(Just to remind you: Rodriguez, who was 20 years old most of that season, also won the batting title with a .358 average. That same year, Juan Gonzalez won the MVP award, just in case you were wondering why some of us consider the BBWAA awards a joke.)
Rollins also had 212 hits last year, easily the most for any player who failed to hit .300 during the season:
Year Player H Avg. 2007 Jimmy Rollins 212 .296 1962 Maury Wills 208 .299 1974 Dave Cash 206 .2999 1967 Lou Brock 206 .299 2006 Juan Pierre 204 .292
If you project Rollins' numbers to a 700 PA season, still a large total (only 19 players reached that mark last season), you wind up with 27 homers, 18 triples, 35 doubles, and 191 hits--excellent numbers for a shortstop, but not historic ones, and not ones worthy of an MVP award.
That doesn't mean that Rollins owes his performance last year to luck and circumstance, because of course his ability to stay in the lineup every day is part of the reason he had so many plate appearances. But he owes the record to a perfect storm of factors. He was exceptionally durable, starting every game for the Phillies, and playing shortstop for all but 17 innings. He generally batted leadoff, giving him first crack at an extra at-bat as the lineup turned itself over. (Surprisingly, he only started 139 games in the leadoff spot, spending the other 23 games in the #3 hole.) Finally, he played for a team with an offense that led the NL in runs, by a wide margin--the better the offense, the more opportunities the batters in that lineup will have to bat.
What's interesting here is that while the Phillies had a terrific offense, it wasn't historically great--they ranked 76 runs behind the Yankees, in large part because they operate without the benefit of the DH. You would think that the list of players with the most plate appearances in a season would be on teams that operated with the DH, yet somehow the top five players listed above all played in the NL. The only player in the top 10 for single-season plate appearances who played in the AL was Ichiro Suzuki, in seventh place with 762 PA in 2004. I have no idea why this is.
The Phillies' offense looks even more pedestrian when you consider that the most important factor in determining a team's overall plate appearances is not runs scored, but OBP. By definition, a team has three outs an inning, 27 outs a game, and 4374 outs a season to work with, and since OBP encompasses almost every plate appearance that does not lead to an out, a team's overall plate appearances should correlate with its OBP almost perfectly. Approximately speaking, a team's plate appearances in a season should equal 4374/(1-OBP). (In reality, this formula overestimates a team's plate appearances, as we will see later.)
This rule, that the higher a team's OBP, the more plate appearances it will have, works pretty well. The 1931 Yankees, with a team OBP of .380 and the all-time record for runs scored with 1067, also had 6473 plate appearances, just 81 shy of the all-time record set by the 1999 Indians. That's pretty remarkable, given that the Yankees only played 155 games--over 162 games that works out to 6765 PA. Lou Gehrig had 738 PA for that Yankees team, just 40 shy of the record, and he presumably spent most of the year in the cleanup spot.
Here's the list of the top team plate appearance totals of all time:
Year Team PA OBP Outs 1999 Cleveland 6554 .373 4106 1996 Boston 6545 .359 4193 1976 Cincinnati 6538 .357 4206 2007 Philadelphia 6537 .354 4225 2003 Boston 6530 .360 4181
In this table, Outs refers to the number of outs we would calculate each team to make if that formula, PA = Outs/(1-OBP), held true.
The first thing to stand out is that plate appearances do not correlate perfectly with team OBP. The Phillies had just the 54th-highest team OBP of the 162-game era, but ranked fourth all-time in plate appearances. The 1999 Indians, despite an OBP a full 19 points higher than the Phillies had last year, only managed 17 more plate appearances. This means that their batters were responsible for about 119 fewer outs than the Phillies' hitters were, in the same number of games. Where did those outs go?
There are a number of culprits. While both teams played 162 games, they played a different number of innings. The Phillies played in 19 extra-inning games, including a pair of 13- and 14-inning affairs each; the Indians played in just 14 such games, just one of which went 13 innings. All told, the Phillies batted in 1456 innings, the Indians in 1442. That's 42 outs right there.
Next, we have to consider outs made by someone other than the batter. The Indians grounded into 134 double plays, the Phillies 125; that's another nine outs accounted for. The Indians were pretty effective on the basepaths, stealing 147 bases and being thrown out 50 times. The Phillies, however, were simply the most efficient base-stealing team in major league history, and it's not close:
Year Team SB CS SB% 2007 Philadelphia 138 19 87.9% 1994 Baltimore 69 13 84.1% 1995 Toronto 75 16 82.4% 1975 Cincinnati 168 36 82.4% 2004 New York (NL) 107 23 82.3%
Rollins stole 41 bases for the Phillies last year, and was caught just six times--and he lowered the team's stolen base percentage. So, that's another 31 outs in the Phillies' ledger. On the downside, the Phillies had more sacrifice bunts, 65 to 54, no doubt the result of playing in the National League. Remove 11 outs, and the Phillies have picked up a total of 71 outs as a result of our homework. That still leaves 48 outs unaccounted for, which may come from a variety of factors--outs made on the bases, additional men who reached base on an error--that are not easily teased out of the stat lines.
What's interesting here is that, in general, you would expect NL teams to have more of these "hidden outs" than AL squads. They bunt more, and they generally also run more, and more stolen base attempts leads to more caught stealings. Somehow, these factors worked to the advantage of the Phillies instead of the Indians.
That won't always hold up. AL teams do, in fact, bunt less and steal less, although they also tend to ground into slightly more double plays. As sabermetric thought takes hold in front offices, moreover, we're seeing a trend throughout baseball towards making fewer unnecessary outs. Stolen base percentages are higher today than ever before. AL teams with strong offenses are eschewing the sacrifice bunt almost completely; last year's Red Sox dropped down just 30 of them. (Billy Beane's A's had just 18.)
The day is coming when a team will not only have an OBP north of .370, but will also avoid making any more outs on the bases than necessary. When that day comes, that team will blow past the 1999 Indians' record for most plate appearances. A team that combines the OBP of the 1999 Indians with the number of outs the 2007 Phillies had to play with would end up with around 6743 plate appearances--over 180 more plate appearances than any other team, an average of 20 per lineup spot.
While Rollins holds the all-time record for plate appearances, the Phillies do not, in fact, hold the record for most overall plate appearances by their leadoff hitter. Their total of 792 ranks 19th all-time. (They didn't even have the most of 2007; the Yankees had 793.) Here's a list of the most leadoff plate appearances for one team:
Year Team PA 1999 Cleveland 805 2006 Philadelphia 800 2003 Boston 799 2004 Boston 798 2000 Cleveland 797 2000 Houston 797
The Phillies had more plate appearances from their leadoff hitter in 2006 than 2007, and Rollins led off in all but four starts that year. So why didn't he set the record in 2006? Because he missed four games, and had only a pinch-hit appearance in two others. Give him five PA in each of those games, and he would have had 786 PA that year, instead of 758.
The day may soon come when a team gets 825 or more plate appearances from the leadoff spot, and when that day meets the day that a leadoff hitter matches the durability Rollins had last year, we can expect to see a hitter reach 800 plate appearances in a season.