November 29, 2005
Smartball and Moneyball
Sabermetric Innovation Helps Power White Sox' Championship
The 2005 World Championship Chicago White Sox got the rap of being a "hustle-ball" or "anti-Moneyball" team.
False. One of the pillars of their success was the ability to deliver on an innovation that's best known as the failed child of Bill James and Theo Epstein: The "Closer by Committee." How Chisox General Manager Ken Williams and Manager Ozzie Guillen delivered value from the discredited concept is enlightening, and, because of the team's championship, it's something that's likely (though not certain) to be imitated. As with most competitive tools, it wasn't invented from scratch, but diffused--in this case, from the other side of Chicago.
CLOSER BY SAVE MEETS CLOSER BY COMMITTEE
The save rule (1969, with modifications in '73 and '75), created a gravitational field where, to achieve comfort, managers, pitchers and fans hydroplaned from merely following the stat to Birkenstocking their behavior, perceptions and desires to conform to the stat. A line of work built up around the stat. In the eyes of many sabermetric analysts the reason for inventing the stat (measuring the value of key relief performances) disappeared. Players pitched for saves, agents negotiated to optimize save opportunities, managers managed to deliver saves, fans rooted for savers, and what seemed to get silted over in an alluvial flow of events was the underlying purpose and design of stopping runs, winning games, effective relief. The "save" moved from being a representation of accomplishment to a goal that drove in-game tactics, roster construction, career development. Closers didn't so much earn saves; they became defined by them...the rules weren't reflecting best practices, they were shaping standard...a classic path for underperformance in any field.
In 2003, the Boston Red Sox and their fresh, determined-to-overcome-comfort front office team started the season with the idea of implementing a bullpen scheme built on the underlying scheme of stopping runs. I can't find who first labeled it "Closer by Committee." But it was named to echo Whitey Herzog's 1985 St. Louis Cardinal bullpen scheme, "Bullpen by Committee" where, having lost Bruce Sutter to free agency, Whitey managed a team to a pennant applying four different pitchers who earned four or more saves while none harvested more than 19.
The Bosox' scheme was doomed to underperform because the front office violated not merely everyone's comfort and the status quo, they violated everything known and tested in the field of change management--how to prepare people for organizational change that will inevitably make the steakholders uncomfortable. There's a solid ethnologic reason there's a laugh-track on TV sitcoms, and it's the same reason movie trailers insist on exposing the plot twists of thrillers; many American people want to know when to laugh so they can fit in and not be ridiculed, and most people don't like surprises.
Management did not work to get buy-in. Inevitably, the manager found it uncomfortable (forced consideration of a few dozen extra decisions every game), the relievers found it uncomfortable (their job descriptions became loosely coupled to their expectations of when they had to bear down, get ready, come on, come out of games), the fans were confused (who's the star? who do I look for as the assigned hero? whose Fleer Ultra is worth the most?). So when the media saw the opportunity to attack the front office's experiment, they knew they already had the support of managers, players and some fans, and the concerted attack was successful.
The Red Sox started searching for a traditional closer and in mid-July acquired Byung-Hyun Kim for the role. When he failed to live up to expectations, they signed Keith "Just Plain" Foulke that offseason. Experiment, at least temporarily, over.
A SECOND TRY IN THE SECOND CITY
Very soon into the 2005 season, the Chicago Cubs' plans started unraveling, the most visible cause being LaTroy Hawkins' inability to close out wins. On April 24, the day after a 19-pitch blown-save/loss to the Pirates accompanied by an unusual quantity of fan booing, Cub manager Dusty Baker decided to take some of the focus off Hawkins by announcing a "Bullpen by Situation." It was a clever ruse he didn't fully put into effect. Dusty, knowing the press needs to write stories, made this new scheme, not yet defined & therefore something newsworthy centre-stage, and making Hawkins' recent shortfalls merely a side story. He didn't ride into the trap of the "Closer by Committee," he invented a way of talking about the scheme, something I wrote about in detail here. Between Baker's April 24 announcement and the arrival of Ryan Dempster as the traditional closer on May 8, there were four classic save opportunities and Hawkins was entrusted with three of them.
But while Baker's "Bullpen by Situation" didn't play out as he described for the Second City's Second Team, Ozzie Guillen and Ken Williams quickly adapted Baker's words into innovative deeds, rendering the words flesh, or perhaps, flash.
The White Sox started the season with their previous season's leading relief talent, Shingo Takatsu, as the reputed "closer." The challenge seems to have been that pitching coach Don Cooper, and perhaps the front office, too, never really trusted Takatsu to be a consistent dominator. In Takatsu's first spring training, Cooper and Guillen kept the 35-year old rookie out of games against A.L. teams. They talked about it openly, but with a positive coating on it (some details here). In reality, it seems they didn't trust that Takatsu's torque-y trickery wouldn't turn to trash when hitters got used to it.
And rarely has a reputed closer with a previous season of outstanding success (59 games, 63 innings, 3.75 WXL facing opposition that otherwise had a .743 composite OPS) been tossed aside so quickly. Kind of like a Fiat Spider bought at an auto auction for a low price, it's almost as though Williams and Guillen were just waiting for the first glitch to drive it off the Navy Pier to collect on the insurance. And that led through a couple of approaches, eventually to a Closer By Situation scheme that helped propel the team to their first World Series appearance since The Turk and Jungle Jim roamed the South Side.
Ken Williams carefully called it "Closer by Situation," avoiding the controversy of either the Red Sox' or the Cubs' designs. He told me the words "by Committee," to him anyway, meant weakness, incompleteness, that no single reliever was good enough. He contrasted "by Situation" as meaning putting the right talent in at the right time, a more positive angle on pretty much the same method. He also misguided the media a little, because while Guillen's application of bullpen talent up until the end game was well within "normal" standards, it was in the handling of the end game, the "closer" role, that Guillen and Williams succeeded in establishing the sabermetric dream of the star-free multiple closer slot where other recent innovators didn't succeed.
CLOSER BY SITUATION--CONTRASTING STANDARD BEHAVIOR
I'm going to use the term "Clean 9th" to describe the classically-perceived situation into which a closer is brought: to face the first batter in the 9th inning of a game that the team views as winnable. Nobody out, nobody on.
I use this as a measure that indicates with some accuracy who the manager views as the team's closer. Before I show you exactly how Guillen and Williams used their scheme, let me establish a few contrasts. I'll compare how Tony LaRussa's Cardinals and Mike Hargrove's Mariners distributed Clean 9ths. I've chosen LaRussa's application of Jason Isringhausen because LaRussa has been (imprecisely) assigned the invention of the recent closer role as a guy who only pitches in Clean 9ths when there's a save opportunity. I chose Hargrove's application of Eddie Guardado because Hargrove's a non-radical adherent of "the book," neither rigidly conformist nor wildly experimental, who I use as a surrogate for "average" behavior for a manager with a reliever generally regarded as a safe choice to close.
The table shows Guardado pitched in 58 games, and in 69% of those games started with a Clean 9th, while he had what I call a save situation in 86% of his appearances. I'm defining that as appearing in the 7th inning or later with the conditions being as defined in the most recent delineation of the save rule.
Percentage of Total Appearances Potential G Clean 9th Save Sit'n* Guardado 58 69% 86% Isringhausen 63 62% 79% Hermanson 57 44% 77% Takatsu 31 26% 35% Marte 66 9% 56% Jenks 32 9% 34% * - Enter game in 7th or later with a lead as defined by the most recent Save rule wording.
Were the White Sox being honest about their "by Situation" appelation? Yes, although it was strenuously applied only in one of four phases to their championship run, phases I'll describe in better detail later.
The four relievers Guillen used as "closer" during the season contrast strongly with both Isringhausen and Guardado, most noticeably in the Clean 9th scenario. The White Sox were more likely to change relievers in mid-8th and 9th inning and used more different arms in "save situations." Dustin Hermanson had a handful fewer saves than Guardado or Izzy, but did it with far fewer Clean 9th entries. Guillen was just not wedded to applying his closer as a Clean 9th entry dude.
CLOSER BY SITUATION--THE DETAILS
The White Sox bullpen went through four approaches during the season, with Guillen and Williams showing a strong ability to experiment and change their changes as the circumstances evolved. The four approaches appear in the following table. When I say "Clean 9th or last," I'm indicating a situation that could be either a Clean 9th or a different inning/base-out situation where the reliever was intended to close out the game in a save situation. And as you'll see, Ozzie wasn't making the pandemic corporate management frell-up of thrashing around re-orging every time someone didn't work out perfectly. His moves were calculated, and made with great sensitivity to the rules of change management, changing in ways that preserved runs and morale both.
From To "Closer" Clean 9th or last Other Clean 9th or last Opening day May 5th Takatsu 10 Hermanson 4 May 11th August 24th Hermanson 30 Marte 6 August 25th October 1st Jenks & Hermanson 8 Marte 4 October 4th October 26th Jenks 5 Buehrle 1# 9 Clean or last means the pitcher was brought in either with a save-level lead in the 9th with no outs and none on, or was brought in to replace the pitcher who was brought into a clean 9th but who left runners on base still holding a lead.
In Phase I, Takatsu was "the" closer, though as you can see, it didn't prevent Guillen from using Hermanson in Clean 9ths or intended last reliever appearances. And when it came time to institute a change of approach, the manager found a cool middle ground. He didn't ride his chosen closer into the ground like Jim Fregosi in the '93 World Series, waiting for him to spontaneously combust; he didn't cast him from Eden with vituperation after a lousy performance that cost the team a game. Guillen threw away Phase I tactics after May 5th's Chisox victory during which Takatsu got the save, using an endless-seeming 28 pitches to walk a pair, strike out one looking and yield a couple of authoritatively-rapped fly ball outs.
In Phase II, the plan was different, more like the recent tradition of closer, though not fully there. Hermanson got the primary close-out-the-game function with left-handed Damaso Marte closing out some wins--not frequently, but more often than other teams would go to a set-up man for such work. Phase II would end the way Phase I had, not on a terrible note, but from the reverberations of lower effectiveness, perhaps stemming from Hermanson's bad back. After five consecutive effective apperarances from August 4th through 10th during which Hermanson closed out five games while allowing a single baserunner, The Springfield Rifle got five days off & struggled on the 16th. Eight days later he got his next appearance and closed out a game Guillen had tried to close with Cliff Politte and Marte, both of whom left runners on base. The next day, Hermanson came on with one out in the 9th and the tying runner on, let the run in but got the win from a 10th inning rally. No loss, but no longer Phase II.
Phase III was a true "Closer by Situation." From August 25 through about September 13, it was an almost purely balanced end game approach based on Hermanson, Marte and Bobby Jenks. Jenks had moved up from lower pressure situations. After the 13th, it was mostly Jenks closing out games. Guillen managed to use the strengths of all three while nursing the weaknesses of each (Hermanson's fragile back, Marte's late season inability to smoke lefties, Jenks' relative lack of experirence under pressure), and when Jenks seemed to emerge without a weakness, Guillen decisively gave the youngster a couple of weeks to hone his approach before the playoffs.
Having already changed their approach twice during the regular season, it shouldn't have been a surprise Williams and Guillen shook up their pitching application design again for the playoffs and Series. While advance scouting might have prepared opponents for the Closer By Situation, Guillen closed games with Jenks the way he had for the last couple weeks of the season. Besides the Double Wide Trailer of Infinite Doom, only Mark Buehrle got a relief appearance designed to finish off a game, and that only once.
The White Sox application of Clean 9ths is a little differnt from the other two teams I used in the core sample. You can see this in the following tables. They show the number of appearances pitchers who started Clean 9ths got in several situations: in games the team was well behind (represented as <= -3), games that were winnable but not save situations (-2, -1, 0), basic save situations (+1, +2, +3) & finally, big leads (>= +4).
Clean 9ths - Cardinals DERA WXL OOPS <= -3 -2, -1, 0 +1, +2, +3 >= +4 Isringhausen 2.53 2.50 .581 1 4 39 (93%) 7 King 4.15 -1.52 .813 3 3 1 3 Tavarez 4.14 1.37 .754 2 5 2 4 Al. Reyes 2.52 2.27 .546 1 4 Flores 4.96 .55 .695 1 3 Eldred 2.60 .64 .700 2 2 3 Thompson 3.93 1.26 .634 1 Clean 9ths - Mariners DERA WXL OOPS <= -3 -2, -1, 0 +1, +2, +3 >= +4 Guardado 3.71 1.39 .669 1 2 40 (95%) 9 Putz 4.05 .06 .706 3 3 1 2 Nelson 4.17 -.64 .690 2 1 2 Villone 3.20 -.68 .658 2 2 1 Sherrill 5.56 .81 .631 2 1 Mateo 3.33 .68 .694 Hasegawa 4.18 -.31 .658 2 2 Thornton 5.13 -.70 .794 6 1 1 Clean 9ths -Chisox DERA WXL OOPS <= -3 -2, -1, 0 +1, +2, +3 >= +4 Hermanson 2.86 2.61 .634 4 25 (57%) 2 Marte 4.14 .50 .789 1 5 6 2 Jenks 3.48 .90 .649 1 3 5 Takatsu 2.58 .39 .901 1 8 1 Politte 2.24 3.27 .575 6 2 2 Vizcaino 3.92 -.35 .757 1 1 5 Cotts 2.44 1.45 .527 1 1 2 Walker 8.53 -.37 .762 2 1 4WXL - Expected wins added over an average pitcher, adjusted for level of opposing hitters faced. WXL factors in the MLVr of the actual batters faced by the relievers. Then, like WX, WXL uses win expectancy calculations to assess how relievers have changed the outcome of games.
THE FUTURE OF CLEAN 9THS
If the diffusion of innovation works normally here, other teams will adopt the South Siders' approach, many without getting the context right, some spot on. That's not automatic. As Bill James noted in his Guide to Baseball Managers, the transition from reliever-as-failed-starter to skilled-relief-specialist was made by a successful team, the 1924 Champion Washington Senators, and no-one followed the successful experiment; it sat on the shelf for about 25 years before it was widely adopted. It could happen to the Closer by Situation, too.
Three of the factors that made it possible for the Chisox to execute so successfully are not available everywhere.
The first is a bullpen stocked with multiple strong talents with a mix of pitch arsenals and deliveries, a polycultural group that can complement each other in getting different kinds of batters out easily.
The second is a team where strong collaborative consciousness takes priority over individual ego. When Hermanson was "demoted" for Phase III, he didn't complain--he acted the good citizen and publically supported the idea that whatever made the team most effective was good enough for him. Williams traces part of the team ethos to a move Guillen and he made in late 2004. When Mike Jackson complained about not getting the work he wanted and Guillen called Williams to report on someone whining, Williams interrupted the manager, "Just waive him...don't even tell me who it is...and tell the team I didn't even want to know who it was when I told you that." The Pale Hose deliberately collected a filibuster-proof majority of players who were comfortable being a "we" team, a predominance large enough to drag the waverers along through social pressure.
The third is that this Closer by Situation wasn't an approach inspired by budgetary objectives. This is an important lesson for managers in all fields. While most of the publically-available enthusiasm for applying the model is about how much cheaper it would be to do without a classic high-priced "closer," the White Sox' design and execution of it wasn't about saving money, it was about preventing runs. As in any endeavor, an initiative created with the goal of trimming costs will usually trim short term costs, but it will rarely, and only accidentally have a positive effect on quality. Williams and Guillen had a qualitative goal and they, at least for this season, achieved it. The financial high benefit/cost ratio was merely a side benefit.
None of the three factors is typical. Following the Closer By Situation model may prove a harder task than just overcoming the comfort of using what have become standard bullpen management practices.James Click contributed research to this column
Jeff Angus is a management consultant and he writes a weekly column on sabermetrics for The Seattle Times during the season. He's the author of "Management by Baseball," which Harper Collins will release next summer, and of the Web log at http://cmdr-scott.blogspot.com.