At the end of the 2006 season, I posted my analysis (on another site) about how many wins a manager is worth to a team. At that time, Maddon was under fire from many for regressing to 61 wins from the previous season's 66. It's worth posting again, considering what we've seen in 2007-09. Here it is:
How Many Wins is a Manager Worth?
One of the most difficult things to measure in baseball is the effect a manager has on his team. How many wins is a good manager worth? Or how many losses? The literature is thin. James Click’s piece (“Is Joe Torre a Hall of Fame Manager?”) in Baseball Between the Numbers cranks and crunches a lot of numbers, but ultimately concludes it’s too hard to figure out—there are too many factors out of the manager’s control.
So I thought I’d try a different approach to the problem. Rather than try to definitively determine the impact of a manager, I thought it would be useful to figure out the greatest possible impact, which would allow me to at least define the reasonable outside envelope of potential success. To do this, I started by examining what might be considered the cream of the managerial crop, which I defined to be those managers who have had a career that has lasted at least 10 seasons in the Majors. This 10-year span does not have to be with one team—few managers have ever lasted that long in one place—but at least this group by sheer longevity has the perception of competence by their prospective employers. They also had the ability to survive in the game, in some cases, building their stature even with losing teams.
From 1903 through 2006, there have been 445 managers in Major League Baseball (http://www.thebaseballcube.com/managers/index.shtml). Of this group, few last long. Only 75 (17% of the 445) have had careers that have lasted at least 10 years. Some current managers, like Mike Scioscia, appear to be well on their way to 10 or more years, but the fact is that managers burn out, are drummed out, or simply fade away as yesterday’s replaceable field general long before they reach that milestone. So it’s worth focusing on the characteristics of this group if we’re looking to figure out what makes for a successful field manager, and how they might help their team in terms of wins.
The first thing to note about this elite group, their collective career winning percentage is .520. That suggests the collective winning percentage of the other 83% of all-time managers is only .480. That is, most managers end their career with a losing record. It is typical to lose a lot, unusual to sustain success. Of all the managers with at least 10 years under their belt, only one—Joe McCarthy—NEVER had a losing season. That unique achievement is remarkable, even considering he mostly managed excellent Cub, Yankee and Red Sox teams. It’s also telling that even Joe McCarthy managed three teams (not just one), in spite of his sustained success. Very few managers have lasted an entire career with one team, Connie Mack being the ultimate example in his 48 years with the Philadelphia A’s.
I surmised that in order to measure the most successful managers (i.e., those with the greatest W/L effect on their team), it would be useful to first examine those whose stints had lasted at least 5 years with a single team. Since 1903, there have been only 94 such instances, the longest of which were Mack (48), McGraw (30), Alston (23) and Lasorda (20). As a separate study, I also looked at all stints of at least 3 years, to see if the effect on their team would be any different. I hypothesized that this group would have a lesser effect on their teams, on average. Retrosheet.org provided the necessary data.
The major findings:
- On average, managers that lasted at least 5-years with their team were worth 7.7 games (on average) to their team. This was calculated by looking at the team’s winning percentage in the 5 years previous to their arrival, and comparing it to the first five years of their stint with the team. This percentage was then normalized to a 162-game schedule, accounting for those who managed before the modern season length, or to account for strike years. On average, these managers took over teams with a previous 5-year W/L % of .485, and they produced a record of .533.
- There is tremendous variation among this group of managers in terms of their success relative to the teams they took over. The most dramatic examples of improvement were John McGraw, Davey Johnson, and Danny Murtaugh, all of who took over terrible teams (Giants, Mets, Pirates), and saw their teams improve by over 30 games per year on average in their first 5 years compared to the previous 5.
- At the other end, Ralph Houk (Tigers and the Yankees in his second tour) and Joe Torre (Mets), saw their teams actually decline by 14 to 16 games in an average year under the first five years of their tenure, as compared to the previous 5 years. Torre’s failure with the Mets is especially noteworthy considering his later success with the Yankees, where in his first five years, his teams were on average 13 games better than the 1991-95 Yankees that preceded him.
- Among managers that lasted a minimum of three years (and there are many), the effect on the team is less pronounced, but generally similar. This group on average improved his team’s fortunes by about 5.1 games annually in that three-year period compared to the previous three (improvement from .484 to .516).
- Again, there is tremendous variation among this group. The ranged from McGraw (+44 games), to Fred Hutchinson of the 1952-54 Tigers, where his teams were 25 games worse each year than the 1949-51 Tigers. No surprise that he didn’t last a fourth year, and somewhat shocking that he even lasted three. Actually, it’s interesting that 25% of all managers with at least a 3-year stint group actually saw their teams regress compared to the previous 3 years before their arrival.
- With an average of +5.1 games and a standard deviation of 11 games (a huge variation), this suggests that a “normal” expected range for all managers lasting at least three years with a single team is anywhere from a 16-game improvement to a 6-game regression. This fact alone suggests it’s very difficult to generalize about the W/L effects of a manager. At the same time, it suggests that factors beyond the control of the manager are often major culprits in their “failure”, and are recognized as such by ownership, which is why such managers remained as long as they did in their roles.
- In general, managers who managed long stints for multiple teams were most successful in their earlier assignments. Chuck Dressen (Reds, Dodgers, Senators, and Tigers) saw improvement with the first two, regression with the last two. Leo Durocher (Dodgers, Giants, Cubs) followed the same pattern. Others like these two included Lou Boudreau, Mike Hargrove, Felipe Alou, Whitey Herzog, Davey Johnson, Al Lopez, John McNamara, Branch Rickey, Buck Rodgers, Buck Schowalter, Casey Stengel, and Dick Williams. There are, however, some notable exceptions, including Sparky Anderson, Bobby Cox, and Bill McKechnie, who either improved over time, or consistently sustained success.
- Some specific examples are enlightening, others unexpected. Both Walt Alston and Tommy Lasorda oversaw teams that regressed in their first three years (and five years, for that matter) compared to the previous periods of the same duration before their arrival (3 to 5 games). This regression nearly cost Alston his job early in his career, but it also has to be remembered that both took over relatively successful programs, and expectations were high. Both were very successful in the long run. Bobby Cox improved the Braves in 1978-81 by 7 games, but did even better with the Jays and Braves later (+26 games in each case!). Tony LaRussa had great success with the A’s (+20 games), but only marginal success (+1 to 3 games) with the White Sox and Cardinals).
- What about Lou Piniella? Right about average. Regressed by about 5 games with the Yankees, but saw a similar magnitude of relative success with the Mariners, Reds, and even the Rays.
- Many managers showed great success with one team, but failure with another (Torre is maybe the best example, but others like him include Fregosi, McKeon, Mauch, Tanner and Dick Williams). This suggests that the “fit” is just as important as the manager’s inherent skills in determining success.
- And what does this all say for Maddon? He inherited a team that had gone .412 in the previous three years (average 66 wins per year), considerably worse than the average team that is taken over by a new manager (.484, or 78 wins). Thus, among managerial stints with similar starting points (those with previous three-year records ranging +/-.025 from .412, or .387 to .437), the average three-year improvement was 14 games, but there is a huge standard deviation of 12 games. Thus, it is reasonable to expect anywhere from a 2 to 26 game improvement as “normal” in this group. Anything more would be called “above normal”, anything less “below normal”. In his first year, Maddon won 61 games, or 5 less than the average for the previous three years. If Maddon can average anywhere from 69 to 80 wins over a three-year period, this would be considered “normal” within expectations. That would essentially mean at least 75 wins per year in 2007 and 2008.
Obviously this study is imperfect, full of assumptions. But the basic assumption is to measure success based on those managers with the greatest longevity in the game, who by definition are the survivors, even if their individual records do not always reflect their success or stature in the game. By comparing Maddon to this group is to compare him against the elite managers in the game from a historical perspective. And yet by this high standard, if he can manage 75 wins this year and next, he will be holding his own within reasonable expectations against this elite company. Anything less, and the debate is open. But you can’t “know” anything until well into the 2008 season.
That’s putting the matter in perspective.<!--EndFragment-->