Does Firing The Manager Mid-Season Help A Team?

The Diamondbacks are approaching the end of a long, largely forgettable season. Interim manager Kirk Gibson enjoyed a solid home-stand, winning five of the six games played, but it's still uncertain whether he will be back for the 2011 campaign. Will new GM Kevin Towers want to have his own man riding the bench? Or has Gibson done enough to keep the position? I imagine we'll know relatively soon after the season ends next week.

But in the meantime, I thought I'd delve a bit more into the numbers, to see what kind of effect firing a manager in the middle of a season has on a team's W-L record? Does it tend to galvanize them into playing better, with the aim of impressing a new boss? Or does the upheaval cause more problems than it solves? After the jump, we have the numbers for all teams that have fired their managers mid-season since 2001, and will take a look to see what they can tell us.

For each case, we give the record a) in the year preceding the change, b) up until the manager was fired, c) after the manager was replaced, and d) the following season. If multiple men e.g. a caretaker, were at the helm after the replacement, their records are combined into one "post-sacking" W-L figure. Where the "post" figure is in italics, that means the manager was fired mid-2009, so it's a 2010 figure, and therefore incomplete. All numbers are through Sunday night though, so we only have a week left to go. There is also some overlap of seasons, when teams did it in consecutive years.

That's 34 cases in all. Let's take a look at the the numbers and see if we can detect any trends. Firstly, it doesn't take long for things to go sour: 41% of the sample had a winning record the year before the manager was fired, and the average record in those seasons was 78-84 - there are ten teams currently below that win percentage this season. The one that stands out in particular were the 2001 Indians, who made the playoffs under Charlie Manuel, who got himself canned by the All-Star Break the next year - GM John Hart had been replaced over the winter. Manuel is the only manager in this study to get fired the season after taking his team to the post-season.

You can get fired at any point - from as soon as six games in, to as late as the penultimate day of the season, when the Phillies dumped Larry Bowa in 2004. If we divide the season in six equal sections of 27 games, five managers got their marching orders in the first sixth, with the numbers for the subsequent sections being eight, seven, seven, three and four. So it seems that May through the All-Star break [usually happening early in the fourth division] are the dangerous times for a manager. Only four of the 34 went down while above .500 in the standings; the most memorable example being Ned Yost, canned while his Brewers were leading the 2008 NL wild-card race.

But when we look at overall win percentages, we see that managerial replacements have only the tiniest effect on a team's success. All told, we're looking at a sample size of about 5,500 games for each year. In the one before the firing, teams won at a .480 rate. In the season where they got rid of the manager, that dropped to .441 before the dismissal. The replacement(s) did do better: very, very slightly, getting the win percentage up to .457 - given the average replacement worked 88 games, that works out at a mere 1.4 additional wins over the rest of the season, compared to the man they replaced.

25 of the 34 posted losing records, compared to just nine with winning ones. The three best belong to the World Series winning 2003 Marlins (.605 under Jack McKeon), play-off reaching 2009 Rockies (.638, Jim Tracy) and 2004 Astros (.649, Phil Garner), who fell one game short of making the WS. It's examples like that which mean that teams will keep changing horses in mid-stream, hoping similarly to catch lighting in a bottle. However, it's worth noting, none of the trio came close to matching that W% the following year, posting numbers of .512, .535 and .549, so each lost close to a hundred points.

Even more disheartening is the general lack of further progress made by franchises in the season after a firing, with improvement by no means certain. 20 of the 34 had better records in the year subsequent to a firing; one posted the same record, and 13 actually got worse. Overall, however, the average win percentage climbed just eight points to .465, about 1.4 extra victories over the entire year. 68% of teams still ended up being below .500 again at the end of the next campaign, and only four (12%) managed more than 86 wins: the 2002 Red Sox, 2003 Cubs, plus the 2005 Astros and Phillies.

To conclude, this seems more evidence that managers don't actually have an enormous amount of effect on their team's W-L record. It's difficult to separate managerial effect from player effect - is Gibson's improved W% because he's a better manager than AJ Hinch, or because Gibson has Daniel Hudson's 7-1 record and 1.69 ERA? However, I was surprised how little of a difference the results showed in teams' records before and after a manager is canned. A mere sixteen points is much less than I'd have predicted before crunching numbers.

While there are cases where it has had a much bigger positive impact (Buck Showalter in Baltimore would be this year's prime example), those are definitely the exception rather than the rule. There seems little rhyme or reason to what works or doesn't, e.g. Jim Tracy had managed three straight losing seasons, before he turned the Rockies around last year. But whether it's Gibson or someone else in the D-backs dugout next season, there's not much evidence to expect this change, by itself, will be sufficient to get us back into contention.