Even before the Philadelphia 76ers replaced Maurice Cheeks with Tony DiLeo on Saturday, it was evident something unusual was going on when it came to the NBA and coaching changes. Cheeks ended up being the fifth coaching change already this season, more than the last two years combined.
When there was a similar run of firings during the 2003-04 season--which ended up tying a modern league record with seven midseason coaching changes--I felt it could be explained by the wide-open Eastern Conference (six of the seven moves came in the East), with multiple teams feeling they could compete with a new head coach.
This year's changes defy such simple explanation, so at last week's Portland/Orlando game, I turned to a couple of experts in the field--head coaches Stan Van Gundy and Nate McMillan. Both have experience with midseason changes--Van Gundy was replaced by Pat Riley when he resigned in December 2005, while McMillan got his first chance as a head coach taking over for Paul Westphal in November 2000.
The two coaches took the question in very different directions.
"I think it gets contagious," said Van Gundy. "It doesn't happen often that first 20 games of the year. I think when that first guy does it, then the second guy feels like he's got some cover because he's not the only guy doing it. Then it's easier for the third guy and the fourth guy. To have four coaching changes before you hit the 20th game - you don't see that very often."
"That happens," noted McMillan. "In this league, you've got to win. There's going to be a time when you as a coach, it's going to be based on winning and losing. If you win, they can still make a change. It just depends on what the organization feels like doing. If you lose, there's a good chance that they're going to make some changes. That's our sport. It's a part of it. I don't think from what I've seen as far as the coaches that have made changes, I don't think they were really surprised that it could happen. It's what we go through. The bottom line here is about winning."
To take a look at the two suggestions and further investigate midseason coaching changes, I used Basketball-Reference.com to research a comprehensive list of such changes dating back to the 1989-90 season. Throwing out a handful of situations where it was obviously the coach's decision to leave--like Van Gundy's brother Jeff abruptly resigning in New York early in the 2001-02 season--there have been 68 midseason coaching changes, an average of 3.6 per season. Here's how they break down year by year, including a look at how many changes were made after 30 games or fewer.
Year Tot <30 Year Tot <30
2009 5 5 1999 4 3
2008 1 1 1998 3 1
2007 3 1 1997 7 3
2006 2 2 1996 3 1
2005 6 1 1995 2 0
2004 7 3 1994 1 0
2003 4 2 1993 4 3
2002 4 3 1992 5 1
2001 2 1 1991 2 2
2000 5 2 1990 3 2
To have five coaching changes in a season is not unusual, though it should be noted that there are other coaches on the hot seat, so teams may not be done yet. Where this year has already been out of the ordinary is in terms of how quickly teams have pulled the trigger on their coaches. We're still not to the 30-game mark, with no team playing more than 25 games so far, but 2008-09 has shattered the high in this span, which had been three coaching changes in the first 30 games.
That seems like some anecdotal evidence of Van Gundy's theory that teams are more likely to fire their coach when they can point to other coaching changes. Looking through the history, there are several other similar runs of coaching changes. In 2004, Boston, New Jersey, New York and Philadelphia all made changes within the span of about a month. Two seasons earlier, Dave Cowens, Tim Floyd and Dan Issel all left in a two-week period.
To attempt to quantify this, I looked at the distance (in games for the team in question) between coaching changes. On average, given the period in which coaching changes realistically happen (the earliest was Cotton Fitzsimmons resigning after an 0-8 start, while the Knicks fired Don Nelson with just 13 games remaining in the 1995-96 season), there is a firing by some team every 16.5 games. Therefore, in theory there's a 50-50 chance that when one coach gets fired, another one should get the axe within the next 16 games. In practice, that happens more often--55.1 percent of the time. It's not overwhelming, but that does offer some data to support what Van Gundy has observed.
McMillan's theory that ultimately things do come down to wins and losses is more subjective and difficult to evaluate. Still, in doing the research I was surprised by how frequently teams going nowhere change coaches. Essentially, midseason coaching changes can be distilled into two main categories--teams that aspire to turn things around and make the playoffs, and teams that know this year is lost and are building for the future.
The sweet spot for coaching changes comes when a team underperforms playoff-level expectations in the very early going, with Philadelphia being a good example this year. The rationale in these situations is obvious. If things go well, there's still time to make a run at the postseason. The downside is limited because the team was already playing poorly; if it plays worse, so what?
The cases where teams came into the season knowing it was devoted to development and not winning are somewhat more nebulous. Sometimes it's a matter of getting the right leader in place for the long term, like Flip Saunders taking over in Minnesota during the 1995-96 season or Hubie Brown replacing Sidney Lowe in Memphis in 2002-03. At others, like Kevin Pritchard's brief run on the bench, it's a chance to evaluate young talent and prepare for hiring a new coach during the offseason. Still, some changes appear to be nothing more than shuffling deck chairs. What exactly did the Cleveland Cavaliers think they would accomplish, for example, by firing John Lucas and replacing him on an interim basis with Keith Smart 42 games into a season devoted to the pursuit of presumptive No. 1 overall pick LeBron James?
There does seem to be a point at which, no matter the expectations, teams simply will not tolerate continued poor performance. Randy Wittman's exit in Minnesota seems to fall into this category, as does P.J. Carlesimo's in Oklahoma City to some extent. The funny thing is McMillan had complete support from the Trail Blazers front office during the rebuilding effort that has produced one of the league's most promising young teams. However, as he pointed out, expectations are continually changing, and if Portland was to struggle this year, McMillan's job might be placed in jeopardy.
The two categories of coaching changes are also useful for evaluating midseason changes. Actually, in this case it probably makes more sense to split those aiming for the playoffs into an additional two groups--teams coming off of a playoff appearance and those who merely aspired to make the playoffs. (The best example of the latter being the 2002-03 Atlanta Hawks, infamously declared "playoff bound" by coach Lon Kruger, who paid the price for his optimism during the season.)
Predictably, the most successful coaching changes have come amongst the teams with playoff track records. Of the six teams who improved their winning percentage by at least 20 percent after a midseason coaching change, the only one which did not make the playoffs the previous season was Brown's 2002-03 Grizzlies.
Let's take a look at the average performance of each type of coaching change:
Type Old% New% Diff
Playoff .399 .487 +.088
Aspiring .406 .402 -.004
Non-Playoff .230 .255 +.025
When broken down this way, the comparison between past playoff teams and playoff hopefuls in particular is very interesting. By far, coaching changes on teams that made the playoffs the previous season are most successful, while aspiring playoff teams who change coaches tend to play at virtually exactly the same level. That seems to indicate that the GMs of these teams might have overestimated the talent level and been too hasty to make moves.
There's another interesting way to look at the data, one John Hollinger remarked upon in a column on midseason coaching changes way back when he was at SI.com. The earlier in the season a change is made, the more successful it tends to be. Teams who change coaches within the first 25 games of the season improve their winning percentage an average of 14.0 percent. That drops to 4.1 percent for teams making changes between the 25-game mark and the halfway point of the season and to a negligible 0.4 percent for teams changing coaches in the second half.
There are a couple of issues with these numbers. First, the early firings include a pair of 0-8 teams (the 1996-97 Suns and the 2002-03 Grizzlies), as well as the 1-10 2003-04 Orlando Magic. I could coach those teams to better records the rest of the way, so of course actual NBA coaches managed the feat. Even removing them still leaves the early group with an average improvement of 9.7 percent. Second, late changes are hurt by the small group of teams going into tank mode. Those teams aside, there aren't a ton of examples of teams in the thick of the playoff race successfully changing coaches and making the postseason.
Add it all up and it becomes apparent that the most successful coaching changes are made by former playoff teams with high expectations early enough in the season that the new coach has a chance to make a difference. Sound familiar? That's exactly what we saw from the Philadelphia 76ers on Saturday. Only time will tell if DiLeo can engineer a midseason turnaround, but the numbers are in Philly's favor.
Kevin Pelton is an author of Basketball Prospectus.
You can contact Kevin by clicking here or click here to see Kevin's other articles.