Same as the old boss?
Changing coaches doesn't always change a team's fortunes
Posted: Monday November 24, 2003 2:07AM; Updated: Monday November 24, 2003 2:07AM
By John Hollinger, SI.com
It was about as unexpected as the sun rising. Once the Orlando Magic dropped 10 straight games to fall to 1-10, it was inevitable that their head coach, Doc Rivers, was fired.
Coaches of underachieving teams, especially to the extent Orlando was, are nearly always handed their walking papers. Just in the past five years, a head coach has been told, "Thanks, but get lost" in the middle of a season 21 different times.
Based on averages, we can expect three more coaches to meet Rivers' fate before this season ends. Given how often it happens, one would expect that it's nearly automatic for the teams to perform better when they fire their coach in the midseason. In reality, the difference has been pretty underwhelming.
Look at the 21 teams that changed coaches during the season. Note that in addition to those who were fired, I included all the coaches who issued their resignation, since virtually all were made under duress. As the chart at left shows, the teams had a winning percentage of .345 before cutting the coach loose, so it's no wonder the guys got the ax. But even after the coach was gone, the teams only won at a .382 clip.
One might point to an improvement of 37 points in the win-loss percentage, but a small improvement is to be expected for one very simple reason: Coaches are rarely fired after a win. Thus, one might expect a team to slightly outperform its previous record after the coaching change. For an obvious example, consider last year's Grizzlies. They started the year 0-8 before firing Sidney Lowe, for a winning percentage of .000. However, it's reasonable to assume that if Lowe hadn't been fired, Memphis still would have won a game (or even two!) at some point, and finished with a better winning percentage.
If we expand the study to go back to 1990, we see a similar trend. In the chart at right, the 50 teams that changed coaches improved an average of 32 points on their winning percentage over that time, with 33 of the 50 finishing with a better record. That's nothing to sneeze at, but again most of the difference can be explained by the fact that coaches on winning streaks don't get fired. Rivers is a perfect example -- the Magic won their first game before dropping 10 straight, but it's scarcely believable that, regardless of what Rivers did, they would continue at that rate over a full season and finish 7-75. After all, that would be two games worse than the worst record in history, and that team didn't have a Tracy McGrady.
Still, based on the track record, the Magic can expect only a modest improvement from their 1-10 start. If their winning percentage improves by the average of 32 points the rest of the way, that would still leave their final record at a rather unappetizing 10-72.
But there's a silver lining: the Magic acted early. Teams that pull the trigger in the first 15 games have tended to have much more success than the others. There have only been six in the past 13 years, but the chart below shows that all six finished better than they started, and five of the six improved considerably.
Two of the six were 0-8, so it's no surprise that the teams improved, but the extent of the improvement is impressive. The 1996-97 Suns started 0-8 and still made the playoffs after they changed drivers, and the 1998-99 Hornets and Lakers also made impressive stretch runs after ho-hum starts.
Based on the 174-point average improvement in those six teams (again, we need to take this with a grain of salt because of the small sample), we would be looking at the Magic to win 19 games the rest of the way and finish the year at 20-62. That's a lot better than 10-72, but it would still be a crushing disappointment.
The reason is that, unlike most of the teams in this study, the Magic had a winning record the year before and were commonly expected to do so again. A 1-10 record is bad anywhere, but especially for a team that expects to be playing in May.
So let's narrow things down further by studying teams that both had a winning record and went to the playoffs the year before. Only 22 of the 50 teams meet the criteria, and the results are more encouraging for Orlando. The teams had an average winning percentage of .450 before the coaching change, and .505 afterward, with 15 of the 22 improving after the switch.
Here's where it gets real interesting for the Magic: The worse the teams started, the greater the impact of the coaching change. Take out the seven teams that fired the coach despite playing .500 or better, and the improvement is from .409 to .488 -- 89 points instead of 55.
And if we only include the teams that started below .400, like the Magic, the trend is even stronger. There are only seven, but they all improved, and the improvement -- from .305 to .512 -- is more than 200 points. It's a decent bet the Magic will experience a similar improvement from their 1-10 start, but here's the punchline: A 200-point improvement in their winning percentage still only would put them at 22-60 when the season ends.
Based on the data from both the playoff teams that disappointed and the teams that changed coaches early, we should expect Orlando to finish with somewhere between 20 and 22 wins. That's better than 1-10, but still light years away from respectability. All of which supports the conclusion that changing coaches is rarely a panacea. Certainly it's the most expedient move available for a struggling team, especially given that most player contracts are guaranteed. Additionally, the study shows that it can achieve a slight improvement in results for struggling teams, especially when we look at playoff teams who have disappointed. But in the bigger picture, the coaching switch can only accomplish so much. More often than not, a midseason coaching change is just a sign of a team with much larger problems to fix.
John Hollinger covers basketball for SI.com and is the author of Pro Basketball Prospectus. Click here to send him a question or comment.