black monday

As a warning, I don't expect most of my readers to bother with this one.  The major themes are football, business, and statistics, and I am fairly certain that only interests a small percentage of my readership.

Today is the day after the last day of the football season, and is typically known as a black Monday when teams who had disappointing seasons announce which coaches are getting fired.  This year, a lot of coaches with relatively successful careers have gotten the boot.  The names wouldn't mean much to people who don't follow the NFL, but they include people like Lovie Smith, Andy Reid, and Norv Turner, all of whom are usually in the playoffs.  And that brings me to a point.  I think that most of these firings are ridiculous, and they point to a mistake that people make when they look at statistics.  there frequently is not enough data available to make the right decision.

One that sticks out a bit to me is the firing of the Chicago Bears coach, Lovie Smith.  I have always detested the Bears, but I have long held quite a deal of respect for Smith, both on and off the field.  While Detroit struggled through ridiculously bad season after season, Chicago was graced with a defensively-minded coach who kept getting them into the playoffs, and was also someone to look up to personally.  Something to envy for sure.  This season, he was canned after a winning season that just missed the playoffs.  A lot had to do with the poor performance of the team's offense, but turn a couple of the team's losses to wins and Chicago would be falling over themselves to keep Smith.  That's the point of my contention.

The win/loss ratio for a sixteen-game football season is not a large enough data set to use to know whether a coach is good or bad.  String along several seasons of mediocre performance, and that would probably be enough, but if you have someone in-house who has a record of success, but one or two mediocre seasons, that is reason to keep rather him rather than to get rid of him.  I see this in other business environments as well.

In my experience and in discussions in my MBA classes, I have been amazed at how willing businesses are to live and die by quarterly numbers.  Most executives and financial decision-makers will have had to have had statistics classes to attain both their bachelors and masters degrees, yet they make decisions (out of necessity or not) that ignore the fact that statistical variance all but dictates that everyone will have periods where their unmanipulated numbers underperform expectations.

Looking to another business source, one of the biggest red flags for the Bernie Madoff scam was that his hedge fund never underperformed.  Statistically this was nearly impossible, even if Madoff was the wisest investor around.  Everyone misses on some quarters if they're being honest.  Rather than his consistent performance being a reason to invest with him it should have been a reason to avoid him like the plague.  People simply aren't wired to look for that sort of red flag, though, and they are wired to punish others based on the appearance of underperformance caused by statistical noise.