LCS: Guide to Hockey

The Regression Effect:

No More Dirk Graham, Less S’Mores

by Howard Fienberg, Correspondent
Issue 114, February 24, 1999

Common wisdom is hard to argue. Being portrayed on the cover of Sports Illustrated is bad luck and will hurt your career. Rookies usually fizzle in their second year in the NHL. Firing the coach will improve the team’s performance.

Being on SI’s cover is, in fact, usually followed by a downturn in one’s career. Why? A plague on both your houses and may Mathew Barnaby come knocking on your front teeth with a chrome stick. Well, that is what the athletes believe. Actually, it is because you usually reach the cover when you are at the pinnacle of your career, not the end. It is simple logic that thereafter, your career will be in some form of decline.

Sophomore players tend to really blow. Bryan Berard had an outstanding rookie season (‘96-‘97) and took home the Calder Trophy as Best Rookie. And the next year? His plus-minus rating was so bad he was getting defensive zone tutoring from Phil Housley. But this season? He’s making end-to-end rushes and providing the power in Toronto’s power play. Must have been the sophomore slump. Teemu Selanne’s rookie season, ‘92-‘93, when he blew away the league with the highest rookie goal total ever, was followed by a rather fallow second season in the harsh cold climes of Winnipeg.

But no. The “slump” is a myth. There is no overall tendency for performance to be lower in a player’s second year than in his first. However, it is common for certain players to find themselves in a “slump” in their second year. Those certain players happen to be those who had the most outstanding rookie stints. Their success was probably bolstered by a bit of random error, and can be expected to decline to average the next year.

The regression effect is when two variables are imperfectly related (like a sophomore and rookie season), extreme values - high or low - in one are often matched by less extreme values in the other. So, a truly poor or absolutely great season will tend to be followed by an average one. Since a player’s performance at different periods of time is imperfectly correlated, we can use the regression effect to guess that a “hot” streak will be followed by an average period. That average period will APPEAR as cold as Winnipeg to an uppity fan, a superstitious player, or the likes of Al Strachan, because they are blinded by the hot point - which was most likely just a statistical irregularity.

So perception enters into it. Who remembers Ryan Smyth’s rookie season? Let me refresh your memory - 2 goals and 9 assists. Real Calder Trophy stuff, right? And the ‘96-‘97 season, he breaks out - 39 goals, 22 assists. He didn’t suffer no danged sophomore slump, because no one noticed him the season before. Just like the cover of Sports Illustrated, we notice the guys who are at the top, and we remember them, to the exclusion of the nobodies. Fans remember dazzling debuts, outstanding rookie seasons. So a belief in the sophomore slump results from a fan’s tendency to only the remember the outstanding bits, and failure to realize that the grand plays of such players are likely to regress in their sophomore seasons.

Which brings me to Dirk “Diggler” Graham. Nice guy, that Dirk. Maybe you don’t think he deserved to be put in the unemployment line. Maybe you think that GM Pulford handed him a deck without any aces. Or, any Kings, Queens, and Jacks for that matter. Hey, let’s face it, Dirk was lucky he had any cards at all, not to mention so many twos and threes. And those twos and threes don’t do much unless you and/or your buddies have had a lot of alcohol. Maybe you think the moon is made out of green cheese. Well, frankly, I don’t much care what you think. Because firing the coach is a sure-fire cure to improve a team’s record. Unless the guy is singing off key and ruining the track, this is so dead wrong, we had to dig up Clarence Campbell to prove it. And boy, was he mad!

Generally, it is true that teams tend to perform better after firing the coach or manager. (Exceptions do exist, like the Lightning. But they are owned by aliens from Planet 10, so they don’t count. Come to think of it, most of their players mail in their performances from Planet 10 as well.) But that upswing in performance is not “caused” by the firing, even is it may be correlated. Certainly there are cases where the dumping of an unpopular coach (Hey, calling Iron Mike Keenan) may have some causal effect on a team’s playing level, but that improvement could just as likely be accounted for by the regression effect. Especially since they are usually fired when a team has hit rock bottom, there is only one way to go - up.

So expect the Blackhawks to start to play a little better, maybe even start winning some games. Yeah, winning, remember those days Chicago? But don’t expect a real turnaround. Luck can only take you so far, and the odds are stacked against it in the windy city.

http://www.lcshockey.com/issues/114/114fien.asp


return to Howard's published works... or Howard's main page....