Does Familiarity Breed....A Hot Start?

Atlanta's GM Don Waddell made this comment in a post-season interview: “Unfortunately, it took the guys the first 15-20 games to come together as a team. We had 13 new players that had been added since the last season so there was a transition period.” (http://www.atlantathrashers.com/)

This statement made me curious about the effect of roster turnover on how NHL teams started the season. After all some clubs saw massive roster changes between the end of the 2003-2004 season and the start of the 2005-2006 season. So I decided to take a look at all NHL teams.

First I went through the every NHL team and totaled up all the games played by players who were also on the final 2003-2004 roster. I then divided the games played by returning players by all games by position players. This gave me percentage of all games played by returning players. I used this as my measure of familiarity. I assume that the more people you have returning on your roster them more likely they are to know each other’s tendencies and therefore have an advantage when playing against other clubs that are still learning how to play with their teammates.

In the table below I have ranked all 30 NHL clubs from top to bottom based upon this familiarity index. The high was provided by Vancouver where almost 80% of all games played were by players who had been with the team in 2003-04. The low was the NY Rangers who with only about 20% of games by returning players, so there was quite a wide difference between extremes.

Next I calculated the team points-per-game earned in October (OCT W %) and the points-per-game rater from November to the end of the season (Nov+ W%). Finally I coded each teams’ October start as better or worse than their winning percentage the rest of the way to see if it fits with the turnover hypothesis.

So what does the data show?

• Of the 10 teams with the most returning player games, 6 out of 10 played better in October than they did the rest of the season.
• Of the 10 middle teams, 5 out of 10 played better in October
• Of the 10 bottom teams, 3 of out 10 teams played better in October.

There is a trend here but it is a fairly weak one it seems. One thing that surprised me is that the Atlanta Thrashers fell into the middle third of the league in terms of roster turnover. Anaheim is a rare example of a team that had a lot of roster turnover and yet made the playoffs. Even more surprising to my mind is that they did a lot of roster changes in mid-season. On the other hand, mid-season changes were not as successful for Boston. Everyone thinks of the Sharks as making the big mid-season roster moves with the Joe Thornton trade, but in the Sharks case it was a case of inserting a star player into a fairly stable roster, rather than making wholesale changes in the Anaheim mode.

TEAM..RET%…OCT W%…NOV+ W%....Hot in Oct. or Not?

VAN…78%….1.50….1.06….BETTER

TB…...74%…..1.33….1.09…. BETTER

BUF…73%…..1.27….1.35…...WORSE

OTT…72%……1.60….1.35…. BETTER

DET…71%……1.83…..1.46… BETTER

SJ…….70%……1.08….1.23…. WORSE

EDM…67%……0.92….1.20…. WORSE

LA…....66%……1.33….1.04…..BETTER

MON….65%…..1.42….1.09….. BETTER

NJ…....64%…….1.09…..1.25….WORSE

MIN…..63%…..1.17….1.00…. BETTER

TOR….62%……1.17….1.09….BETTER

DAL.….62%……1.18…1.39….WORSE

NAS….58%……1.70….1.24…. BETTER

CAL…54%…….0.77….1.35…. WORSE

COL…53%…….1.18….1.15…. BETTER

CAR….51%…...1.55…..1.34…. BETTER

ATL…50%…….0.55….1.18… WORSE

STL….49%…….0.55…0.72… WORSE

CBJ…..46%……0.46…0.94…. WORSE

NYI….44%……0.91….0.96…. WORSE

WAS….43%…..0.73….0.87…. WORSE

ANA….41%…..1.08….1.21…. WORSE

BOS….37%……1.00….0.88…. BETTER

PHO….36%…..0.79…..1.03…. WORSE

PHI….36%……1.30…..1.22…. BETTER

PIT….35%…….0.64….0.72…. WORSE

FLA….34%…..1.08…..1.03…. BETTER

CHI….25%……0.55….0.83…. WORSE

NYR….20%…..1.15…1.23…. WORSE