We now have three different Olympics to look at, and I wondered if there were any perceptible differences in player performance based on whether they played in the tournament that year or not. Let's look first at team winning percentage for players who played in the games in the years before and after the games:
W% | 1998 | 2002 | 2006 |
Oly-1 | 524 | 530 | |
Oly | 511 | 537 | 536 |
Oly+1 | 527 | 528 | 513 |
The 2006 Olympics don't have a previous season due to the 2004-05 lockout (thanks, Gary.) There's no solid pattern here. What if we compare winning percentage before and after the Olympic break?
1st/2nd | 1998 | 2002 | 2006 |
Oly-1 | -14 | -7 | |
Oly | 8 | -13 | -24 |
Oly+1 | -3 | 23 | 46 |
Again, nothing here. Clearly there is no impact at the team level. What about at the player level?
Games Played:
GP | 1998 | 2002 | 2006 |
Oly-1 | 70 | 69 | |
Oly | 72 | 73 | 71 |
Oly+1 | 69 | 68 | 70 |
Players play a slightly-higher number of games in the Olympic years. But this is selection bias – a player won't make the team if he's injured. What about scoring?
Pts/82 | 1998 | 2002 | 2006 |
Oly-1 | 55 | 57 | |
Oly | 50 | 51 | 67 |
Oly+1 | 54 | 49 | 57 |
Still nothing to hang our hats on. How about the 1st/2nd half difference in points-per-82 games?
1st/2nd | 1998 | 2002 | 2006 |
Oly-1 | -1.8 | 1.8 | |
Oly | -0.5 | -0.1 | -0.9 |
Oly+1 | -0.1 | -2.3 | 1.0 |
Again, nothing. We see the same lack of trend year-over-year in +/-, TOI, shooting percentage. I did find one stat where players got worse in the second half of the Olympic season:
Shot% 1st/2nd | 1998 | 2002 | 2006 |
Oly-1 | 0.6 | 0.6 | |
Oly | -0.6 | -0.2 | -0.9 |
Oly+1 | 0.2 | 0.0 | 0.1 |
I would still argue that this means nothing. I essentially have data with no trends in it, but the noise in my observations is large enough that some statistic will consistently point in the right direction without actually meaning anything. If participating in the Olympics has a measurable impact on player performance, I can't see it.
I'll take a look at the goalies next time…