February 9, 2009

Back-to-Back game statistics.

Mc79hockey had a post about handicapping the playoff race, and one of the indicators was back-to-back games. One poster commented that his analysis showed that back-to-back games resulted in 6% hit to the road team’s chances of winning (45% vs. 39%). I decided to further investigate this research and I looked at 5 season’s worth of data (ties excluded):
- In all games there were 2466 road wins in 5542 games played (44.5%)
- In the 1432 back-to-back road games there were 593 wins (41.4%)
- If you exclude the back-to-back road games you get (45.6%)
- Almost 25% of road games are back-to-back!
- A difference of about 4%

However, there was no discussion in regards to the rare home back-to-back games. For example Columbus has 6 remaining back-to-back games, but 3 of those are at home. What are those chances?
- In all games there were 3076 road wins in 5542 games played (55.5%)
- In the 612 back-to-back road games there were 332 wins (55%)
- If there is a home back-to-back effect it is much smaller

Now on further investigation of the road back-to-back games showed that the teams who are bad perform even worse than the above averages would suggest:
For example, if a team wins about 31% games the above results would suggest that they would win about 22% of the back-to-back games, but the result is probably closer to 18%. Or on the other end if a team would normally wins 70% of their games the average suggests that they would win about 61% of the back-to-back road games when in fact they’ve won about 66% (or essentially no road back-to-back effect). So the better the team is the less they are affected by back-to-back games.

I was too lazy to look at the result of the combined back-to-back games (both home and away teams are playing back-to-back, which is likely a large majority of the home back-to-back games).

Please note: when I say "back-to-back road games" or "back-to-back home games" I am referring to the second game of the set of two games. The first game is not included in the statistics.
Just a quick example:
For example Vancouver played two games one day apart on December 22, 2008 & December 23, 2008. (won first game, lost second). I am only counting and considering the results of the second game on December 23, 2008 (a loss).


Mogen David said...

Just to clarify. Are these the backend of the back to backs or either game (assuming that a back to back could be road-raod, road-home and home-road). I made the assumption that it was the backend and got called on it when I quoted your numbers.

JavaGeek said...

backend (or to clarify: the 2nd of two back-to-back games)