Vulgar Statistics: Goalies In Back To Back Games

This is a debate that rages every season, doubly so with Lindy Ruff continually mismanaging his goalies.  The conventional wisdom seems to favor splitting goaltenders for the two games, but do the stats back that up?  And if you do, is it better to play your best guy on the first night or the second?  Hop aboard the magic hockey bus, let’s find out.

I started with the Sabres of the previous three years.  I like going back three years for things like these because it’s a nice number, and because I think it’s reasonable to assume that a given player’s game doesn’t suffer much due to aging over that span.  These are the numbers:

• Same Goalie Both Nights – (36-14-6) — 3.13 gpg — 2.32 gapg
• Different Goalies – (23-34-9) — 2.55 gpg — 3.17 gapg
• Miller Both Nights – (28-12-4) — 3.16 gpg — 2.20 gapg

Granted that doesn’t tell too much of a story beyond how bad Patrick Lalime was.  Enroth brought the ‘backup’ numbers up a bit, but not nearly enough.  It was time to go league-wide, also over the past 3 years.

• Same Goalie Both Nights – (247-234-64) – (.512 pts pct) — 2.64 gpg — 2.84 gapg
• Different Goalies – (343-321-97) – (.514 pts pct) — 2.73 gpg — 3.00 gapg

What a grand crapshoot.  I then wondered if there was a breakdown between different goalie combinations.  Was it better to go Starter – Backup, or Backup – Starter?  Since this is a little more complicated, I scaled it back to just last season.

• Starter – Starter – (125-101-35) – (.546 pts pct) — 2.65 gpg — 2.65 gapg*
• Backup – Backup – (32-33-5) – (.493 pts pct) — 2.41 gpg — 3.01 gapg
• Starter – Backup – (137-129-45) – (.513 pts pct) — 2.61 gpg — 2.86 gapg*
• Backup – Starter – (79-73-17) – (.518 pts pct) — 2.80 gps — 3.10 gapg*

*Given that I looked at hundreds of scores over two days, I believe I missed one game in each of these, with no way to easily fix it.  This puts the margin of error at roughly +/- .001 pct. point, and +/- .01 goal, or pretty much negligible.

Okay, so it did matter, but only a little bit, and there are definitely teams that have studied this.  Most seem to think that their best bet is to win the first game with a rested team and a rested goalie, but a few (The New York Rangers) rely on their starting goalie to bail out a tired team and almost always start him second.  Given the atrocious 66-94-27 record and 3.10 GAA of a backup in the second game (whether they played one or both), perhaps this strategy merits some reconsideration.

Final Thoughts:

The best way to pick up points is to play your starter in both games, but it does depend on your personnel   Teams with two good goalies (Boston) have done very well splitting them almost every single time.  I think if I were a coach, my inclination would be to rest my starter during the week and give him both weekend games.