aka, Everything You Always Wanted To Know About 5-on-3s, But Didn't Know Where To Ask

I guess I shouldn't have been surprised, but as it happens, JavaGeek (the Hockey Numbers guy) had done some work on 5-on-3 situations. His table -- which I can't certify, but smells right -- is here, the magic numbers (for me) being the actual amount of time teams spent in 5-on-3 situations in the 2005/06 season.

I took that data (btw, thanks Chris) and the Team TOI by Situation I got from mc79hockey in the summer (btw, thanks Tyler), and went about assessing what kind of impact 2-man advantages had, how common they were, etc.

First, some figures, most of which are straight off of JavaGeek's table:

• There were a total of 1370 5-on-3 powerplays, an average of 46 per team. 2-man PPs per team ranged from a high of 62 to a low of 31; 2-man PKs ranged from a high of 62 to a low of 27.
• The average duration of a 5-on-3 was 48 seconds
• Ignoring duration (for the moment), the average success rate on 5-on-3 PPs (i.e. Goals divided by Opportunities) was 25.7%; team success rates ranged from 42.9% down to 16.0%. In terms of 3-on-5 PK, success ranged from 56.8% all the way up to 88.9%.

Times. There was a total of about 1100mins of 5-on-3 time, or about 5.2% of total PP time. Put another way, the mythical average NHL game broke down like this:

• Total length: 60:46
• Even strength: 42:09
• 1-man PP/PK: 17:22
• 2-man PP/PK: 0:54
• Empty net: 0:21

Scoring. This is what I was really interested in.
The first two columns are rates per 60 minutes, the 3rd is the difference between the two, and the 4th is the sum. 5-on-3s account for 5.2% of all PP time, and 13.9% of all PP goals.

Clearly the advantage conferred by a 2-man advantage is a massive one, but I'm actually surprised at how small the effect is on the overall PP figures (as always, your mileage may vary). It's good to finally know that the average mark for 5-on-3 scoring is about 1 goal every 3 minutes. Also, congrats to the Leafs: that "0.05GAA" is entirely due to the only 3-on-5 goal scored the entire season, by Mike Richards (IIRC) against Toronto.

For Gary Bettman fans, I think the 4th column is the most interesting one. Relative to ES, putting a team on the PP increases the chance that a goal will be scored by 35%; putting a team on a 2-man PP increases the chance that a goal will be scored by 273%. Hypothetical: if you ran the NHL and wanted to see Total Scoring increase, would you be able to use this information effectively?

If there's anything else you want to know about this, drop questions in the Comments. My head hurts a bit from the spreadsheet, so I may have forgotten to put all of the relevant (or interesting) data in the post.