Zack Stortini, pictured above, took 23 shots in 2008-09. He scored six goals on those modest shot totals, good enough to lead the Oilers in shooting percentage at 26.1%.
Hockey or Die! (and Oilogosphere) commenter Ender took me to task yesterday for my post on shooting and save percentage. While he had a number of different points, the one that intrigued me was that team shot totals could fluctuate enough to compensate for some of the drastic shooting percentage changes we will see; in other words, a team like the Oilers (15.2 SH%) might drop back into the standard league range (say, 10%) but shoot enough that it wouldn’t matter. Intuitively, that made no sense to me, but however much I trust my intuition it doesn’t do much for anyone else.
Thus, I decided on a simple test: I’d break the Oilers season last year into 16 five-game chunks (ignoring games 81 and 82) and see how much shots and shooting percentage varied. For good measure, I included shots against and save percentage as well. Here’s the raw data:
- Games 1-5: +154/-143 shots; 7.79/.895 SH%/SV%
- Games 6-10: +136/-200 shots; 9.56/.920 SH%/SV%
- Games 11-15: +142/-148 shots; 11.97/.912 SH%/SV%
- Games 16-20: +121/-192 shots; 11.57/.927 SH%/SV%
- Games 21-25: +151/-146 shots; 9.27/.904 SH%/SV%
- Games 26-30: +127/-187 shots; 11.81/.920 SH%/SV%
- Games 31-35: +142/-160 shots; 7.75/.881 SH%/SV%
- Games 36-40: +138/-149 shots; 12.32/.913 SH%/SV%
- Games 41-45: +128/-156 shots; 10.94/.891 SH%/SV%
- Games 46-50: +158/-148 shots; 10.13/.926 SH%/SV%
- Games 51-55: +144/-154 shots; 6.94/.890 SH%/SV%
- Games 56-60: +139/-145 shots; 12.23/.903 SH%/SV%
- Games 61-65: +134/-175 shots; 10.45/.903 SH%/SV%
- Games 66-70: +144/-178 shots; 11.81/.904 SH%/SV%
- Games 71-75: +141/-150 shots; 5.67/.900 SH%/SV%
- Games 75-80: +132/-160 shots; 9.84/.925 SH%/SV%
Over these five game segments, shots varied on average by 5.4% from the mean. The total range from minimum to maximum was 26.7%. That contrasted sharply with shooting percentage, which on average varied by 16.5%, and had a range from minimum to maximum of 65.2%. Total goals was almost as bad, with a 14.4% average variance and a range from minimum to maximum of 64.0%.
In other words, shooting percentage varies by roughly 300% more than shot totals. I’d imagine we could narrow both down by focusing on, say, just home games or just road games but using this test it seems clear: when prognosticating, it’s a much better idea to look at the shot clock than at either shooting percentage or total goals scored.