clock menu more-arrow no yes

Filed under:

More on Score Effects and Corsi Numbers

New, comment

Earlier, I wrote about the effect that the score has on shot counts.  In a nutshell,when a team is down, it tends to outshoot its opponents.  This effect is not constant throughout the game - there's very little difference in shot counts for the first two periods, but in the 3rd period, shot ratios are very extreme:

Fenwick_vs_time_medium

So what happens if we adjust shot totals for home/road, score and time?  Surprisingly, not much - the standard deviation of the score effect adjustment is 0.45%.  Over the last five years, the 2008-09 Boston Bruins had the largest expected increase in shot ratio, and that was barely 1%:

 

Season Team Fenwick Raw Fenwick Norm Difference Reg Win%
2008-09 bos 50.3 51.3 1.064 646
2006-07 ott 51.0 52.0 0.956 585
2005-06 det 57.0 57.8 0.787 707
2009-10 was 51.4 52.1 0.720 659
2006-07 nsh 47.6 48.3 0.711 622
2006-07 sj 52.4 53.1 0.705 628
2005-06 cgy 51.2 51.9 0.697 561
2007-08 det 58.6 59.3 0.694 659
2008-09 sj 54.4 55.1 0.684 646
2008-09 was 54.6 55.3 0.678 610

 

At the other end, the adjustments are slightly more extreme - not surprisingly, because it's easier to put together a legendarily bad team than it is to build a comparably good one:

 

Season Team Fenwick Raw Fenwick Norm Difference Reg Win%
2006-07 edm 48.2 47.4 -0.812 390
2009-10 edm 45.2 44.4 -0.825 329
2007-08 atl 42.9 42.1 -0.830 415
2008-09 tb 47.4 46.6 -0.834 293
2005-06 stl 48.5 47.7 -0.837 256
2009-10 tor 53.1 52.3 -0.879 366
2008-09 col 48.1 47.2 -0.885 390
2006-07 la 50.3 49.4 -0.923 329
2006-07 phi 47.2 46.0 -1.221 268
2005-06 pit 45.0 43.7 -1.374 268

 

Again, this year's Toronto Maple Leafs disaster makes an appearance - bad goaltending and a strange unwillingness to play to the score when they were winning makes them the rare truly bad team that legitimately outshot their opponents.

Overall, I think we don't have to worry too much about score effects at the team/season level - even the best and worst teams over the last five seasons are unlikely to need more than a 1% adjustment.  Of course, score effects are incredibly significant at the game level and can not be ignored over a small number of shots.