:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/32321903/486753391.0.jpg)
Now let's look at Power Play Save Percentage. I am going to look at 2009-10 through 2013-14. I have shown repeatedly that different goalies are different at Even Strength. As always, in order to do regression analysis I constructed logits (ln odds of a save). I threw out the goalies whose logits were zero or undefined. All the regressions are weighted by number of shots faced. Lastfirst is the name of the goalie variable.
Goalies are all the same, teams are all the same. > LinearModel.7 = lm(PPLogit ~ lastfirst + Team, data=Goalies5, weights=PPSA) > anova(LinearModel.7) Analysis of Variance Table Response: PPLogit Df Sum Sq Mean Sq F value Pr(>F) lastfirst 107 1198.1 11.197 1.0260 0.4374 Team 28 416.7 14.882 1.3637 0.1200 Residuals 162 1767.9 10.913 It looks like there is a small correlation between ESSP and PKSP. ESSP explains about 17% of the variability seen in PKSP (I.e., a little). > cor.test(Goalies5$ESVPCT, Goalies5$PPSVPCT, alternative="two.sided", + method="pearson") Pearson's product-moment correlation data: Goalies5$ESVPCT and Goalies5$PPSVPCT t = 7.8363, df = 296, p-value = 8.393e-14 alternative hypothesis: true correlation is not equal to 0 95 percent confidence interval: 0.3157524 0.5043694 sample estimates: cor 0.4145025 Graphically, the relationship looks like: From the graph we can see that seven goalies at the low end are driving the apparent correlation. If we take out those seven, who faced at most 4 shots and a collective total of 18, we get > cor.test(PPminus$ESVPCT, PPminus$PPSVPCT, alternative="two.sided", + method="pearson") Pearson's product-moment correlation data: PPminus$ESVPCT and PPminus$PPSVPCT t = 0.8386, df = 289, p-value = 0.4024 alternative hypothesis: true correlation is not equal to 0 95 percent confidence interval: -0.06608471 0.16332683 sample estimates: cor 0.04927089 So, in reality, nothing there. Total randomness. Conclusions There is no team effect. There is no goalie effect. In general, PP Save Percentage is not related to ES Save Percentage. "Absence of proof is not proof of absence." From the ES data, we know that it takes several thousand shots to tell goalies apart. The busiest goalies in this data faced less than 300 shots over the 5 seasons. There may well be a goaltender effect. However, if it is far too small to see in 5 years of action and it certainly will not be visible in 1 year. If a 0.920 goaltender sees 50 PP shots in a season, the 95% Confidence Interval for observed PP Save Percentage is 0.808 to 0.978. Over the 5 years, goaltenders faced a total of 8862 shots and made 8063 saves for an overall PP Save Percentage of 0.90984.