Last month, Jonathan Garcia and I attempted to find out if Pekka Rinne is better when he faces more shots in a game. We concluded that shot totals have no effect on Rinne's save percentage. Jen LC at The Committed Indian came to a similar conclusion with Corey Crawford. Matt Cane at Puck++ found an entirely different conclusion when he looked at league-wide data however. He found that even strength save percentage improves as shot totals increase in the short run. An expanded analysis from the Rinne post gives us a solid conclusion among all this noise.
Most investigations into the relation of shot totals and save percentage use all starts including games in which a goalie was pulled. I've always felt that including pulled starts significantly confounds the data, which is why Jon and I only included games in which Rinne played 55 minutes or greater for our initial analysis. Put another way, the data is altered by including a start in which a goalie faced 5 shots and let in 3 goals but was pulled after 10 minutes. There is no way of knowing how he would have finished the game.
So, an analysis including only starts 45 minutes and greater of 5on5 time should increase the reliability of the data since early exits are excluded while 5on5 data will exclude any special teams situations. In fact, we can go one step further and only include goalies who have faced at least 3,000 5on5 shots over the past 4 season. This way, we can analyze goalies who we know are (or were) established. In total, the analysis includes 30 goalies and over 3,800 goalie appearances.
We end up with a fairly even distribution of per game shot totals when using these parameters. Shot totals lean a little to the high side, but the difference shouldn't change our conclusion significantly.
The question we are looking to answer now is "do higher shot totals correlate with higher save percentages among goalies who faced over 3,000 5on5 shots over the past 4 seasons, in games where they played 45 minutes or greater of 5on5 time?" The reason we segment the data so much is because we want to investigate the relationship between shot totals and save percentages during "normal" circumstances.
The r-squared value will hold the answer of whether there is any correlation between shot totals and save percentage. R-squared can only be a positive value between 0 and 1. The closer to 0, the weaker the correlation. Conversely, the closer the value is to 1, the stronger the correlation. Clare Austin reminded us that save percentage is not linear on our Rinne post so, this time around, the analysis utilizes a logarithmic regression to estimate the relationship.
With an r-squared of 0.03639, we can see there is essentially no correlation between shot totals and save percentage during 5on5 play. The answer to our question is a resounding "NO".
An analysis focusing on only 5on5 situations shows that, in relatively full appearances, there is no correlation between save percentage and shot totals. When it comes down to it, there is no reason to believe goalies will be more effective if they face higher shot totals.