# Predicting Goal Scoring Regression (Part 3)

Hello everybody and welcome in for the third and final article in this series. I’ve been taking some time to put together what I hope you’ll agree is some quality content and information that you likely have not encountered elsewhere. I’m trying to tackle the problem of goal scoring regression and learning what statistics may predict an impending decline in that area. If you want to check out the previous articles in this series you can find part 1 here and part 2 here. I began my analysis for this article with a hypothesis that players that have a significant shooting percentage (S%) jump from the previous season would regress in the following season. I went back five years and compiled the following list of players who scored more than 20 goals with a S% increase of 7% or more over the previous season:

I’m a bit hesitant about this metric because it is often tied to players who simply had an extremely low shooting percentage for no apparent reason the season prior. A low shooting percentage the previous year makes it very easy to have a high S% difference and earn a place on the list. Lee in 16/17, Aho in 18/19, and Kadri in 16/17 all fit the bill here. That being said, the average goals/60 minutes played decline in the following season is -0.36, or an average decline of -8.5 actual goals scored in the following season. These numbers fall in line with our previous two metrics and seem to be evidence of a pattern at the very least. As before, let’s apply these criteria to players in the 2019/20 season and see who gets this ignominious distinction:

Two players who jump out as obvious candidates to simply dismiss from analysis are Jaden Schwartz and William Nylander; both had ridiculously low shooting percentages in 2018/19 for players of their caliber and positive regression should have been expected for anyone who has watched either player in live game action for more than thirty seconds. Pageau also had a very low S% in 2018/19 at 4.82%, but his career average prior to last season was just 9.3%, meaning last season’s 17.2% still marks a major aberration for him and he should be considered an obvious regression candidate. Killorn, Schenn, and Zibanejad have made one or both of my previous two lists and need to be viewed as near locks for significant goal-scoring regression. A bonus player who just missed making this list: Jack Eichel, who posted a 6.6% shooting percentage jump and made an appearance in my first article in this series.

One may ask why this third analysis is necessary if it is not revealing additional players to be wary of in terms of goal scoring regression for the upcoming season; it’s a fair question. I believe there is some value here simply because S% is the most visible stat I have used in this series. For your own quick and dirty analysis, you can 1) look for a high S% in the 15%-plus range (which eliminates players who simply had an extremely low S% the previous season) 2) check that player’s previous season or career average to see if there is a significant difference of 7% or more and 3) reduce expectations of goal scoring if the player meets both criteria. Hopefully this is a useful, easy-to-use tool you can add to your arsenal when trying to determine if that big breakout season was an aberration or simply a sign of things to come.

If you’ve enjoyed this content I hope that you’ll take a minute and fill out my five-question survey here as I explore the possibility of doing this full-time. Make sure you follow Apples & Ginos on Twitter for more content and to ask any fantasy hockey questions you may have.

Thanks for reading, you are much appreciated!

NGN