What the new Sportsvision data could tell us about shooting percentage

It’s become accepted among the advanced stats community that shooting percentages, especially at the team level, can be influenced strongly by random variance, and given enough time, will eventually regress to the mean.

At the individual level, however, we know that player skill also plays a role in percentages (shooting percentages, at least; it’s still unclear whether or not players have an effect on save percentage). This makes intuitive sense; a shot taken by Steven Stamkos or Vladimir Tarasenko is going to be much more dangerous than one taken by Shawn Thornton or Ryan Reaves.

Just because there is an aspect of skill that influences player’s shooting percentages doesn’t mean that variance doesn’t still play a significant factor. For all forwards who have taken at least 150 shots in two consecutive seasons since 2006-2007, personal shooting percentage has an R^2 of 0.096.

SH per 1


Though that correlation is very weak, there are players who have been able to buck the trend and consistently post high (or low) shooting percentages, like the two shown in the graph below, courtesy of war-on-ice.com.

SH per 2

Steven Stamkos’s 40-game rolling average rarely drops below 10%, and consistently jumps over 20%. Sean Bergenheim’s 40-game rolling average rarely gets over 10%, and usually hovers between 5-10%.

So, with players, regression to the mean refers to a return to their personal mean, or the average number of shots that they convert on, separate from the rest of their teams. There can be other factors that affect personal shooting percentage, of course (being traded to Pittsburgh and getting to play with Sidney Crosby could result in an increase in shooting percentage), but overall, the percentages will regress.

With the NHL debuting new player tracking technology at the All-Star game this year, it’s assumed that there will eventually be tracking chips in every player and puck during every NHL game, providing data analysts tons of new information to work with.

This new information will surely help explain some of the random variance seen in shooting percentages, and with that in mind, here are some things that should be available thanks to the new data, and may end up being proven to result in higher personal shooting percentages for players.

Time and Space

This one makes intuitive sense, but can’t really be quantified until we have player-tracking technology.

We can reasonably assume that most players at the NHL level have a high quality shot, or at least a shot that can beat a goaltender about 30% of the time (think shootouts). For most of the action during a typical game, players don’t have tons of space to get shots off; opponents pressure them, or their shooting lanes are clogged.  When they manage to make space for themselves, however, (or when their opponents give it to them) their chances of scoring increase.


Mark Scheifele has all day to tee this one up from just inside the circle. With this much time to shoot, it’s expected that Scheifele picks his corner, and boy does he ever with this shot.  It’s a bullet, directly into the top corner. The poor goaltender had no chance.


Brent Seabrook has enough time to get control of the puck, take it to the middle of the ice, pick his spot on Niklas Backstrom, and then deliver a rocket to the back of the net. Thanks to the fact that there isn’t a defender pressuring him, Seabrook is able to keep his eyes focused on his target the entire time, and Backstrom really has no chance of stopping this one.


Mike Hoffman is able to walk in, pretty much inch by inch, and pick his spot behind James Reimer. Roman Polak simply backs off, and Hoffman has all day to set up, aim, and then fire off the rocket that finds the back of the net.

Shot speed

Though shot speed influencing shooting percentage also makes intuitive sense, it is also impossible to quantify with the data available right now, mainly because there isn’t a radar gun pointed at every shot, or a tracking chip placed in the pucks used during games.  It still can be seen, however, that players who consistently have high personal shooting percentages have reputations around the league as snipers, and have quick, accurate shots that are difficult for goalies to stop.


Vladimir Tarasenko is tearing up the league this season, posting 36 goals in 76 games, good for 4th among all skaters. He also sports a 13.8% shooting percentage in all situations this year, which is partly due to his blistering wrist shot. The puck just jumps off of his stick, and is practically past Jonathan Quick before the athletic netminder even has a chance to move.


Nikita Kucherov gets this puck up and over Jhonas Enroth’s shoulder from a close distance, before Enroth really has time to react. The impressive aspect of this shot isn’t that it was a rocket, but that it was elevated so quickly; even the announcer had to take a moment to state “Man, that got upstairs quickly.” Enroth is a smaller goalie, standing at 5’11. He didn’t really have a chance on this shot.


Alex Ovechkin simply powers this puck into the back of the net. The whipping motion of his stick launches the puck on goal, and James Reimer really has no chance as the puck flies through his five-hole and straight into the netting behind him.

When shot speed and time and space are combined, it makes for a lethal combination. Here’s some hockey eye candy, as Nikita Kucherov is being given tons of time and space to deliver this perfect shot to the back of the net.


Player speed

This one is closely related to shot speed, but it’s just a tad bit different. Sometimes, players are moving so fast before they take their shots that they pull the opposing goaltenders out of position, opening up holes to shoot for.


Erik Gudbranson is not known as a speedster, but here he flies past all four Islander defenders before slipping the puck past Jaroslav Halak. Halak clearly isn’t ready for this shot, and he’s out of position when it’s taken. Gudbranson’s speed catches Halak off guard, and as a result, it opens a hole near side for Gudbranson to put the puck into.


Brandon Saad is absolutely flying through the neutral zone on this play, splitting the defense and forcing Tobias Enstrom to turn in order to play him correctly. As a result, Michael Hutchinson is slightly out of position, and Saad’s shot finds the small gap that is left behind.


A section on player speed positively influencing shooting percentages would feel empty without an appearance by Sidney Crosby, whose speed and skill produces points and highlight reel goals night in and night out. Here, John Gibson gets pulled out of position as Crosby speeds through the Anaheim defense, and the puck finds it way through one of the open holes that are created by Gibson’s movement.

The shot quality debate isn’t going to die down any time soon. Though there is currently tons of random variance present, it is clear, at least on a personal level, that some type of shot quality does exist, and certain players can have an effect on their shooting percentages. Most of the data we currently have allows us to focus on shot location. The new information from the Sportsvision data will help focus on other factors besides shot location that can influence shooting percentages, and will hopefully help to further differentiate random variance and skill within the percentages.

One thought on “What the new Sportsvision data could tell us about shooting percentage

  1. Pingback: Breaking Down Theory Behind Advanced Stats: Percentages

Leave a Reply

Your email address will not be published. Required fields are marked *