Is a team’s shooting percentage considerably affected during the time following a saved shot? As we know, not all shots are equal. Aside from location, shots occurring in rapid succession appear to be more problematic for a goalie. The defensemen cannot adequately reset from their previous shape to a one that will best defend the next potential shot.
Unlucky rebounds are the prime reason why we would think shooting percentage is higher immediately after a save. Goalies and defensemen may both be out of position while an offensive player picks up the puck with a great scoring chance. The logic seems reasonable. So let’s test this hypothesis so we will know to what extent these situations impact scoring.
I looked at all shots for the past three regular seasons and recorded how many seconds elapsed until the next shot, and also whether or not it was a goal. I reset the counter after each same-team shot, if the other team recorded a shot, or if the whistle blew. All “initial” shots to kick off the counter do not appear in this data since we’re only looking at the time after a shot (However, I did look at these shots and not surprisingly, goals were scored at the league average shooting percentage).
Here are the results:
As expected we see a large spike in the following few seconds after the initial shot. Shots that were recorded at the same time as the initial shot went in 41.4% of the time. This makes sense. The shots happened in such rapid succession that they occurred during the same second. It could also be scorekeeping error. I do not know how often they go back to look at the time on the clock for every shot, especially if there is a flurry of activity in a short amount of time.
The other surprising part about the beginning of the graph is that shots occurring 2 seconds after the initial shot converted at a higher percentage than shots occurring 1 second after. It’s a little puzzling since 0-second shots have such a high Sh% value, and common sense would suggest that as more time elapses, the less quality of a shot you’d have on offense. There could also be some missing spillover from the shots that were recorded during the same second.
We do see Sh% bottom out at 7 seconds with a 7.0% conversion rate during even strength play and 7.4% conversion rate during all play. That is also a curious finding. You have a lower chance of scoring a goal on a shot given that you had a shot saved 7 seconds prior, than if you were to randomly pick any other random shot during gameplay.
Lastly we see that as time goes on past 7 seconds, Sh% increases. I would have imagined that Sh% would level out after some amount of time following a save, so this is also interesting. One explanation is that this could be due to selective sampling. For instance, we recorded 4132 shots that occurred 1 second after an initial shot, but only 260 shots occurring 59 seconds after an initial shot. Perhaps if we combed through more data, these percentages would level out.
Also, keep in mind the way I programmed it is that the only way a shot counter will reset after an initial shot is if the other team gets a shot on net, or if the whistle blows. As the chart moves to the right, we are seeing instances where a team gets a shot off, but then keeps the shift going and also does not allow the other team to shoot during that time span. I first considered all situations and saw these results thinking that the numbers were skewed by teams on the power play or empty nets, but we see the same trend for EV situations.
Now that we can quantify the effects of allowing a follow up shot soon after a save, we can see which teams are giving up an abundance of rebound goals and also which goalies are suffering from seeing a lot of shots in succession.