So, there is some strange stuff happening in the NHL in recent years and I am not quite sure what to make of it. Let me explain.
This first chart shows the percentage of shot attempts that were blocked, missed the net or were on net. This is 5v5 regular season data only, as will everything in this post unless specified otherwise.
As you can see, shooters are far less likely to hit in 2024-25 than they were in 2021-22 or earlier. There was a slow rise in shots on net rate up until 2021-22 when it reached a high of 55.2% of shot attempts being on goal. However 3 seasons later that has dropped down t 46.6%. What’s interesting is that it seems that initially it was attributable more to defenders blocking shots more but the percent of blocked shots flattened out in 2024-25 and the drop in on goal percentage dropped primarily because shooters just missed the net more. It makes you wonder if defenders focused more on blocking shots which forced players to try to shoot around the blocking defensemen and then missing the net. Let’s explore more.
(Edit: I was reminded that starting in 2023-24 the NHL started tracking shots that were blocked by teammates in addition to shots blocked by opposing players. This would account for the increase in blocked shots the past two seasons, but this will not impact most of the analysis below which does not consider blocked shots).
League average save percentage has steadily dropping since 2015-16.
While the percentage of shot attempts on goal has dropped significantly in recent years, this hasn’t seemed to change a the fairly steady downward trend in save percentage. Shots on goal are still becoming more difficult to save and at a fairly steady rate.
However, if we look at what I will call Corsi Save % (saves / all shot attempts) we see something interesting, but maybe not surprising.
So, there was a downward trend in Corsi Save Percentage up until 2021-22 and in the past two seasons it has notched up again. This is not surprising when you consider that defensemen are blocking shots at a higher rate and shooters are missing the net at a higher rate.
The league average combined (both teams) shot attempts per 60 minutes of 5v5 ice time has fluctuated a bit but aside from 2020-21 (COVID interrupted season) it has been in the range of ~83-85.5 shots attempts per 60 minutes since 2016-17 season.
While writing my article on the Ottawa Senators improved defense, I noticed some trends in Ottawa’s shot locations that can explain Ottawa’s improved defense this season. Basically, the Senators have improved their defense by pushing the oppositions shots away from the middle of the ice. I wondered if there was anything with shot locations that might explain some of what we observe above. I developed a shot zone analytics tool to look at various data points based on the location the shots were taken from. I’ll be using some of that here so let me first explain some terminology.
I developed this loosely based on the shot zones that the NHL developed for their NHL Edge data with some modifications. The image below shows my zones and how I have labelled them. This is similar to the NHL Edge data but I decided to split ‘Zone 6’ into three separate zones based on angle to the net.
Due to the large number of zones I will define several larger zones as follows:
- Home Plate (HP) – Zones 4, 6a, 6b, 6c, 9, 10 and 11 combined.
- Non-Home Plate (Non-HP) – All non HP zones (1,2,3,5,7,8,12,13,14,15,16)
- Crease – Zone 4 only
- Center HP – Zones 6b and 10
- Left HP – Zones 6a and 9
- Right HP – Zones 6c and 11
First, lets look at blocks per 60 for the larger zones. The NHL records blocks as the location where the shot block occurred and not where the shot was taken.
It is clear that total shot block rates (top blue line) have risen dramatically the past two seasons. Ignore the dip in 2020-21 as this is partly due to the dip we saw above in shot attempt rates. The key takeaway is the jump in the past two seasons. So, what accounted for this jump? Well, both home plate and non home plate shot blocks rose, but what is interesting is that among the home plate shot blocks, it was left and right home plate zones that accounted for the rise and the center home plate zone actually fell off the past two seasons after a steady increase.
It is unfortunate that we cannot get an idea of where the shots were being taken from so we can’t calculate a shot blocked rate for all shot attempts taken from a particular zone. However, it is clear that shot blocks are happening more towards the outside rather than in the middle of the home plate zone.
We can look at unblocked shot attempts per 60 though which may tell us if teams are shooting more or less from specific zones.
There has been some mild fluctuations over time, but generally the takeaway here is that in the past two seasons there has been a drop of in unblocked shot attempts in the center home plate zone and a slight increase in non-home plate unblocked shot rate. If teams are taking fewer shots from the center home plate zone and more from the non-home plate zone then this explains why there are fewer blocked shots in the center home plate zone and more outside the home plate zone. Does this hold for shots on goal?
Yes, this shots on goal charts looks very similar to the unblocked shot attempt chart above.
We saw above that the percent of shots attempts that missed the net has risen. We can’t look at all shot attempts by zone since we don’t have the shot location for blocked shots, but we can looked at the percentage of unblocked shot attempts that missed the net wide or high by zone.
Wow, lots to unpack here. First, other than crease shots, every other shot zone is quite similar in terms of the percentage of unblocked shot attempts that missed the net high or wide. This is interesting because I might have thought that outside shots might have had a higher missed net rate than shots taken from more directly in front of the net because of the angle to the net makes the net ‘smaller’. This doesn’t really seem to be the case. While non-home plate shots do have a slightly higher chance of missing the net, center home plate shots miss the net at a slightly higher rate than left or right home plate shots do. In fact, center home plate missed net rates are closer to the non-home plate missed rate shots. Crease shots miss the net are much lower rates which is maybe expected. It’s hard to miss the net if you are shooting from 5′ in front of it.
The most interesting thing in this chart is that every shot zone has seen their missed rate increase the past few seasons. Non crease shots rose from a range of 25-29% up to 32-36%. Crease shots have increased much more dramatically. Crease shot missed net rates rose from a low of 5.8% in 2015-16 to 10.5% in 2021-22 and have nearly doubled since then rising to 21.1% in 2024-25. This kind of dramatic change is unheard of in hockey analytics where changes are typically slow, gradual and small. I am not sure what is happening here, but here are a couple of theories:
- It’s possible that some teams are starting to leverage player and puck tracking and actually are changing their playing style. I have my doubts that this explains the above because I am pretty certain that most teams do not yet have the analytics teams to do this yet.
- It has something to do with player and puck tracking that has been put in place over the past several seasons. It could be that manual trackers were tracking goalie stops as saves puck tracking now allows the NHL to determine that some percentage of them would have gone wide and are now being classified as missed shots, even though the goalie stopped the puck.
Option two makes the most sense but the NHL has had player and puck tracking data for four seasons now (NHL Edge data goes back to 2021-22 season), however I would have expected more of a step up rather than a steep incline. It is possible though that the NHL spend 2021-22 and 2022-23 refining their data collection processes and computational algorithms which led to an incline rather than a step. If that were the case then, for analytical purposes, 2020-21 and earlier should be treated as one dataset and 2023-24 and beyond as a second data set with 2021-22 and 2022-23 being transitional years. I look forward to seeing next seasons data to see if things stabilize or there are still changes being observed.
One more chart to look at, save percentage by zone.
Non Home plate save percentage has remained pretty stable (dropped 0.4 percentage points from 2015-16 to 2024-25) however the home place save percentage has dropped much more significantly (down 2.5 percentage points). Here is where things get a little interesting. The right home plate zone has dropped just 0.8 percentage points, while the center home plate zone has dropped 2.5 percentage points and the left home plate zone has dropped a whopping 4.8 percentage points.
In conclusion, I am not sure what to make the dramatic changes we have seen in the past few years. It is clear that the NHL’s shot data is changing, what I am less clear about is the why. I have difficulty believing that it is all due to changing playing styles as these changes are generally much smaller and occur much more gradually. More likely, we may be seeing is an NHL data problem during the transition years from manual shot tracking to more automated shot tracking using puck and player tracking data. We’ll be able to better see if this is in fact the case in another year or so when we can see whether data has largely stabilized again closer to the 2024-25 levels.
One final point on goalies and save percentage. We have seen that in the last couple seasons a much larger percentage of shots have been identified as missing the net than had previously been classified as having missed the net. These are missed shots that were previous allocated as saves (assuming this is due to tracking changes, not playing style changes). This means we should expect that goalie save percentage should drop as you are removing some saves, but not removing goals against at a similar ratio. So if we expect save percentage to drop due to changes in tracking the past couple seasons, but we only see that save percentage has dropped at a similar rate it did the 6 or 7 years prior, it might in fact mean that if not for the tracking changes we may have actually seen save percentages start to rise. After years of improved offenses and falling save percentages, is the defending side catching up and we can’t see it because it is hidden behind shot tracking changes? With that question to ponder, I’ll leave you this final chart on league average Fenwick save percentage (includes missed shots as saves).
