Your eyes are lying about soft goals in hockey (Puck Momalytics)

(Jen Lute Costella is our new analytics writer, breaking down the fanciest of stats for you each week. She's a mom. She's writing for Puck Daddy. Hence, she calls this slice of stats heaven Puck Momalytics.)

If you have ever watched a hockey game, you have heard about a goalie giving up a softy. Soft goals make fans very upset and are often the things that stand out in our memory of games. The goalie may make 15 spectacular saves on high quality scoring chances but that shot he should have had burns into our brains and colors his entire performance. 

Once analysts pick up the “soft goal” thread and weave it into their discussion of a goalie, people cling to it and confirmation bias leads them to look for the soft goals. Goalie reputations are built and laid waste to in such a manner. It can be a tough label to shed because confirmation bias loves, well, confirmation. If two seasons ago in the playoffs a goalie gave up a few softies, but since then has given up very few of them, it doesn’t matter.

The moment he gives up another one, what we think is true is confirmed and who doesn’t love to be right?

So what is a soft goal?

Often, a shot from the perimeter that finds its way to the back of the net is considered a soft goal. If the goalie is completely screened by an opposing player or some sort of deflection happens on the way to the net, we are usually more forgiving of the goal. If that perimeter shot doesn’t deflect or the goalie’s vision is not obviously completely obstructed, the goals are labeled as soft. Our understanding of what “soft goals” are goes to the very heart of what we think of as the “eye test” in hockey.

The stats versus eye test debate has grown tiresome for many hockey fans. It really boils down to this: Sometimes, our eyes lie to us.

We have all learned about hockey in various ways, but a lot of our understanding of how the game works is from watching the game. It’s understandable that many people would feel like their knowledge of hockey is under attack by those promoting statistical analysis. If you have learned the game by playing it and/or watching it for years, your sense of pride can feel threatened by people boiling all of that down to a set of numbers. The thing is, that’s really not what stats are intended to do.

As fans or players of the game, we are pretty adept at picking up on generalities and forming a common knowledge base. We can fairly easily identify things like the overall skill of a player, speed on breakaways, whether the team is playing sloppy or sharp and the like. If we want to get into a deeper analysis of how a team or player is performing, statistics are a necessity. This doesn’t mean that you aren’t a real hockey fan if you choose to ignore the stats, but it does mean that you are probably missing some information to form your opinions. Humans have confirmation bias. This often skews what we look for and what we pay attention to when watching hockey.

The speed and fluidity of hockey results in so much data streaming through our eyes and into our brains that we simply cannot possibly process all of it accurately. This is the basic reason for keeping stats in any sport, not even just hockey. It’s really hard to remember what a player did in a few games at the start of the season when the playoffs are starting. So while the eye test certainly has value, it is not reliable for long term analysis of large amounts of input or data. Sports analysts track statistics to have access to reliable information free of the biases that we as humans innately carry with us.

We have constructed a common knowledge base as hockey fans that allows us to recognize things like soft goals. Whether or not we actively realize it, we know that shots from certain areas of the ice carry a greater likelihood of scoring than others. Everyone has experienced that moment of holding their breath when there is a scramble in the goal mouth only to exhale in relief because the puck stayed out or to give an exasperated sigh (or curse word depending on how you roll) because the puck went in. Most hockey fans aren’t holding their breath when a forward skates the puck into the zone and slaps a shot at the net from the boards. Our experience has programmed us to expect that those shots will be stopped. When those shots go in, our brains simply cannot fathom how the goalie let that happen and those moments stick with us.

One of the newer developments, at least in publicly available stats, is the ability to break down a goalie’s save percentage based upon shot location. Shot location data in the NHL can be troublesome, because frankly, sometimes the people recording it just get it wrong. Luckily, with enough shot data, statisticians can build in some corrections for these errors and give us a reliable notion of a goalie’s save percentage against shots coming from different areas of the ice.

To illustrate how these breakdowns can be useful in analyzing a goalie’s strengths and weaknesses, I pulled the numbers for goalies facing at least 4000 shots during 5 on 5 play from the start of the 2010-11 season to the present from stats site war-on-ice.com. The larger the sample of shots against we use, the more reliable the information is with regard to statistics so this is why I’ve made the cutoff for the group of goalies to look at so high.

At war-on-ice.com, they separate the ice into three sections, High sh% areas, Medium sh% areas and Low sh% areas.

The area in blue/green directly in front of the net represents the High shooting percentage zone. The red area that forms what looks like an arrow toward the net is the Medium shooting percentage zone. The remaining areas in yellow, mainly around the perimeter, represent the Low shooting percentage zone. Keeping this in mind, I calculated what percentage of the total shots faced came from each area to give some context to the numbers.

The markers in yellow show how tightly grouped this selection of goalies is in terms of Sv% for Low sh% shots despite having some variability in the percentage of shots faced from those areas of the ice. The orange markers show a little more of a spread in Sv% against shots from the Medium sh% area. The red markers show the largest variation in Sv% on shots from the High sh% area.

Looking at the Low sh% area graph, we see that comparatively speaking, goalies tend to face the largest percentage of shots from these areas. The percentage of shots faced from the specific area runs from left to right at the bottom of the graph. Save percentage against shots from these areas runs vertically. Of the goalies having faced at least 4000 shots at 5v5 play since 2010, we see that Henrik Lundqvist (NYR), Jonas Hiller (CGY) and Ryan Miller (VAN) have seen the lowest percentage of their total shots against from these areas, while Tuukka Rask (BOS), Pekka Rinne (NSH), Antti Niemi (SJS) and Steve Mason (PHI) have the greatest percentage of shots coming from the Low sh% areas.

The goalies with the highest Sv% against shots from these perimeter areas over the selected time frame are Marc-Andre Fleury (PIT), Tim Thomas (BOS/FLA/DAL) and Carey Price (MTL).  That may be a little surprising given our preconceived notions about some of these goalies. It is important to note that the Sv% scale (vertical) is very small, only covering from 96.6% to 98.0% because goalies tend to have a lot of success in stopping these shots.

The fact that the Sv% on shots from the perimeter areas is much higher than what we usually see in whole Sv% numbers again reinforces what our common knowledge tells us. Goalies are far more likely to stop these shots and thus we often call them soft goals when they find the back of the net.

The Medium sh% graph above tells a different story. Again, the percentage of the total shots faced by these goalies from the Medium sh% area goes from left to right along the bottom. These percentages are quite a bit lower than those we saw on the Low sh% graph. The Sv% on these shots is also quite a bit lower by comparison.

Ryan Miller has faced the greatest percentage of shots against from this area followed by Roberto Luongo (FLA) and Ondrej Pavelec (WPG). Craig Anderson (OTT) has the highest Sv% of this group followed by Lundqvist and Hiller, while Niemi registered the lowest with a Sv% of 90.01% against these shots.

The High sh% graph is far different than the others. The first thing that jumps out about this is that the spread of the values for Sv% is much wider and much lower than the others. The percentage of shots faced from these areas is also the lowest, which is logical considering it is a small area of the ice and often is very difficult for skaters on the offensive attack to get to in the first place. Lundqvist has had the highest percentage of the total shots he has faced come from this area. He also has the second highest Sv% in these situations. Tuukka Rask claims the crown for the highest Sv% against these shots.

The group of goalies near or above the 84% Sv% mark on this graph is pretty impressive. There is a wealth of Stanley Cup winners and finalists and Vezina winners and finalists in this group. It may be that our opinions about goalies would be more accurate if we considered the saves against these higher percentage shots a bit more heavily in forming them. Food for thought.

Just to illustrate my point about the variability of a goalie’s Sv% in small sample sizes, below is a graph of the Sv% and S60 (Shots against per 60) for many starting goalies around the league this season.

All of the numbers above are at 5 on 5 play this season. From left to right is the shots faced rate (SA60), from which we can see Darcy Keumper (MIN) facing the lowest shots against rate. Anderson (OTT) and Jonas Enroth (BUF) have faced the highest shots against rates to start the season. So far the highest Sv% at 5v5 among the regular starters have come from Roberto Luongo (FLA), Pekka Rinne (NSH), Corey Crawford (CHI), Jonathan Quick (LAK), Craig Anderson (OTT) and Frederik Anderson (ANA).

It is expected that these will regress a bit as the season goes on, but it is interesting to keep an eye, or rather, statistics on.

*All data herein collected from war-on-ice.com