How can the Blackhawks be 24th in expected goals but 5th in actual goals?

Expected Goals takes this information into account by looking at all unblocked shot attempts:

Shot type (Wrist shot, slap shot, deflection, etc.)
• Shot distance (Adjusted distance from net)
• Shot angle (Angle in absolute degrees from the central line normal to the goal line)
• Rebounds (Whether or not the shot was a rebound)
• Rush shots (Whether or not the shot was a rush shot)
• Strength state (Power Play, 5v5, etc)

Using this information, it looks at how dangerous every shot that took place on the ice. It doesn’t account for player talent level. For example a Steven Stamkos shot from the faceoff dot is equivalent to a Nick Schmaltz shot from the faceoff dot.

I should note that there are other models that take in account a players shooting talent, but the one on Corsica.Hockey, which i use, does not.

Back on topic.

So we have basically an outline of where the shots are coming from on the ice. Its almost like a heat map presented in numerical form. Someone with great expected goals is going to have a heat map that shows them taking a bunch of shots closer to the net offensively, while defensively they keep the opponent to the outside.

Expected Goals is closely related to scoring chances, but more in depth.

Corsi, or shot attempts, ranks every shot attempt that occurs on the ice as the same. 1 shot equals 1 shot no matter where it occurs on the ice. It would be similar to looking at hits in baseball.
Scoring Chances go more in depth. It groups shots that occur closer to net as higher rated. But it still groups chances together. A higher danger chance is one chance. All the high danger chances are grouped together. All medium danger chances are grouped together. All low danger chances are grouped together. To return to our baseball analogy, instead of looking at only hits, we are looking at hits that land in the infield and hits that land in the outfield. We are more in depth than just hits, but we are still grouping hits together, A hit off the wall and a blooper over the second baseman’s head are both rated as one outfield hit.

eXpected Goals gives a rating to every unblocked shot attempt based on the percentage of that shot going in. Adding all these ratings together, we get expected goals for and against. In baseball, this would be like analyzing the exact location that every hit landed and giving each hit a rating based on that information. It would be a more in depth stat than hits, and a more in depth stat that hits that land in the outfield vs the infield. Likewise, Scoring Chances are more in depth than Corsi. Expected Goals are more in depth than Scoring Chances.

So how can the Blackhawks be 24th in expected goals but 5th in actual goals?

One, expected goals doesn’t take goaltending into account. It simply looks at how likely a shot should go in against league average goaltending. Corey Crawford and Scott Darling have been amazing this season and are covering up for some of Chicago’s poor play.

The expected goals stat says opponents are taking many high danger shots against Chicago. It doesn’t account for the fact that Chicago’s goaltending has been incredible. So this is one of the reasons why we find a discrepancy between the two stats.
Secondly, I already mentioned that the Corsica model doesn’t take player talent into account. Players like Panarin and Kane take more low danger shots, but they are so talented, that they can still score. They can get medium range shots past the goalie better than most players. Players like Kane and Panarin will outperform their expected goals numbers.

For the most part its goaltending, but it’s also the talent level of the Blackhawks that has them outperforming their expected goals numbers.

It’s still alarming though. In their Cup seasons, they were good at expected goals. It’s unlikely Crawford and Darling will be amazing for the whole season. Likewise Kane and Panarin can’t carry the team. Luckily, the team has been playing better lately with the reemergence of Jonathan Toews.

Remember how i said expected goals was like a numerical heat map of shot locations? Let’s take a look at an actual Blackhawks shot heat map via HockeyViz.com so I can show you what I’m talking about.

Shot For (Offense)

screen-shot-2017-02-26-at-10-15-34-am

In the above heat map, the net is at the top of the graphic with the blue line near the bottom.

Blue = below league average in shots from this area. Basically they don’t shoot much from these areas.

Red = above league average number of shots. They do shoot often from these areas.

We see that Chicago doesn’t take many shots from directly in front of the net. They shoot the most from the point and the left circle.

Shots Against (Defense)

screen-shot-2017-02-26-at-10-15-47-am

The net is at the bottom of this graphic with the blue line at the top. Red again means a lot of shots happen in this area. Blue means not that many shots occur here.

Ah, we see the Blackhawks really struggle here as they give up more shots right in front of the goalie than most teams do. Shots in the slot are usually high danger chances and the above graphic shows that Chicago gives up many shots in and around the slot or goalie area.

So when looking at expected goals for % (which is expected goals for vs expected goals against) think of a heat map. For a poor expected goals team like Chicago, think of the above heat map.

Lastly, I guess I should show you the heat map of a good expected goals team. Let’s use the Stanley Cup champion Pittsburgh Penguins who are fourth in xGF%

Shot For (Offense)

screen-shot-2017-02-26-at-10-27-22-am

Shots Against (Defense)

Screen Shot 2017-02-26 at 10.27.31 AM.png

By glancing at the heat maps, Pittsburgh looks like an amazingly gifted offensive team with okay defense.

Further reading:

The Problem with Scoring Chances by @LoserPoints