Expected Goals, Actual Goals, and the Early Days of the 2022-23 New Jersey Devils

It has been a stressful start to the 2022-23 New Jersey Devils season. I am writing this prior to Thursday’s game in Long Island. Already we have seen both goaltenders get lit up for four goals in their first two starts, Miles Wood declaring he’s “sick and tired” of losing after the first game of the season, and fans chanting “Fire Lindy” in the third period of the home opener and after an 0-2 first period against Anaheim on Tuesday. There has been more than enough and justified teeth gnashing about the Devils’ performances - even in the win over Anaheim. This is the result of the past two seasons being bad and two offseasons of thinking it would get better - without the team fully delivering.

Among the discourse, the topic of expected goals have come up. I believe this is mostly due to them being espoused by some - including one play by play announcer, Bill Spaulding. That is, the Devils have been fantastic in terms of expected goals for (xGF) and expected goals against (xGA). Which is little solace when the Devils concede four actual goals in one period to Detroit in a 2-5 loss. This has led to some of the usual, tired arguments about how “analytics don’t matter” which is surely news to the 32 NHL teams that employ people who track and report on them and at levels beyond on what is publicly available. Others, myself included, lament the goaltending and defensive mishaps - even if the numbers suggest the Devils’ defense is actually fine. Rather than letting it cycle out, let us dive into the topics and possibly learn something about this season’s team.

Expected Goals: A Refresher

Expected goals, as a stat, is a result of a statistical model based on NHL data that the game’s scorer collects. There are multiple models with their own minute differences, but in general an expected goal model determines how likely it is for a shot to be a goal based on its shot location, shot type, and other available data. The biggest driver of xG for a player or a team is the shot location. You ever heard of the cliche, “Good things happen when you go to the net?” Expected goals pretty much justify that cliche as fact. The closer you are to the net, the more likely you are going to score. This makes sense. A shot in the slot is more dangerous than a 50-footer from the right or left points.

However, shot location is only one important factor in what makes a shot a goal. The execution - the finishing, if you will - matters. The pre-shot puck movement matters; a shot by itself in a spot is different than a one-timer from that same spot. Whether the shot is contested by the opposition matters. Whether the goaltender is prepared for the shot matters. The expected goal model does not consider those things. Mostly because they are not publicly tracked. This is how a team can rack up a high xGF and end up with fewer goals. Or put up a low xGA and end up with more goals allowed.

Those factors may be more apparent to those who watch the game because you can recognize those factors. I think that is where the divide in the discourse is in. The model does not account for them - they cannot - and so it is dismissed despite what it does represent. Memories are not at all trustworthy and the resources to go through the tape for every shot taken is not always available. Over a long enough time, a high xGF or xGA will do a decent enough job to tell us whether a team or a player is putting in good performances or not. For a single game or a few games into a season, it can be easily to misinterpret if someone just brings up a xGF or xGA value and leaves it at that. A little more investigation is needed.

Expected & Actual Goals Short Look at the Offense

The 2022-23 New Jersey Devils certainly warrant it in their first three games. In 5-on-5 play at Natural Stat Trick, they put up a xGF of 2.31 in Philly, a 3.65 against Detroit, and a 3.27 against Anaheim. Those last two are exceedingly good. The Devils only broke the 3 xGF mark six times last season (and went 5-1 in those games) and their high was 3.22. The Devils bested their best mark from last season twice already. Among all games played in the NHL this season before Thursday’s games, the Devils have two of the top ten highest xGF values in the league in 5-on-5 play. Even when more games are played, the Devils legitimately put in an impressive array of offensive attempts in their first two home games.

What does this mean? A couple of things. One: It means the Devils have absolutely crushed Detroit’s defenses and Anaheim’s defenses. They racked up plenty of scoring chances and especially high danger chances in both games. (Aside: I have to look at this later, but getting more than 12 HDCF in 5-on-5 appears to be exceptional. Devils had 16 and 18 in their last two home games) They were not at all stuck on the perimeter and firing hopeful pucks towards the net. I

Two: This also justifies the frustration with the lack of goal scoring, especially against Detroit. Two goals out of a 3.65 xGF? How can that be? Likely due to finishing, the goalie playing really well (Alex Nedeljkovic did play well), and puck luck. The best I can suggest is to just keep it up. The Devils are not going to score more goals by shooting fewer pucks. And with as many high danger chances they have made so far, it is not like they can do even more. Fortunately, they did that against Anaheim and the Devils did get rewarded for their efforts in the game. The goals by Ondrej Palat and Nico Hischier were absolutely high danger goals. The scores by Dougie Hamilton - a great example of how pre-shot movement is deadly for a defense - and Dawson Mercer were at least in the “house” for scoring chances. Hopefully, it leads to more efforts so more can be scored. Frustration is as much as an output as a load of goals scored.

Three: I would hold off before anointing the 2022-23 Devils offense as a machine. They played three teams that did not make the playoffs last season and likely will not this season. Those two impressive xGF marks were at home, too. The Devils will be traveling more and playing more difficult competition very soon. If they are still out-performing their opposition by the model, then we can say the Devils’ process has been potent. Whether it leads to actual goals, well, those other factors have to follow. I think they will. We shall see in time.

Now, that’s all about xGF, what about the xGA?

Expected & Actual Goals Against - with Pictures

The thing about hockey is that if you have the puck, you’re creating chances, and you’re shooting the puck a lot, then it means the other team does not. Sure, counter-attacks do not need a lot of time to happen and they can with a miss, a block, and/or just a good play by the defense. For the most part, though, flooding the opposition on offense in 5-on-5 means the other is pinned back. At least in terms of expected goals, the Devils have demonstrated that they can have a good defense with a constant offense. In 5-on-5 play, the Devils’ out did Philly 2.31 xGF to 1.29 xGA; out-did Detroit 3.65 xGF to 1.02 xGA; and out-did Anaheim 3.27 xGF to 1.3 xGA per Natural Stat Trick. Similar to their xGF values, the Devils’ xGA values are among the lowest among all games played in the league so far this season prior to Thursday’s games. And they would have ranked among the best the Devils put up in terms of xGA from their games last season. They put up an xGA fewer than 1.3 just five times, although they won just two of those games.

Of course, the actual goals tell a totally different story. They allowed 3 goals in 5-on-5 at Philly in a 2-5 loss, they allowed 4 goals in 5-on-5 at home to Detroit in a 2-5 loss, and they allowed 2 goals in 5-on-5 at home to Anaheim in a 4-2 win. Nine goals allowed in three games? More than a combined 3.61 xGA? Seems like goaltending remains bad in New Jersey. Some have looked at the difference in values and concluded this. It must be largely the fault of Mackenzie Blackwood and Vitek Vanecek. Another season where a difficult save is seemingly hard to come by in Newark. However, those who have attended or watched the games point out defensive miscues involved with those same goals allowed. How can a defense be seen good with the kind of failures that lead to these goals against the Devils?

I myself plant a foot in both camps because I think they are both correct to a degree. However, I cannot show that with stats alone. It is early enough in the season where it is easy enough to look at the goals allowed and see what went wrong. We can see how a low xGA from a team in three games can give up more goals than it suggests.

Game #1 - Devils at Philadelphia

All pictures for Game #1 come from this highlight video from the NHL’s Youtube page.

The first goal against was this put back by Wade Allison on a shift right after Alexander Holtz - You remember him, right? Young guy. Scored a goal and then exiled to the fourth line in the next game - scored a goal. Here’s a key moment on the play:

NHL Highlight Video

Right after a defensive zone faceoff loss by Michael McLeod, the play goes back to the points. Ivan Provorov gets it from Tony DeAngelo and takes a seemingly harmless shot from the boards. McLeod contests him. Severson picks up Noah Cates in front. Everything looks fine here. The eventual goal scorer, Wade Allison, is all alone to Blackwood’s left. I have no idea what or who Ryan Graves thinks he is covering. He’s just hanging out in the slot.

While it appears Blackwood is screened, he easily stops the shot. He did not catch it or hold it though. He gave up a juicy rebound to his left, which makes it easy for Allison to score.

Was this a bad goal to allow? Yes. Blackwood should have frozen the shot. Or at least not send the rebound away. Was there a miscue by the Devils skaters? Yes. Graves covering air seems like a miscue to me.

The second goal was scored by Travis Konecny from a bit of distance. A play by Kevin Hayes at the blueline set him up to go one on one with Brendan Smith. Here is the moment of the shot:

NHL Highlight Video

Smith is actually doing the right thing here by trying to contest the shot. He cannot step up to Konecny unless he wants to risk being burned by him. He cannot drop deeper and give up more space to Konecny. Keeping him to the high slot is not a bad move and his stick is out to at least challenge the shot. Unfortunately for him, Konecny is going to fire a shot between Smith’s legs towards the left (Blackwood’s left) post. One could argue Blackwood is being screened but the shot is also from well over 30 feet away (40 feet?) and Blackwood sees the shot enough to stretch to his left. The shot just beat him.

Was this a bad goal to allow? Yes. Was there a miscue by the Devils skaters? Not really. Seeing Graves lose a battle to Hayes did not help. Seeing Tomas Tatar too inside on the play was noted by the broadcaster. But this was a shot Blackwood could have done better on. Simple as.

The third goal was a Lindy Ruff special. What’s worse than one goal allowed? Giving up a second right afterwards! And this one is a doozy.

NHL Highlight Video

A missed pass by Tanner Laczynski goes to the far boards. James van Reimsdyk is first to the puck and tosses it around the corner to behind the net, where Laczynski recovers. Marino had Laczynski initially, goes behind the net with him, and then moves up to the goal line as Smith tries to engage with Laczynski behind the net. This is an overload or a swarm. Marino is positioned such that Laczynski should not be able to come out by the net and Smith is actively engaging him. In theory, this should be fine as long as the forwards cover what is going on behind them. And if Marino fills a passing lane to whatever is happening behind them.

Both do not happen. Ondrej Palat is covering nobody in the slot. Mercer is turning after van Reimsdyk’s play. Morgan Frost is in the right circle entirely wide open. And Marino is not denying a passing lane that way. Full credit to Laczynski for making a not-so-easy pass right to the tape of Frost’s stick. Frost got all of the puck on a one-timer (nice half-hearted effort from Palat after the pass was made) and hammered it in past Blackwood.

Was this a bad goal to allow? Well, not for the goalie! Blackwood was hung out to dry here. Was there a miscue by the Devils skaters? Yeah! Marino did not deny a passing lane out that Laczynski took advantage of. Worse, no one had Frost! Nobody! Palat was protecting an empty slot! Of all three goals, this was the most enraging to me because this one was absolutely on the skaters. Two Flyers were in the picture compared to three Devils and somehow that pass and that shot was made for a goal against. If the Devils are overloading a lot more, then a play like this may be bound to happen again - and I will not like it.

Game #2 - Devils vs. Detroit

From the Philly game, I have two bad goals and two goals where I fault skaters. What about that Detroit game with the infamous second period of four goals allowed? Dylan Larkin’s last second goal was a 4-on-4 situation, so these will be pictures of the other ones. All pictures for Game #2 come from this highlight video from the NHL’s Youtube page.

NHL Highlight Video

The first of Detroit’s four-spot in the second period was a classic “seeing eye” shot. A blocked shot on Dylan Larkin went out to the right point. Ben Chiarot figures to just shoot his shot. You can see that it’s not the best idea in the world. Where’s the shooting lane? Graves and Severson are right in front. Mercer is near the lane. Vanecek is standing up because, well, he can’t see anything if he is in a stance. Graves is absolutely in his way here. But Chiarot lets it fly anyway. By the time Vanecek goes down into a form, it is too late. The puck is heading to get inside the left post - and it does somehow.

Out of all of the goals, this one may be the most unfortunate. Was this a bad goal to allow? Well, Vanecek did not even see it until it was too late. Was there a miscue by the Devils skaters? Not really. It’s not like Graves or Severson were horribly out of position. This was a reaction to a play after a block. Normally, this one can be chalked up to bad luck and move on. But, as you know, it did not end here.

NHL Highlight Video

The play begins with Chiarot rimming a puck out from behind his goal line towards the left point. Then Chiarot took a hit. Graves could not keep the puck in play and Jakub Vrana chipped it past him. It is a 2-on-1 situation with Severson back. Vrana makes a quick pass to Lucas Raymond, who passes it back on for the zone-entry. Here, you can see two things of note. One, Nico Hischier amazingly got back to at least impede Raymond. Two, Damon Severson amazingly does not know what to do in a 2-on-1. He stayed in the middle on the entry and only here does he try to get close to Vrana. You know, while Hischier caught up to Raymond. Severson is not even on his skate blades here as Vrana loaded up his shot at the dot. Vrana’s shot beats Vanecek short-side and high. He saw it all the way and it torched him. This made it 1-2 and the fans did not like it.

Was this a bad goal to allow? I would say so. Getting beaten short-side on an open shot without anyone in the way and knowing it was a 2-on-1 coming at him is not a good goal. Was there a miscue by the Devils skaters? Yeah. Graves failing to keep the puck in created the 2-on-1 and Severson made a total mess of this play. How does a defenseman play 570 games and still not know what to do in a 2-on-1? How?

NHL Highlight Video

This one is a different kind of error by the goalie. Vrana was behind the net and tried to make a pass out in front. Vanecek paddled the pass away. Only to knock it up to the zone instead of back to the corner or trying to grab it. The result is in the above picture. The puck came right to an uncovered David Perron, who was crashing the net. Perron finished the drill. One can question why no one picked up Perron. I question why Vanecek put the puck at risk or why no one told him otherwise.

Was this a bad goal to allow? Yeah. Vanecek served it up for Perron and ate a GA for it. Was there a miscue by the Devils skaters? Despite the picture showing a lot of Not-Defense, I do not think they were prepared to account for this happening. Just a mistake from the goalie.

NHL Highlight Video

The fourth goal in the second period was in 4-on-4 play. The fourth 5-on-5 goal came in the third period. Shortly after Tomas Tatar lost the puck to Dominik Kubalik along the right (Vanecek’s right) side boards, Kubalik tossed a quick pass to Raymond, and darted ahead for a quick give-and-go. As Dougie Hamilton took out Raymond with a hit after the pass, Jonas Siegenthaler had to stretch to his left to get to Kubalik. The forward had the poise to keep the puck away from him as he essentially went around Siegenthaler. While Vanecek tracked this puck, Kubalik lifted the puck past Vanecek’s glove for the goal. The fans were already mad at the coach; this play added to it. I cannot blame them.

Was this a bad goal to allow? Yeah. While this play was the result of a turnover, Vanecek had the time to prepare. He is in position to make a stop. The contest from Siegenthaler was not that effective. Yet, Kubalik beat him. It was a good shot. It would have been tough, but part of the goaltender’s job is to make some tough stops. Was there a miscue by the Devils skaters? Yeah. Tatar lost the puck to Kubalik to begin this whole mess. Hamilton stepped up on Raymond, which forced Siegenthaler to scramble. And Siegenthaler looks bad getting beaten wide by Dominik Kubalik as he put it past Vanecek. Like the 2-on-1 goal against, this was a big mess on both sides.

Game #3- Devils vs. Anaheim

The Devils stemmed the bleeding by giving up just two goals against Anaheim. Both were in the first period. Both were out of four shots on net until the official scorer changed it to six at intermission. An upgrade of 50% saved to 66.7% saved is still pretty rotten. Fortunately, the game got better and a post bailed out Mackenzie Blackwood in the second period. All pictures for Game #3 come from this highlight video on the NHL’s Youtube page.

NHL Highlight Video

This play is a worst case scenario for defensive zone faceoffs. Erik Haula lost this draw. The puck came back to Jakub Silfverberg. He rifled it through traffic to beat Blackwood. Blackwood is prepared for something but he could not have seen this shot coming. While it would have been great if Haula won this draw, the real culprit was what created this defensive zone faceoff. It was a long pass attempt through the neutral zone to Miles Wood that missed badly. The Devils like to look for those leading passes to begin with. It has been common that whenever Wood is on the ice, the Devils look to dump-and-chase to use his speed. It failed. Icing was called. Isac Lundestrom beat Haula clean off the dot. Silfverberg shot and scored through traffic.

Was this a bad goal to allow? No. As much as Mackenzie Blackwood has to earn some of benefit of the doubt, he was screened on this one. No real chance for him here. Was there a miscue by the Devils skaters? I would say that long pass up ice that led to the icing was a mistake that the Devils paid dearly for. Right off the draw? I guess Haula losing would be it. It is not like the Devils lined up in some odd formation for the draw. Normally, this would be chalked up to just a bad outcome, but when you’re 0-2-0 and you go down 0-1 to a team who was barely attacking the Devils, the reaction was understandably harsher.

NHL Highlight Video

The final goal to look at is just eye-roll inducing. Anaheim gets close to the net for one of the few times in the first period. Earlier on the play, Brendan Smith and Mason MacTavish were tied up and it led to Blackwood losing his stick. Then Cam Fowler took a shot from above the left circle (Blackwood’s left) and misses the net. MacTavish, still covered by Smith, recovers the puck off the endboards, and tosses it back in front. Silfverberg, who curled into the middle from the right side (Blackwood’s right) is there to get this puck and put it past Blackwood. The picture here tells you a thousand words. There is one (1) Duck by the crease. There are four (4) Devils within 15 feet of him. None of those Devils are going to do anything to Silfverberg. None of them stuck with him as he circled into the middle. None of them adjusted to protect the slot. None of them seemingly talked to each other. And since Blackwood is stickless, he is even more prone. Silfverberg scored. It’s 0-2 with less than five shots on the scoreboard. The crowd was seething.

Was this a bad goal to allow? No. If he had a stick, I would be more critical of Blackwood here. Even so, the Devils pretty much let a Duck go right to his doorstep and beat him dead to rights. Was there a miscue by the Devils skaters? Absolutely. Look at the picture. This is a breakdown.

Conclusions from the Pictures

Out of those 9 goals, I counted 5 goals where I observed that the goaltender allowed a bad goal and I counted 6 where the Devils skaters made a mistake or had a bad play that created the goal allowed. There were multiple goals where I would blame both the goalie and the skaters. And, goodness, this did not look good for Severson and Graves. The main conclusion is that those claiming the Devils goaltending has not been good enough and those claiming the Devils defense has made costly errors are both correct to degrees.

Related to expected goals, the pictures show how a team putting up a low xGA count could still yield a high actual GA count. The expected goals model has nothing to do with goaltending other than being a point of comparison. The expected goals model has nothing to do with whether there was coverage on a shot, a shot created off a turnover or a bad bounce, or a play made because someone erred in coverage. By the model, it implies the Devils’ defense has been rather solid and the goaltenders have let them down. They are not By the actual goals allowed, the blame needs to be shared beyond the crease on some plays.

What this does not really show is a failed or failing system. I know I have been guilty of blaming the coaching staff for these errors and, well, I should walk some of that back. This is just 9 plays in 5-on-5 where the Devils got beat and it cost them on the scoreboard. Some are certainly fixable and should be fixed by the coaching staff. But this is just a snapshot of what has been happening in games. A peek at the defense; not the whole picture. We know from the low xGA counts that the Devils have not been giving up loads of high-danger chances or even shots in their games. So it is not as if the Devils are bleeding lots of shots, which would indicate a real systemic issue. It may be that when the Devils fail, they have a tendency to fail so big that it is likely to cost them. Mistakes that seem, well, easy to correct from a viewer’s standpoint. Mistakes that make you wonder what the pros are thinking. Or even if they are communicating with one another. The Devils’ goaltending has been such a problem, that the last thing you would want either Blackwood or Vanecek do is put either in positions to bail out these mistakes. And, well, they did not sometimes.

This is how you get this situation where 9 goals have been allowed in 5-on-5 despite a model stating that it should’ve been 3 or 4 - which is mostly based on shot location anything else. The other factors, well, they have been factors.

Concluding Thoughts & Your Take

As the season goes on, if the Devils continue to “out perform” their xGA, then it will be more reflective of the goaltending being straight awful for a third straight season and/or the Devils making catastrophes in their own end to contribute to their goaltending problem. Stats tend to better reflect how a team performs with more data and there has been just three games so far. But there does not need to be an argument over whether the Devils’ goaltending was an issue in their first three games or their defensive faults were an issue. The answer is both. The expected goals model is not really reflective of either; but that is not the fault of the model. There is no need to throw that out.

By the way, I did write this all up before Thursday’s games. The Devils did beat the Islanders 4-1 and, similar to their games against Detroit and Anaheim, they absolutely controlled the run of play. They out-shot them 41-17 on the road. With just 11 forwards, mostly leaning on 9 since Lindy Ruff only trusts three lines to attack. Which is hard to argue against since, again, they out-shot the Isles 41-17. The expected goals count in 5-on-5 ended at 2.56 xGF and 1.19 xGA, and the score more reflected that with a 3-0 score in 5-on-5 play and 4-1 total. As I stated earlier, the offense just needed to keep attacking and get that finish - and they did come with three goals against a stud goalie in Ilya Sorokin. And the defensive breakdowns were few with Blackwood stepping up. This is very encouraging. If Blackwood can be more consistent in net (he gave up nothing bad in the Anaheim game or the Isles game, for that matter), then defensive miscues may not often end up in goals against to doom Devils to losses. Which would mean a better season for everyone. The true tests are yet to come, but the larger lesson is that the expected goals model is not completely junk and the actual goals are starting to become more favorable.

By the way, Damon Severson played a way better game in Long Island than he did in any of these three games. As did Ryan Graves. Both were called out in this post multiple times, and both need to be better. I hope they can build off of yesterday’s game to do so.

Now you’ve read my take on all of this expected vs. actual goal issues; my thoughts for the offense; and went over some pictures of goals against to hopefully prove my point of the Devils’ issues in those games. I want to know what you think. Do you have a better understanding of what the expected goals does and does not do? Did you agree with the thought of the offense getting more actual goals if they keep it up? If not, did you see the Islanders game? What did you learn from the pictures of the goals allowed? What should the Devils try to fix, even coming after two straight wins to even up the record? Please leave your answers and other thoughts about expected goal models and actual goals in the comments. Thank you for reading.