clock menu more-arrow no yes

Filed under:

Let’s Experiment in EHM: A (Mostly) Corsi-Optimized Roster for 2019-20

New, comments

After showing how a hockey team optimized for WAR could dominate in Eastside Hockey Manager, this week’s experiment is to show how a team mostly optimized for Corsi impact would perform in the game.

Pictured: Brad Marchand scoring of the most crucial goals of the entire campaign.
Pictured: Brad Marchand scoring of the most crucial goals of the entire campaign.

Welcome to the fourth experiment in the Let’s Experiment in Eastside Hockey Manager (EHM) series. Last week’s experiment saw Jack Hughes play his 2019-20 season in several other leagues before playing 2020-21 in the NHL and the results of how he did. In general, he crushed it in the other leagues before having a tougher time in the best league in the world. This week’s experiment is another exercise in optimizing a roster for a given stat. As suggested by MedicSBK from a few weeks back:

How about utilizing the highest ranked Corsi players from the 2018-19 season on a team together to see how they perform when compared to a WAR optimized team?

This comment came in response to the WAR-optimized Team post, where they steamrolled most of the NHL in the regular season and went 16-1 in the Stanley Cup Playoffs. It is a good point of comparison. I did indicate in last week’s post that I would be using 2019-20 data instead of 2018-19 to be consistent with the WAR-optimized team. Let us get right to it.

The Experiment

Replace the 2019-20 Devils roster with a NHL roster optimized for Corsi based on the 2019-20 season. Play through the 2019-20 season with this Corsi-optimized roster.

The Approach & The Confounding Factors

As CJ put together a linear optimization algorithm to create an optimized roster with WAR, I asked him to do so for Corsi for all NHL players who played at least 500 minutes in the NHL in 2019-20. This way we filter out called up players who only played a little bit and therefore have an extraordinarily high or low Corsi value. Rather than using CF% directly, CJ opted to use Corsi Impact, which is generated from Relative Adjusted Plus Minus (RAPM) Corsi for and against. The general idea with RAPM is that it takes into account how a player is deployed (ice time), the game situation (score, venue) and comes with an expected value for the stat. This is compared with their actual Corsi. The more they exceed their expectation, the higher of the Corsi impact. This method, while more complex than just optimizing a player’s CF%, does allow a more diverse lineup with respect to team selection instead of just a roster filled with Golden Knights, Canadiens, and Hurricanes.

Goalies do not have this value and they really do not influence it. They stop pucks. That is their main job. So I arbitrarily decided to go with the goaltending tandem of the team with the highest CF% in 2019-20. That would be the Vegas Golden Knights per Natural Stat Trick. In order to stay consistent with the WAR-optimized lineup, the base database I used was for the start of the 2019-20 season. This means my tandem was Marc-Andre Fleury and Malcolm Subban. I informed CJ of this before he ran his algorithm to give me a cap-compliant roster.

CJ gave me a list of 23 skaters, of which they all fit under the cap by at least seven million dollars. This meant I had to make some adjustments, which is why this is technically a mostly optimal Corsi team instead of an optimal one by CJ’s method alone. I cut a defenseman and a winger to bring it down to 21 skaters: fourteen forwards and seven defensemen.

The Corsi Impact method actually yielded many of the league leaders in CF%. Tomas Tatar and Shea Theodore, the CF% leaders among forwards and defensemen respectively at Natural Stat Trick, were on the team. Five of the top ten in CF% among forwards were on the team. The lowest CF% on the team was Adam Fox’s 51.59%, but his Corsi Impact exceeded +146 events -which was higher than half of the blueline. Most of CJ’s results were deep in the “green” in terms of CF% and the cap restriction meant I did not have the luxury of Mark Stone ($9.5 million) or Sebastian Aho ($8.454 million).

Knowing I had plenty of cap space, I made two replacements. I decided to replace Thomas Chabot and his sub-50% CF% with Dougie Hamilton. While Chabot’s Corsi Impact was quite good and his CF% was arguably submarined by being a Senator, Hamilton has been a run-of-play king for years and finished just short of Theodore in CF%. So Hamilton was in. I also decided to take out Anthony Cirelli. While his CF% was solidly in the 53% range, his Corsi Impact was the lowest among CJ’s results. We also saw Cirelli in the WAR-optimized lineup and saw he was not very good as a fourth liner. I decided to replace him with Jordan Weal, who was both cheap ($1.4 million) and finished tenth among all forwards with a 56.55% CF%. I had hopes for Weal to be a better fourth liner than Cirelli. Current me has bad news for past me, but I will get into it later.

So those replacements are the confounding factor. I think they are justifiable moves. That said, in retrospect, I wish I made some different ones. Here was the roster to be used (pay no mind to the in-game date):

The 2019-20 (Mostly) Corsi Optimized Devils
The 2019-20 (Mostly) Corsi Optimized Devils

You can see the problem from the roster. The blueline is strong and the winger depth is fantastic. (Aside: AI Tom Fitzgerald is a bit clueless to claim Max Pacioretty as the fourth best left winger on the team and that Kevin Lebanc was the best right winger). But it is weak in the middle. Weal’s position in the game is LW/C/RW, which means he can play any of the three forward positions. I learned he was not very good at the one I needed him to play at. I also learned that two of those three centers listed here did the job I expected. The other question mark was in net. Fleury has experience and Subban has some upside, but which versions of them were I going to get? As I found out, I was going to get a very familiar experience in the crease.

By the way, please note that some of these players were on the WAR-optimized roster: Brad Marchand, Roman Josi, Valeri Nichushkin, Oliver Bjorkstrand, Andrei Svechnikov, and Vince Dunn. Had I not switched him out for Weal, Cirelli would be another.

The Limitations

Before going into the results, I want to re-emphasize the limitations I put on myself for this season. Just with the WAR-optimized game, I had to use this roster as much as possible. I was not allowed to waive, demote, or trade any of the 23. Additionally, as I was using the Devils for this game, I have to use Binghamton and their system for any call-ups in the season. I could only call someone up from Binghamton if I had fewer than twelve available forwards, six defensemen, and/or two goalies. Lastly, in order to ensure this roster, I had to play the game in full. I could not do what I did for last week’s experiment where I ran the game and looked at the results.

The Experiment Results: The 2019-20 Regular Season

I loaded the modified database with this team for the Devils, started a new game, and took inspiration from one of the godfathers of the hockey analytics scene for the GM name: Vic Ferrari. He has revealed himself as Tim Barnes, who works for the Washington Capitals and has a Stanley Cup ring.

The preseason did not have much happening until nearly the end of it. The Corsi-optimized Devils crushed exhibition games as I set the lines and the tactics to mirror the WAR-optimized team. Basically: a lot of shooting, creative passing, an offensive mindset, and rarely any dump-ins. I did get find out two bits of bad news:

  • I called up some of the Binghamton team to fill in spots for training camp scrimmages. I sent them all back by the end of camp. I could not send Jeremy Groleau back to Binghamton as he was not 20 years old. Thankfully, I did not call up Ty Smith (I put him on Binghamton in the WAR-optimized game) as I would have had to send him to Spokane. This becomes relevant much, much later in this post.
  • Svechnikov suffered a month-long injury in the final preseason game. I had the replacements - Lebanc, Toffoli, and Bjorkstrand - ready to go, not that they took command until one of them got going.

The regular season started out with disappointment. The Devils were winless in their first three games. Despite how good the wingers and defensemen played, Fleury and Subban struggled and Eller and Weal were in the five to six range in terms of rating. While it was early, it started to confirm my concerns about the roster. Then the Devils ran off ten straight wins to ease any concerns. Svechnikov came back, which pushed Bjorkstrand and Lebanc to the scratch list, joining Jeff Petry who struggled and lost his spot to Fox.

Most of the season consisted of the team dropping a game or two and then rattling off a winning streak of six to ten games. More and more of the games were closer and required post-regulation play. The Corsi-optimized team went to a shootout more often, which was largely positive as Marchand, Josi, and Tomas Tatar were quite good at it. Still, I could tell this team was a bit of a drop off from the WAR-optimized lineup in season. Fleury and especially Subban were prone to having some rough nights in net. Eller and Weal nearly guaranteed sub-seven rating performances on most nights, with Weal often getting a five (which is bad, the “average game” rating for everyone is a six). The team suffered injuries more often. Usually for a week or two; Svechnikov missing October was the most severe. However, it meant a lot of juggling of lines and pairings. Especially when Couturier and Weal suffered injuries for a week each. I eventually found out and settled on Kevin Labanc playing center in Weal’s place. Labanc is merely “competent” in EHM at center and yet performed much better than Weal, who is “accomplished” at the position.

That said, there were still plenty of excellence on display. The Devils maintained a point percentage above 70% since their first winning streak ended throughout the entire season. Nikolaj Ehlers, who was my first line right winger, was a goal scoring machine through to NHL Trade Deadline with 38 goals in 61 games. He had a shot of being the first Devil to ever score 50 in a season. Alas, his goal scoring slowed down in the final 21 games to just four; but he continued to amass assists to stay near the league leader in points by season’s end. Ehlers was a star. Pacioretty resembled Panarin from the WAR-optimized team as being one of the league’s best despite being on a second line. Marchand was very good yet again and managed to avoid a suspension (!). The first pairing of Roman Josi and Dougie Hamilton was brilliant as both averaged ratings above eight. Svechnikov missed twelve games and still managed to nearly put up a point per game, nearly 200 shots (199), and finished in the top fifty in scoring in the NHL. The team as a whole averaged at least 3.8 goals per game and often outshot opponents by large margins. It was not as dominant as the WAR-optimized team, but they were similarly running through opponents. There were just more hiccups and off nights, particularly from the goalies.

By the end of the season, the (mostly) Corsi-optimized Devils managed to pull away from the super-competitive Pittsburgh, Columbus and Washington teams in the Metropolitan and get ahead of Buffalo (!) to win the Eastern Conference and the President’s Trophy. They finished with a record of 59-13-10 with a 7-6 record in shootouts. Was this as good as the WAR-optimized team? No. It was still a 128-point season, which would be the most in franchise history if it happened in real life, with a 78% point percentage. They still qualified first for the postseason by the start of March. This team was expected to be Cup contenders by the board in the game and they were definitely contenders.

The final Eastern Conference standings for the mostly Corsi-optimal Devils
The final Eastern Conference standings for the mostly Corsi-optimal Devils

The East was no joke in 2019-20. Washington, Pittsburgh, and Columbus were all high-level teams. Even with the winning streaks, the Devils did not take over the top spot until 2020 and pulled away when the other three teams faded a bit. Buffalo ended the season on fire but they were not catching the Devils for home ice throughout the playoffs.

Leading producers in this 2019-20 EHM Season.
Leading producers in this 2019-20 EHM Season.

Ehlers’ goal scoring really cooled off in the stretch run to the season. He finished just shy of the Art Ross, which was sniped by Stamkos by the season’s end. Still, Ehlers was the main producer on the Devils. Marchand finished a solid tenth and you can see Pacioretty in 22nd. Both Pacioretty and Marchand primarily play left wing; while I kept Marchand on the first line, I sometimes wonder if I should have put Pacioretty there instead. While he is not in this screen shot, a special mention goes to Svechnikov finished 35th in scoring with 68 points in 70 games.

For defensemen, Hamilton finished with 54 points. This was one point shy of Nate Schmidt for the league lead in points. Roman Josi was the only other Devil to make the top fifty in scoring by a defenseman; he finished tied for 21st with 42 points. The mostly Corsi-optimized Devils only had one rookie make the top fifty among rookie skaters in points: Adam Fox, who had 29 and finished sixteenth among rookies. The only rookie who was on the Devils was one call-up appearance for Mikhail Maltsev when three forwards were injured prior to a game. He did not register a point. He did well in his one game.

Here is the full list of Devils by average rating, which represents the how the player performed. It also includes their production, so you can see how much they produced. Of note, eight players finished with fifty or more points (Ehlers, Marchand, Pacioretty, Couturier, Svechnikov, Gallagher, and Hamilton) and three scored at least thirty goals (Couturier, Pacioretty, Ehlers)

Devils roster by average rating.
Devils roster by average rating.

Pacioretty and Ehlers finished second and third in the entire NHL in average rating (Leon Draisaitl led the league with a stunning 8.44.) and led the Devils. Still, anything above 7.5 is quite good and this team had plenty of players hit that mark. Hamilton, Marchand, and Josi were great and Svechnikov nearly joined them in the eight club. What makes Svechnikov’s line more impressive is that he was kept to the second power play units and the third line for a majority of the season since Marchand and Pacioretty were ahead of him on the depth chart. Speaking of non-first line wingers being great, Brendan Gallagher had a great season. I do not think he got enough credit as he deserved. Tyler Toffoli had a bad training camp and preseason, but I was encouraged to see him perform as well as he did. It kept Labanc and Bjorkstrand to the scratch list but he earned it. They were also kept to the scratch list due to good seasons by Nichushkin and Tatar among the bottom six. Among the defense, everyone did well enough. Petry was the “weak link” but a 7.16 average rating being the lowest is still quite good. I did not know what to expect from Sean Walker and I wondered if Vince Dunn could be solid again. Dunn was and Walker was very good in the role I used him in.

At center, I expected Sean Couturier to be very good, hoped Danault and Eller would be good, and crossed my fingers Weal would not be worse than Cirelli. Couturier was very good and I missed him even if he was out for only a week or two. Danault was acceptable; I expected better ratings but I am not going to be mad about 22 goals and 57 points. Eller was a disappointment and Weal was considerbly worse than Cirelli. I should have kept Cirelli. Or played Labanc at center earlier than I actually did.

I was confused to see Fleury get a 7.54 and Subban get a flat 7 rating. Look at these save percentages:

The Devils goalies on the mostly Corsi-optimized Devils.
The Devils goalies on the mostly Corsi-optimized Devils.

Your eyes are not deceiving you. Fleury was just above a 90% and Subban posted a percentage that may be would have been acceptable when he was born in 1993. I do not know how they got the ratings they did in the game. This was similar to the Devils goaltending I had with the re-do of the season. I could have even saved more money if I used Blackwood and somehow found a loophole or justification to replace Schneider and maybe then I could have swung for Aho or even Paul Stastny (two of the top centers in CF% in 2019-20) to strengthen the center position. Alas. I set my own requirements and these are the results.

The Experiment Results: The 2020 Playoffs

The results continued into the playoffs. Our Hated Rivals qualified for the postseason by season’s end. The mostly Corsi-optimized Devils swept them. Fleury decided to up his game, Ehlers put up five goals including the first goal of the entire series jus 33 seconds into Game 1, and the Devils won both games at the World’s Most Overrated Arena 6-1 and 4-1. Next.

Pittsburgh, who was the second highest scoring team in the East and finished second to the Devils in the division, was the next opponent. They came off a sweep of Columbus, which was a bit more dramatic as their first two wins came in overtime. It seemed like another sweep was in the books for the Devils. They won Game 1 by a 4-0 shutout, they won Game 5-3 after a 3-3 first period, and they won Game 3 by 3-1 with Ehlers’ first two goals of the series. However, the Pens provided a bump in Game 4. The Devils saw a two goal lead blow up in the third period with Kahun scoring an equalizer with less than five minutes in regulation. Fourteen minutes and twenty one seconds into overtime, Marchand decided to live up to his real-life reputation. He took a cross-checking penalty and said something to get a game misconduct. Less than a minute later, Rust scored to hand the Devils their first playoff loss. How would the mostly Corsi-optimized Devils respond? By another shutout in Game 5, a cruising 3-0 win with another goal by Ehlers. Fleury was fantastic throughout most of this series and this team was onto the Eastern Conference Finals.

Their opponents would be Tampa Bay. The Bolts finished seventh in the East and had a tall order in front of them in the first round. They preserved Toronto’s tradition of not playing hockey games in May by beating them 4-2 in the series. Up against a really, really good Washington team, Tampa Bay swept them as no one could stop Stamkos or really contain him. While the Devils dumped all over Tampa Bay in the season series, the playoff series would be much more dramatic and epic.

Game 1 proved that the Lightning would be no pushovers. Rocco Grimaldi, whom they acquired, was a sparkplug and got things going with the series’ first goal. The Lightning’s play off the puck kept the roaring Devils offense to just 30 shots. And Andrei Vasilevskiy was in full effect, stopping all but one of them. The Devils lost 1-3. Game 2 was a tighter affair but yielded more concern. The Devils scored first and second in the game thanks to Ehlers in the first period and Svechnikov in the second period. But they blew the lead in the third period to Stamkos and a really awkward goal to Cirelli (sigh). Early in overtime, Palat sent the fans at the Rock hope unhappy for a 2-3 loss and a 0-2 series deficit. Were these Devils in trouble? No. Just as the cliche goes, they had to take it one game at a time and they did just that in Tampa Bay, Florida. The line of Pacioretty, Gallagher, and Danault had a fantastic night in Game 3. They made Vasilevskiy look less like a superhero and more like an average goalie. The unit was responsible for three goals against the goalie and an empty netter to boost the score to 5-3. It was fine win, slightly marred by Adam Fox being injured for five weeks with a hip pointer. In came Jeff Petry for Game 4. In that game, the second line were not as productive, but they struck again with support from Toffoli (who was on the fourth line with Nichushkin and Labanc at center) and Walker. The Lightning did come back from a 4-1 game to make it 4-3 in the third period, but Josi sealed the win in the game’s final minute with an empty netter for a 5-3 win a tied series. The road team won every game in the series so far, which is uncommon to see in the playoffs.

The uncommon event continued on in the series. In Game 5, the first period was scoreless, Ehlers scored halfway through the second period, and Tampa Bay ran up three goals in the third to stun the Devils and put them on the brink. The 1-3 defeat meant that Game 6 was a must win at the Amalie Arena. Just as with Games 3 and 4, they took care of business. An early penalty led to a smashing power play goal from the point by Hamilton, Labanc scored minutes after a Grimaldi equalizer, and Josi sealed the win with an empty netter. The Devils out-shot the Bolts 42-22 and Fleury ensured that no more pucks would get past him after Grimaldi’s score. A Game 7 was needed. This one was tense. while the Devils piled up the shots, not much was getting through Vasilevskiy. Ehlers’ scored his 13th of the playoffs early in the second period to send the Rock into rapturous applause. Tampa Bay silenced them when Killorn scored seven minutes later. With 4:40 left to play, Svechnikov converted a power play to put the Devils up 2-1. Could they hold on for 4:40? No. Tyler Johnson increased the anxiety of Devils fans worldwide with a goal within the final two minutes. Overtime was necessary. One bad play could decide it all. One bad decision could decide it all. It turned out that the bad move by Cedric Paquette would do it. He cross-checked Nichushkin down in open ice at 68:58. He was incredibly unhappy about the call. Couturier won the draw, Hamilton fired a low and hard shot on net, Vasilevskiy kicked the rebound long to the circle to his right where the draw took place, and Marchand hammered the rebound. It beat Vasilevskiy. The Rock exploded. The Devils cleared the bench. A home team finally won a game in this series and it was the most important one. The Devils were on to the Stanley Cup Finals.

Meanwhile, Anaheim was waiting. They swept Edmonton in the first round, prevailed over Chicago in five games in the second round, and dispatched Colorado in the Western Conference Finals. Hampus Lindholm was on another level with 16 assists in fifteen games, Rickard Rakell had ten goals and seven assists to lead the offense, and John Gibson was sensational. The mostly Corsi-optimized Devils escaped a tight series that could have gone either way. Could this be another series like Tampa Bay? One that left me, the player, on the edge of my seat?

No. Not really. Game 1 was a romp for the Devils. As Toffoli struggled by the end of the Tampa Bay series, I put in Bjorkstrand. He rewarded my choice with a first period goal to open the scoring. Ehlers converted a power play and Marchand slammed in a bad Ducks turnover to make it 3-0 by the end of the second period. Marchand scored quickly in the third to make it 4-0 and Kase scored a consolation goal later on. A PPG by Couturier ended a fun 5-1 win. In Game 2, it was a bit more of the same with the Devils dominating the Ducks. The power play did most of the damage with two quick PPGs by Hamilton and Couturier to open the game. Gallagher made it 3-0 with one of his own. While Anaheim’s Sprong responded with a PPG in the second period, it was another cruising win. Gallagher put the exclamation mark on it for a 4-1 win.

As the Devils out-shot the Ducks by 30 shots in the first two games and won big, I decided to give Subban a game just to get him a playoff game. Fleury was doing great but surely this offense could overcome any miscues by Subban. This thinking lasted for forty minutes in Game 3. Ehlers scored a quick double and Pacioretty made it 3-0 but Silfverberg and Steel made it 3-2 by period’s end. Those two goals against were out of fewer than ten shots. Couturier scored in the second period to restore a two goal lead. Then a horror show happened in the third period. Ritchie and Kase scored nearly a minute apart in the third to tie up the game and send Subban packing. He gave up four goals on fifteen shots. Fleury could not save the day. A turnover yielded a go-ahead goal by Ritchie and Lindholm secured the win with an ENG. The Devils lost Game 3, 4-6. This would be the end of the Ducks’ resistance in the series. With Fleury back in net, he rebounded in Game 4. As did the team. While Steel matched a Svechnikov goal in the first period, Danault and Marchand scored in the second period. Comtois made it a one goal game late in the second, but an early third period PPG by Gallagher made it a 4-2 game. Fleury did well to maintain it and the Devils were one win away. However, they would have to do it without Eller and Ehlers. The loss of Ehlers was larger as he had a whopping 16 playoff goals. The NHL record for most goals in a single playoff year was 19. The Devils would just have to do it without him and with Weal in the lineup to replace Eller.

To add to the challenge, Shea Theodore was injured in practice prior to Game 5. With Fox and Theodore out, I had to call up a Binghamton Devil. Binghamton had a bad season and did not make the playoffs, so there were a lot of idle players. Remember how I was glad I did not call up Smith to fill in a scrimmage spot in training camp? This meant he was in Binghamton for the entire season, just as he was in the WAR-optimized game. This also meant he was able to be called up to fill in on defense. Smith would make his NHL debut in a potential Cup clinching game. I did not need him to be great; he could just be OK and that would be fine. Just like I hoped for Weal. Just like I hoped for Bjorkstrand, who had to come in as the other right wingers - Gallagher, Tatar, Toffoli - moved up a line.

Game 5 was tense and the Ducks struck first when Rakell scored with less than five minutes left in the first. But the Devils had a flair for timing in this one. Gallagher finished a pass from Couturier with one second left in the period to tie up the game. Within the first minute of the second period, Toffoli took a loose puck knocked away from Svechnikov and slammed it past Gibson to make it 2-1. Adam Henrique gave the Rock plenty of mixed feelings when he tied up the game. On the next shift, Kevin Labanc had an instant response. Biorkstrand took a hit in the offensive zone, recovered his own puck, sent it across to Labanc, and with Steel in his grill, hammered a one-timer that fooled Gibson. The Devils were up 3-2 and sought to increase the lead. However, Gibson kept stopping everything. The Ducks were banking on an error or something where they could beat Fleury. In the regular season, maybe that would have worked. But not this version of Fleury. Not in this postseason. Not in this game. The Devils held on to win 3-2 and win the Stanley Cup.

The (mostly) Corsi-optimized Devils won the Cup!
The (mostly) Corsi-optimized Devils won the Cup!

The mostly Corsi optimized Devils did not rampage through the postseason like the WAR-optimized team. That squad went 16-1. This team went 16-5. Still, they won the Cup. They had a franchise-historical performance out of Ehlers with 16 goals (only seven players ever had more than 16 playoff goals in a year), Couturier was sensational, Gallagher and Pacioretty were wonderful, Hamilton was a stud, and Fleury was fantastic. It was not an easy run and the ECFs really could have been the end. But they achieved the goal the board set at the beginning of this season: The Cup was theirs. Let’s look at some stats:

Devils playoff stats by points
Devils playoff stats by points

I am still mystified by Ehlers putting up 16 goals in the postseason. He picked the right time to start heating up in the postseason. It would have been a long shot in Game 5, but I wished he did not get hurt in Game 4 so he had a chance to at least get closer to Kurri’s record of 19. Couturier was seemingly ever present on the scoresheet and he edged Ehlers for the league lead in postseason scoring. Similar to the regular season, many of the names who produced and had great average ratings did the same in the playoffs. Danault out-produced his rating again; Toffoli filled in well in a lesser role, and everyone contributed. Smith’s one call-up appearance was great and he is the only skater without a point.

The only ones dragging in terms of performance were Weal and Eller. I took out Weal and put Labanc as a center during the postseason. I can live with Labanc’s 6.8 since he was out of position. It is telling he out performed Eller, who primarily plays center, and Weal, who could play center, even with how I used him. That was a surprise. The biggest one: Fleury.

The Devils goaltenders in the playoffs
The Devils goaltenders in the playoffs

Fleury finished a few thousandths above 90% in save percentage during the regular season. He really turned it up with a 92.2% in the postseason. He had two shutouts and he was consistently good. His disasterous fill-in performance in Game 3 of the Finals took down his save percentage a little bit. However, among all of the Devils, I consider him to be the one who stepped up the most. While it was not 100% the case, he helped maintain leads and keep games close. Even the five losses were within two goals at most. In retrospect, I never should have given Subban that game but it ultimately was a mistep and did not risk the final result of the Cup.

The Experiment Results: The 2019-20 Awards

So the mostly Corsi-optimized Devils were not as dominant in the regular season and in the playoffs as the WAR-optimized Devils. As such, they were also not as dominant when it came to the 2019-20 awards. Here is a list of their results:

  • Art Ross Trophy - Third place, Nikolaj Ehlers with 92 points (Winner: Stamkos with 94)
  • Rocket Richard Trophy - Third place, Ehlers with 42 goals (Winner: Tie - McDavid and Guentzel with 44. Not sure how McDavid got it.)
  • Frank J. Selke Trophy - Second place, Sean Couturier (Winner: Ryan O’Reilly)
  • GM of the Year - Winner: Vic Ferrari
  • Lady Byng Memorial Trophy - Winner: Ehlers (92 points, 14 PIM)
  • King Clancy Trophy - Second place, Max Pacioretty; Third place - Ehlers (Winner: Draisaitl. This is effectively the Best Average Rating award)
  • Jack Adams Award - Winner: John Hynes
  • Conn Smythe Trophy - Winner: Marc-Andre Fleury (Second place: Ehlers, Third place: Brad Marchand)

The Corsi-optimized Devils were just not supremely dominant enough to own the league in many awards. I would have thought Hamilton would be a finalist for the Norris, but alas, that was not the case. I was surprised to see that only Ehlers made it to the NHL All-Star Second Team. In terms of monthly awards, Ehlers was the NHL’s Player of November and Pacioretty was the NHL’s Player of February. Yet, they opted for Landeskog and Laine as left wingers for the All-Star teams instead of Pacioretty or even Marchand. Oh well.

Experiment Conclusions

As MedicSBK wanted to see this team compared to the WAR-optimized team, then there is little doubt that the WAR-optimized team was better. They were more complete and impressive on paper. The centers and the goalies make it no contest. The WAR-optimized team had MacKinnon, Point, Pettersson, and Cirelli down the middle with McCann as depth. The center group of Couturier, Danault, Eller, and Weal does not compare. Similarly, the WAR-optimized team had a goaltending tandem of Hellebuyck and Kuemper. That is better than Fluery and Subban. Throw some more injuries (quantity, not so much in length) for the mostly Corsi-optimized team and some breaks not going their way and that team fell quite a bit short of the WAR-optimized team in terms of results and performance. While both teams out-shot and out-scored their opposition by a lot, the WAR-optimized team was truly dominant in both the regular season and the playoffs. They won 69 games in the regular season, lost just one game in the entire playoffs, and won several trophies on top of the President’s Trophy and Lord Stanley’s Cup. The mostly Corsi-optimized team, while extremely good and still managed to win the President’s Trophy and the Cup, did not dominate to that level.

There is an argument that this Corsi-optimized team was not truly optimal. Again, I did make two adjustments against CJ’s results. Hamilton clearly worked; he was one of the best defenders in the league. Weal did not. However, it could be argued that using a different form of Corsi would have yielded different results. Instead of Corsi Impact based on RAPM, perhaps straight up CF% or solely a score and venue adjusted CF% would have been more in the spirit of what MedicSBK was looking for. Nonetheless, the approach taken was used and I can at least say that was done to a degree where the team was very good and successful - just not at the level of the stupefying powerhouse that was the WAR-optimized team. In that sense, I declare this experiment to be a success.

The Next Experiment & Your Turn

It turns out that updating a team with one player is remarkably tedious unless there is some easier way in the EHM Editor that I am not figuring out. So I apologize to EliasStillRocks, but I am going to shelve the All-Hischier team for later. I will get to your other idea, a WAR-minimal roster, nearer in the future.

The next experiment will be an idea I will need all of your help with. One of the things I have seen on Youtube with respect to Football Manager is a “Let’s Ruin” series (example) wherein the player takes a top tier team and tries to kneecap them as much as possible. Across the hockey family of blogs at SB Nation, we will having a two-week long Rivalry Week (a Rivalry Fortnight?). Fittingly, we will do the same in EHM. The next experiment is to see how much we are going to ruin Our Hated Rivals - the New York Rangers.

The goal of the next experiment is to make the Rangers as terrible as possible. Ideally without getting fired in 2019-20 so we have a season to do a lot of short and long-term damage to the Worst of the Original Six. I will use the same database to start at the 2019-20 season as to avoid using data updates and roster moves they made in real life. What I would like from you are suggestions to make the Rangers suck. Move big name players or futures for picks? Trade the first rounder for someone with a heinous contract like Jack Johnson? Staple Panarin and Buchnevich to the fourth line? Demote Kakko to Finland? Fire all of the scouts? Replace the head coaches with ones with terrible stats? Bring back Jimmy Vesey for Adam Fox? Change the head coach to Lindy Ruff? Make everyone dump-and-chase all the time? There is a lot of things you can do as GM of a team in EHM so there is a lot of ways we can make this team become the new Ottawa or Detroit. I want your ideas to do it, so please leave them in the comments. By Thursday, I will start the game and run it through January 1, 2020. Next week, we will evaluate how bad they are and then I will ask for one more round of suggestions. The final results will be posted in two weeks at the end of the Rivalry Fortnight.

Additionally, please let me know what you thought of this EHM Experiment. How did you react when you found out how much the mostly Corsi-optimized team accomplished and how far off they were from the WAR-optimized team? Was that ECF against Tampa Bay truly epic? What would you have done for a Corsi-optimized team? Please let me know your answers and any other ideas for EHM Experiments that you would like to see in the comments. And don’t forget those ideas to ruin the Rangers for Part 1’s experiment, which will go up next week. Thank you for reading.