TomasHertlsRooster
Don’t say eye test when you mean points
This is a long, long post with a TL;DR at the bottom. The post will attempt to compare and contrast the team before and after the “system change”.
Let’s say that our “system change” was implemented after our regular season overtime loss @ Vegas. All numbers are 5V5.
Our season results before the system change:
12-8-2 (97 point pace)
11 ROW (14th in NHL)
6th in CF/60 (60.94)
10th in CA/60 (55.23)
5th in CF% (52.46%)
10th in FF/60 (44.51)
1st in FA/60 (38.51)
4th in FF% (53.61%)
10th in SF/60 (31.75)
1st in SA/60 (27.56)
3rd in SF% (53.54%)
12th in SCF/60 (26.97)
17th in SCA/60 (25.98)
12th in SCF% (50.94%)
4th in HDCF/60 (11.71)
22nd in HDCA/60 (10.78)
9th in HDCF% (52.07%)
30th in GF/60 (1.86)
6th in GA/60 (2.04)
22nd in GF% (47.76%)
29th in oiSH% (5.87%)
15th in oiSV% (92.60%)
25th in PDO (98.5%)
For those who don’t know what this means: the team was elite at suppressing shots and carrying the rate of shots. They were very good at suppressing goals, creating scoring chances, creating shots, and carrying the rate of scoring chances. They were poor at limiting scoring chances. They were very poor at creating goals for, but this is entirely because of their poor on-ice shooting percentage.
Given the fact that an above average percentage of their scoring chances were of the high danger variety, and that their shot generation, scoring chance generation, and high danger chance generation were all above average, I’m not convinced that team really had an offensive problem. The only difference between that team and an above average offensive team (GF/60 being the only measure of offense we’re using here) was their hilariously low on-ice shooting percentage which was totally not sustainable for a team with the offensive talent that we had.
However, the argument could also be made that our goals against was unsustainable. We had the 15th best save percentage, but given the amount of HDCA we gave up relative to the amount of shots and scoring chances we gave up, one would expect our GA/60 to be due for somewhat of a regression. (Or increase in this case) The team was elite at suppressing shots but mediocre to poor at suppressing scoring chances and high danger chances.
Personally, I think the biggest factor here is Brent Burns. Burns’ offensive confidence was terrible; he only had 1 goal at this point in the season and it came in the final game of this sample size. This played a big factor in our poor on-ice shooting percentage. In addition, his defensive play was also terrible; he made boneheaded error after boneheaded error which led to a massively inflated HDCA/60; Burns’ was 3rd worst in the team in this stat behind Paul Martin, who only played in 2 games, and Melker Karlsson, who had only played half as many minutes and was our worst player all season. Burns had elite shot based metrics because his possession game was strong and he took shot after shot but the amount of crucial errors he made led to a high amount of high danger scoring chances against, and his poor shot selection and poor shooting percentage led to a high amount of shots and fenwick for without many goals actually being scored.
Below, I’ve attached the WOWYs for the Sharks and Brent Burns in that time frame. I didn’t attach every single statistic because I don’t want to flood the page with more and more stats on stats and because I think these are the most important numbers. I’ve also added where that stat would rank among all teams at that point in the season.
Sharks without/without Brent Burns, through the first 22 games:
With:
65.6% OZFO
47.75% HDCF - 24th
12.24 HDCA/60 - 28th
2.42 GA/60 - 19th
27.2 SCA/60 - 23rd
3.40% OiSH - 31st
1.32 GF/60 - 31st
33.3% GF - 31st
Without:
48.3% OZFO
56.87 HDCF% - 1st
9.86 HDCA/60 - 10th
1.8 GA/60 - 2nd
25.22 SCA/60 - 14th
7.74% oiSH - 13th
2.27 GF/60 - 20th
55.81% GF - 5th
Their possession and shot rate metrics much stronger with Burns, but were still well above average without. It’s pretty clear, though, that Burns, despite starting much more of his shifts in the offensive zone and being deployed with much stronger players, completely tanked our defensive numbers. However, it does match the eye test; he really was that bad defensively. It also makes sense that he tanked our oiSH% when his shot selection was very poor, not as dynamic as it was in past years, and he was also just unlucky on top of that.
In summary, the team for the first 22 games, before the system change, was far from bad. Brent Burns’ poor play at both ends of the ice, which was egregiously visible via the eye test, was both A) Clearly not sustainable and B) Possibly the only thing separating that team from one of the top teams in the NHL.
Our season results after the system change:
33-19-8 (101 point pace) + 4 points
29 ROW (14th in the NHL)
6th in CF/60 (60.78) +/- 0 spots, -0.16
22nd in CA/60 (59.9) -12 spots, -4.67
13th in CF% (50.38%) -8 spots, -2.08%
11th in FF/60 (44.28) -1 spot, -0.23
17th in FA/60 (42.45) -16 spots, -3.94
11th in FF% (51.05%) -7 spots, -2.56%
16th in SF/60 (31.11) -6 spots, -0.64
12th in SA/60 (30.08) -11 spots, -2.52
14th in SF% (50.84%) -11 spots, - 2.7%
10th in SCF/60 (27.67) +2 spots, +0.7
19th in SCA/60 (26.95) -2 spots, -0.97
12th in SCF% (50.66%) +/- 0 spots, -0.28%
7th in HDCF/60 (11.54) -3 spots, -0.17
21st in HDCA/60 (11.22) +1 spot, -0.44
13th in HDCF% (50.71%) -4 spots, -1.36%
9th in GF/60 (2.55) +21 spots, +0.69
23rd in GA/60 (2.61) -17 spots, -0.57
17th in GF% (49.42) +5 spots, +1.66%
9th in oiSH% (8.21%) +20 spots, +2.34%
26th in oiSV% (91.31%) -11 spots, -1.29%
20th in PDO (99.5%) +5 spots, +1%
For those who don't know what this means: The rate at which the Sharks controlled shots and high danger scoring chances got much worse. Their plain scoring chance rate generation marginally improved, but their rate of high danger scoring chance prevention and creation, along with every single one of their shot generation and their shot prevention metrics decreased. At this point in the season, they were mostly middle of the pack by all metrics; a far cry from the near elite team they were at the start of the season.
Their overall goals for percentage had improved, but this was mostly due to an improvement in their on ice shooting percentage. Their increase in on-ice shooting percentage can be mostly explained by the fact that they took far less shots, but scored on a ridiculously higher amount of those shots. A higher percentage of the shots and chances they did get were of the high danger variety, but they still created less high danger chances than they did before. It can be expected that this would lead to a small bump in their shooting percentage, but the bump in overall goals for shows that the bump in shooting percentage was a lot more than we might be able to expect.
Sharks with/without Brent Burns, through the final 60 games.
With:
59.29% OZFO
52.28 HDCF% - 10th +14 spots, + 4.53%
11.95 HDCA/60 - 25th +3 spots, +0.29
2.55 GA/60 - 20th -1 spot, -0.13
26.86 SCA/60 - 19th -4 spots, -0.34
7.20% oiSH - 25th +6 spots, +3.8% (THIS MORE THAN DOUBLED!)
2.39 GF/60 - 16th +15 spots, +1.07
48.42% GF - 21st +10 spots, +15.12%
Without:
42.1% OZFO
49.54 HDCF% - 15th -14 spots, -7.33%
2.66 GA/60 - 25th -23 spots, -0.86
27 SCA/60 - 21st -7 spots, -1.78
8.92% oiSH - 3rd +10 spots, +1.18
2.66 GF/60 - 6th +14 spots, +0.39
50% GF - 17th -12 spots, -5.81%
The team as a whole seriously improved when Brent Burns was on the ice, and got seriously worse with him off the ice, after the system change. The team was under or very close to 50% in all shot and chance based metrics without Burns on the ice while they were comfortably above 50% in all metrics without Burns on the ice before the system change. The Sharks' improvement with Burns on the ice makes a lot more sense and it has a much bigger correlation with his play - he stopped taking such low danger shots, which led to an increase in his on-ice shooting percentage. His improvement on defense also can be largely attributed to him when considering the level of high danger chances that the Sharks were allowing with him on the ice in the first 22 games Vs. the final 60, and when considering whether or not he actually has a heavy direct impact on that. (He does)
Conclusion:
The Sharks as a whole got much worse after the system change was implemented. There are two primary reasons that their goals for percentage marginally improved.
1. Brent Burns heavily improved his play from the worst 5v5 player in the NHL in the first 22 games to a Norris level player in the final 22 games, and something in between in the other 38. This has the ability to totally tank a team or totally boost them when a player plays as any minutes as Burns, and when the player has as much of a discrepancy between their good play and their bad play.
2. The Sharks' on-ice shooting percentage heavily increased after the system change. However, I believe this is more due to regression to the mean (both in Brent Burns and in the team as a whole) than I believe it is due to the system. It makes sense that the system would marginally increase on ice shooting percentage, since the system marginally decreased shots more than it decreased high danger chances, and it actually increased scoring chances. But I do not believe the system change has anywhere near as large of an effect on our oiSH (and our GF) to conclude that our new system is responsible for it, or that the massive increase in oiSH% we saw is sustainable.
One thing to consider, is that the Sharks lost Tim Heed and Joe Thornton right around the time the system was implemented. The vast majority of the pre-system 22 were with Heed and Thornton and the vast majority of the post-system 60 were without Heed and Thornton. It is possible that that change in personnel, which removed two of our strongest possession and scoring chance driving players, had more of a negative effect on the team than our change in "system" did. I would be open to some statistical analysis looking into this as well at a different time but I am currently exhausted after all of the work I did on this.
Let’s say that our “system change” was implemented after our regular season overtime loss @ Vegas. All numbers are 5V5.
Our season results before the system change:
12-8-2 (97 point pace)
11 ROW (14th in NHL)
6th in CF/60 (60.94)
10th in CA/60 (55.23)
5th in CF% (52.46%)
10th in FF/60 (44.51)
1st in FA/60 (38.51)
4th in FF% (53.61%)
10th in SF/60 (31.75)
1st in SA/60 (27.56)
3rd in SF% (53.54%)
12th in SCF/60 (26.97)
17th in SCA/60 (25.98)
12th in SCF% (50.94%)
4th in HDCF/60 (11.71)
22nd in HDCA/60 (10.78)
9th in HDCF% (52.07%)
30th in GF/60 (1.86)
6th in GA/60 (2.04)
22nd in GF% (47.76%)
29th in oiSH% (5.87%)
15th in oiSV% (92.60%)
25th in PDO (98.5%)
For those who don’t know what this means: the team was elite at suppressing shots and carrying the rate of shots. They were very good at suppressing goals, creating scoring chances, creating shots, and carrying the rate of scoring chances. They were poor at limiting scoring chances. They were very poor at creating goals for, but this is entirely because of their poor on-ice shooting percentage.
Given the fact that an above average percentage of their scoring chances were of the high danger variety, and that their shot generation, scoring chance generation, and high danger chance generation were all above average, I’m not convinced that team really had an offensive problem. The only difference between that team and an above average offensive team (GF/60 being the only measure of offense we’re using here) was their hilariously low on-ice shooting percentage which was totally not sustainable for a team with the offensive talent that we had.
However, the argument could also be made that our goals against was unsustainable. We had the 15th best save percentage, but given the amount of HDCA we gave up relative to the amount of shots and scoring chances we gave up, one would expect our GA/60 to be due for somewhat of a regression. (Or increase in this case) The team was elite at suppressing shots but mediocre to poor at suppressing scoring chances and high danger chances.
Personally, I think the biggest factor here is Brent Burns. Burns’ offensive confidence was terrible; he only had 1 goal at this point in the season and it came in the final game of this sample size. This played a big factor in our poor on-ice shooting percentage. In addition, his defensive play was also terrible; he made boneheaded error after boneheaded error which led to a massively inflated HDCA/60; Burns’ was 3rd worst in the team in this stat behind Paul Martin, who only played in 2 games, and Melker Karlsson, who had only played half as many minutes and was our worst player all season. Burns had elite shot based metrics because his possession game was strong and he took shot after shot but the amount of crucial errors he made led to a high amount of high danger scoring chances against, and his poor shot selection and poor shooting percentage led to a high amount of shots and fenwick for without many goals actually being scored.
Below, I’ve attached the WOWYs for the Sharks and Brent Burns in that time frame. I didn’t attach every single statistic because I don’t want to flood the page with more and more stats on stats and because I think these are the most important numbers. I’ve also added where that stat would rank among all teams at that point in the season.
Sharks without/without Brent Burns, through the first 22 games:
With:
65.6% OZFO
47.75% HDCF - 24th
12.24 HDCA/60 - 28th
2.42 GA/60 - 19th
27.2 SCA/60 - 23rd
3.40% OiSH - 31st
1.32 GF/60 - 31st
33.3% GF - 31st
Without:
48.3% OZFO
56.87 HDCF% - 1st
9.86 HDCA/60 - 10th
1.8 GA/60 - 2nd
25.22 SCA/60 - 14th
7.74% oiSH - 13th
2.27 GF/60 - 20th
55.81% GF - 5th
Their possession and shot rate metrics much stronger with Burns, but were still well above average without. It’s pretty clear, though, that Burns, despite starting much more of his shifts in the offensive zone and being deployed with much stronger players, completely tanked our defensive numbers. However, it does match the eye test; he really was that bad defensively. It also makes sense that he tanked our oiSH% when his shot selection was very poor, not as dynamic as it was in past years, and he was also just unlucky on top of that.
In summary, the team for the first 22 games, before the system change, was far from bad. Brent Burns’ poor play at both ends of the ice, which was egregiously visible via the eye test, was both A) Clearly not sustainable and B) Possibly the only thing separating that team from one of the top teams in the NHL.
Our season results after the system change:
33-19-8 (101 point pace) + 4 points
29 ROW (14th in the NHL)
6th in CF/60 (60.78) +/- 0 spots, -0.16
22nd in CA/60 (59.9) -12 spots, -4.67
13th in CF% (50.38%) -8 spots, -2.08%
11th in FF/60 (44.28) -1 spot, -0.23
17th in FA/60 (42.45) -16 spots, -3.94
11th in FF% (51.05%) -7 spots, -2.56%
16th in SF/60 (31.11) -6 spots, -0.64
12th in SA/60 (30.08) -11 spots, -2.52
14th in SF% (50.84%) -11 spots, - 2.7%
10th in SCF/60 (27.67) +2 spots, +0.7
19th in SCA/60 (26.95) -2 spots, -0.97
12th in SCF% (50.66%) +/- 0 spots, -0.28%
7th in HDCF/60 (11.54) -3 spots, -0.17
21st in HDCA/60 (11.22) +1 spot, -0.44
13th in HDCF% (50.71%) -4 spots, -1.36%
9th in GF/60 (2.55) +21 spots, +0.69
23rd in GA/60 (2.61) -17 spots, -0.57
17th in GF% (49.42) +5 spots, +1.66%
9th in oiSH% (8.21%) +20 spots, +2.34%
26th in oiSV% (91.31%) -11 spots, -1.29%
20th in PDO (99.5%) +5 spots, +1%
For those who don't know what this means: The rate at which the Sharks controlled shots and high danger scoring chances got much worse. Their plain scoring chance rate generation marginally improved, but their rate of high danger scoring chance prevention and creation, along with every single one of their shot generation and their shot prevention metrics decreased. At this point in the season, they were mostly middle of the pack by all metrics; a far cry from the near elite team they were at the start of the season.
Their overall goals for percentage had improved, but this was mostly due to an improvement in their on ice shooting percentage. Their increase in on-ice shooting percentage can be mostly explained by the fact that they took far less shots, but scored on a ridiculously higher amount of those shots. A higher percentage of the shots and chances they did get were of the high danger variety, but they still created less high danger chances than they did before. It can be expected that this would lead to a small bump in their shooting percentage, but the bump in overall goals for shows that the bump in shooting percentage was a lot more than we might be able to expect.
Sharks with/without Brent Burns, through the final 60 games.
With:
59.29% OZFO
52.28 HDCF% - 10th +14 spots, + 4.53%
11.95 HDCA/60 - 25th +3 spots, +0.29
2.55 GA/60 - 20th -1 spot, -0.13
26.86 SCA/60 - 19th -4 spots, -0.34
7.20% oiSH - 25th +6 spots, +3.8% (THIS MORE THAN DOUBLED!)
2.39 GF/60 - 16th +15 spots, +1.07
48.42% GF - 21st +10 spots, +15.12%
Without:
42.1% OZFO
49.54 HDCF% - 15th -14 spots, -7.33%
2.66 GA/60 - 25th -23 spots, -0.86
27 SCA/60 - 21st -7 spots, -1.78
8.92% oiSH - 3rd +10 spots, +1.18
2.66 GF/60 - 6th +14 spots, +0.39
50% GF - 17th -12 spots, -5.81%
The team as a whole seriously improved when Brent Burns was on the ice, and got seriously worse with him off the ice, after the system change. The team was under or very close to 50% in all shot and chance based metrics without Burns on the ice while they were comfortably above 50% in all metrics without Burns on the ice before the system change. The Sharks' improvement with Burns on the ice makes a lot more sense and it has a much bigger correlation with his play - he stopped taking such low danger shots, which led to an increase in his on-ice shooting percentage. His improvement on defense also can be largely attributed to him when considering the level of high danger chances that the Sharks were allowing with him on the ice in the first 22 games Vs. the final 60, and when considering whether or not he actually has a heavy direct impact on that. (He does)
Conclusion:
The Sharks as a whole got much worse after the system change was implemented. There are two primary reasons that their goals for percentage marginally improved.
1. Brent Burns heavily improved his play from the worst 5v5 player in the NHL in the first 22 games to a Norris level player in the final 22 games, and something in between in the other 38. This has the ability to totally tank a team or totally boost them when a player plays as any minutes as Burns, and when the player has as much of a discrepancy between their good play and their bad play.
2. The Sharks' on-ice shooting percentage heavily increased after the system change. However, I believe this is more due to regression to the mean (both in Brent Burns and in the team as a whole) than I believe it is due to the system. It makes sense that the system would marginally increase on ice shooting percentage, since the system marginally decreased shots more than it decreased high danger chances, and it actually increased scoring chances. But I do not believe the system change has anywhere near as large of an effect on our oiSH (and our GF) to conclude that our new system is responsible for it, or that the massive increase in oiSH% we saw is sustainable.
One thing to consider, is that the Sharks lost Tim Heed and Joe Thornton right around the time the system was implemented. The vast majority of the pre-system 22 were with Heed and Thornton and the vast majority of the post-system 60 were without Heed and Thornton. It is possible that that change in personnel, which removed two of our strongest possession and scoring chance driving players, had more of a negative effect on the team than our change in "system" did. I would be open to some statistical analysis looking into this as well at a different time but I am currently exhausted after all of the work I did on this.