Calgarypuck Forums - The Unofficial Calgary Flames Fan Community - View Single Post - A statistical look at line combos and D pairings

Kovaz · 12-19-2017, 11:24 AM

There's been a fair bit written about our play as a team, and specifically how our results have lagged behind our play. Any time you're not scoring as much as you should be, a variety of explanations get thrown around. "We're shooting too much from the perimeter." "We're not getting the bounces." etc. You also see a lot of vitriol thrown at specific players or lines. So, I thought I'd take a closer look, line-by-line, to see which combos are driving trends in our overall results. All data taken from corsica.hockey.

Before I get into the charts, some quick explanations. These are based on my belief that you can break generating goals into three separate factors: shot volume, shot quality, and finishing.

Shot Volume being the total number of shot attempts directed towards the net. This is a reasonable proxy for offensive zone time, so a high shot volume implies consistent zone entries and good puck retrieval and forechecking. I used Shot Attempts per 60 minutes as my metric for this.
Shot Quality being the likelihood that any given shot attempt will score. This should ideally measure the ability of a line to generate good scoring chances when they have the puck in the offensive zone. I used Expected Goals per Shot Attempt to attempt to measure this.
Finishing is simply how often a line is actually scoring on the chances they are getting. A line that's getting lucky bounces and picking corners will have a high finishing score, where a line that's hitting posts and getting robbed will have a low finishing score. I used Goals per Expected Goal to measure this.

The nice part about using those 3 stats in particular is if you multiply the 3 of them together, you get:

(Shot Attempts / 60 minutes) X (Expected Goals / Shot Attempt) X (Goals / Expected Goal) = (Goals / 60 minutes)

Which is exactly the thing we want to maximize. So essentially, we're starting with each line's goals for and against, and using those 3 metrics to explain what's driving a high or low goal rate. Defensively I used the same 3 metrics - the only difference being that I labeled Finishing "Goaltending (inverse)".

The likely biggest shortcoming of this approach is the quality of the Expected Goals calculation. It's quite possible that it doesn't completely capture shot quality - so there's likely some shot quality effects in the finishing scores I calculated. That said, it's the best I'm aware of, and the results were interesting.

NOTES

All values in the charts are normalized against the league average
Shot Volume and Goals are display as rates per 60 minutes of ice time
All values are setup so that a bigger bar = more goals (so for offensive stats, to the right is good. Defensive stats, to the left is good)
EDIT: All numbers are 5v5

First, the team as a whole:

We generate significantly more shot volume than we allow, while shot quality is pretty much a wash - we generally keep shot quality low in both directions. However, our finishing has been pretty bad while our goaltending has been pretty much league average overall.

The top line is above the team numbers in shot volume, shot quality, and finishing overall - unsurprising for our runaway offensive leaders. Defensively, they allow a lower volume but a higher quality than the rest of the team, but they've gotten some excellent goaltending to bail them out. Overall unsurprising numbers, and they support the eye test so far.

Here's where things start to get interesting. The 3M line generates a completely ridiculous shot volume, which should be unsurprising considering they have the puck about 75% of every shift. However, their shot quality is significantly below the team average, AND they can't score a goal to save their lives. They're easily the biggest drivers of our team's low finishing numbers, and that also explains Frolik's drop in scoring pace. Them turning it around offensively could be all we need to make a jump 5v5

The final 2 lines that met the ice time cutoff. The Hathaway line is absolutely shooting the lights out, while the Jagr version was a bit snake-bitten. With Hathaway they've been much stronger defensively while generating more shot volume - whereas with Jagr their shot quality was through the roof. Looking at this, I'd be in no rush to bump Hathaway off that line. Also I'd add the caveat that easier zone starts and competition could be a factor here.

The 4th liners

None of our 4th line combinations have stuck together long enough to meet the ice time cutoffs, so I looked at just their individual stats. The big theme here is lack of offense, in particular the finishing. I think it's pretty much a given that our 4th line is going to get out-shot with the personnel we have available, but I think the defensive stats are at least encouraging. Stajan in particular has put up respectable numbers - if my 4th line is suppressing both shot volume and shot quality I'm pretty happy. This also confirms some of what my eye test was telling me: specifically that Versteeg's 5v5 play has left a lot to be desired, and F. Hamilton is our weakest player.

The top pair's numbers are similar to the team as a whole: crazy high shot volume, ok shot quality, poor finishing. They also suppress both shot volume and quality quite well, but they've been let down a bit by the goaltending.

The second pair's poor start is echoed by the underlying numbers, but it's not as bad as it could be. Defensively, they've been about league-average but have received some excellent goaltending. However, offensively their shot quality leaves a lot to be desired and our finishing with them on the ice has been even worse.

After years of having the 3rd pair be a total tire fire, it's so nice to have a 3rd pair with a positive goal differential and strong underlying numbers. The only somewhat sore spot is the shot volume against, but if your 3rd pair is above league average you're in a great spot.

A popular request for a 2nd pairing, and there's some merit to it. Brodie-Stone have massively stronger offensive numbers than Brodie-Hamonic, while defensively they're a bit weaker due to much higher shot volume against, offset largely by lower shot quality.

The final pair with enough ice time to make the list, Bartkowski-Stone simply couldn't generate a thing offensively. Defensively the numbers were ok at least, with a very high shot volume against but solid shot quality suppression. But it's night and day when you replace Bartkowski with Kulak.

So there you have it. When you break it down by line, I think there are some interesting nuggets of information hidden in there. The biggest conclusions I'd draw are:

We need the 3M line to start scoring again. They are almost singlehandedly driving our "played well, but lost" feeling as of late
Keep Hathaway with Bennett-Jankowski. That line has excellent stats across the board, and it wouldn't shock me if they could handle some tougher competition
The 4th line is going to be a weakness any way you slice it, but at the very least they haven't been as bad defensively as I thought. Their poor numbers are more due to a lack of finishing talent than anything (although they still get hemmed in their own zone more often than I'd like)
I think Brodie-Hamonic as a pair are improving as of late, and I'd like to take a closer look at the trends in their numbers, but Brodie-Stone have definitely had better numbers to this point in the season
Replacing Bartkowski with Kulak on the 3rd pair has fixed one of our biggest achilles heels over the last few years

Also I'm open to feedback on the charts, data presented, process, etc.

12-19-2017, 11:24 AM	#1
Kovaz Scoring Winger Join Date: Nov 2016 Exp:	A statistical look at line combos and D pairings There's been a fair bit written about our play as a team, and specifically how our results have lagged behind our play. Any time you're not scoring as much as you should be, a variety of explanations get thrown around. "We're shooting too much from the perimeter." "We're not getting the bounces." etc. You also see a lot of vitriol thrown at specific players or lines. So, I thought I'd take a closer look, line-by-line, to see which combos are driving trends in our overall results. All data taken from corsica.hockey. Before I get into the charts, some quick explanations. These are based on my belief that you can break generating goals into three separate factors: shot volume, shot quality, and finishing. Shot Volume being the total number of shot attempts directed towards the net. This is a reasonable proxy for offensive zone time, so a high shot volume implies consistent zone entries and good puck retrieval and forechecking. I used Shot Attempts per 60 minutes as my metric for this. Shot Quality being the likelihood that any given shot attempt will score. This should ideally measure the ability of a line to generate good scoring chances when they have the puck in the offensive zone. I used Expected Goals per Shot Attempt to attempt to measure this. Finishing is simply how often a line is actually scoring on the chances they are getting. A line that's getting lucky bounces and picking corners will have a high finishing score, where a line that's hitting posts and getting robbed will have a low finishing score. I used Goals per Expected Goal to measure this. The nice part about using those 3 stats in particular is if you multiply the 3 of them together, you get: (Shot Attempts / 60 minutes) X (Expected Goals / Shot Attempt) X (Goals / Expected Goal) = (Goals / 60 minutes) Which is exactly the thing we want to maximize. So essentially, we're starting with each line's goals for and against, and using those 3 metrics to explain what's driving a high or low goal rate. Defensively I used the same 3 metrics - the only difference being that I labeled Finishing "Goaltending (inverse)". The likely biggest shortcoming of this approach is the quality of the Expected Goals calculation. It's quite possible that it doesn't completely capture shot quality - so there's likely some shot quality effects in the finishing scores I calculated. That said, it's the best I'm aware of, and the results were interesting. NOTES All values in the charts are normalized against the league average Shot Volume and Goals are display as rates per 60 minutes of ice time All values are setup so that a bigger bar = more goals (so for offensive stats, to the right is good. Defensive stats, to the left is good) EDIT: All numbers are 5v5 First, the team as a whole: We generate significantly more shot volume than we allow, while shot quality is pretty much a wash - we generally keep shot quality low in both directions. However, our finishing has been pretty bad while our goaltending has been pretty much league average overall. The top line is above the team numbers in shot volume, shot quality, and finishing overall - unsurprising for our runaway offensive leaders. Defensively, they allow a lower volume but a higher quality than the rest of the team, but they've gotten some excellent goaltending to bail them out. Overall unsurprising numbers, and they support the eye test so far. Here's where things start to get interesting. The 3M line generates a completely ridiculous shot volume, which should be unsurprising considering they have the puck about 75% of every shift. However, their shot quality is significantly below the team average, AND they can't score a goal to save their lives. They're easily the biggest drivers of our team's low finishing numbers, and that also explains Frolik's drop in scoring pace. Them turning it around offensively could be all we need to make a jump 5v5 The final 2 lines that met the ice time cutoff. The Hathaway line is absolutely shooting the lights out, while the Jagr version was a bit snake-bitten. With Hathaway they've been much stronger defensively while generating more shot volume - whereas with Jagr their shot quality was through the roof. Looking at this, I'd be in no rush to bump Hathaway off that line. Also I'd add the caveat that easier zone starts and competition could be a factor here. The 4th liners None of our 4th line combinations have stuck together long enough to meet the ice time cutoffs, so I looked at just their individual stats. The big theme here is lack of offense, in particular the finishing. I think it's pretty much a given that our 4th line is going to get out-shot with the personnel we have available, but I think the defensive stats are at least encouraging. Stajan in particular has put up respectable numbers - if my 4th line is suppressing both shot volume and shot quality I'm pretty happy. This also confirms some of what my eye test was telling me: specifically that Versteeg's 5v5 play has left a lot to be desired, and F. Hamilton is our weakest player. The top pair's numbers are similar to the team as a whole: crazy high shot volume, ok shot quality, poor finishing. They also suppress both shot volume and quality quite well, but they've been let down a bit by the goaltending. The second pair's poor start is echoed by the underlying numbers, but it's not as bad as it could be. Defensively, they've been about league-average but have received some excellent goaltending. However, offensively their shot quality leaves a lot to be desired and our finishing with them on the ice has been even worse. After years of having the 3rd pair be a total tire fire, it's so nice to have a 3rd pair with a positive goal differential and strong underlying numbers. The only somewhat sore spot is the shot volume against, but if your 3rd pair is above league average you're in a great spot. A popular request for a 2nd pairing, and there's some merit to it. Brodie-Stone have massively stronger offensive numbers than Brodie-Hamonic, while defensively they're a bit weaker due to much higher shot volume against, offset largely by lower shot quality. The final pair with enough ice time to make the list, Bartkowski-Stone simply couldn't generate a thing offensively. Defensively the numbers were ok at least, with a very high shot volume against but solid shot quality suppression. But it's night and day when you replace Bartkowski with Kulak. So there you have it. When you break it down by line, I think there are some interesting nuggets of information hidden in there. The biggest conclusions I'd draw are: We need the 3M line to start scoring again. They are almost singlehandedly driving our "played well, but lost" feeling as of late Keep Hathaway with Bennett-Jankowski. That line has excellent stats across the board, and it wouldn't shock me if they could handle some tougher competition The 4th line is going to be a weakness any way you slice it, but at the very least they haven't been as bad defensively as I thought. Their poor numbers are more due to a lack of finishing talent than anything (although they still get hemmed in their own zone more often than I'd like) I think Brodie-Hamonic as a pair are improving as of late, and I'd like to take a closer look at the trends in their numbers, but Brodie-Stone have definitely had better numbers to this point in the season Replacing Bartkowski with Kulak on the 3rd pair has fixed one of our biggest achilles heels over the last few years Also I'm open to feedback on the charts, data presented, process, etc. Last edited by Kovaz; 12-19-2017 at 01:45 PM.