September 02 2016 10:59AM
Alex Novet did some really fantastic work on neutral zone playing styles recently, taking Corey Sznajder's incredible work on zone entries and turning it into something that can help define player roles.
For this article I'll be drawing on Novet's work once again in attempt to better define player chemistry.
What I did for this article is take the 30 forward lines and 30 defense pairs that played the most time together in the 2013-2014 season (from which Novet's and Sznajder's work is based) and looked at how each line and pairing was composed based on the neutral zone playing style of each player. I then did the same with the 30 forward lines (minimum 70 minutes played together) and and 30 defense pairs (minimum 110 minutes played together) that played together the least and compared.
The idea here is that if a forward line or defense pair played a lot together, they must either have chemistry, or at the very least, perceived chemistry by the coaching staff. Otherwise, why would they be kept together? I then applied the same logic in reverse for the infrequent forward lines and defense pairs.
There are obvious limitations to this. For one, top-end players are likely to get more ice-time and lower-end players are more likely to get less ice-time, which might make a given player more or less likely to show up on either the frequent list or the infrequent list. Also, the type of neutral zone player someone is playing with might influence their own role. For example, if a player is playing with two Drivers, that player might have less opportunity to attempt carry-ins, which can influence how they're defined. Of course, different systems and the ways different teams attempt to gain entry into the offensive zone is yet another influence on possible roles and line/pairing composition. A player in one system might be a Driver, but put them in another system and they might be an Opportunist.
Still, I think I found some things that are at least intriguing, so I went ahead anyways and listed some possible suggestions that the evidence makes.
You can see the full list of most frequent, and infrequent, forward lines and defense pairs at the bottom of this article.
Here's what I found:
- 20 of the 30 most frequent forward lines had at least 2 of the same player type (14 Driver, 5 Balanced, 1 Passenger).
- 22 of the 30 most infrequent forward lines had at least 2 of the same player type on them (9 Driver, 6 Passenger, 5 Balanced, 2 Opportunist).
This suggests that if you're going to have two of the same player types on a line, it might be better to have those players be Drivers or Balanced and not Passengers.
- Of the 14 lines with 2 or more Drivers on the most frequent forward lines, the other player type was: Opportunist 7 times, Balanced 5 times, Driver 1 time, and not known 1 time.
- Of the 9 lines with 2 or more Drivers on the infrequent forward lines, the other player was Balanced 4 times, an Opportunist 2 times, a Passenger 1 time, a Driver 1 time, and a Dump Truck 1 time.
This suggests that Drivers may work particularly well with Opportunists (and to a lesser extent Balanced players).
- Of the 5 lines with 2 or more Balanced players on the most frequent forward lines, the other player was Balanced 2 times, a Driver 2 times, and a Passenger 1 time.
- Of the 5 lines with 2 or more Balanced players on the infrequent forward lines, the other player was a Passenger 2 times, a Driver 1 time, an Opportunist 1 time, and a Dump Truck 1 time.
Balanced players may go better with other Balanced players or with Drivers, though this one might not tell us anything.
- On the line with 2 Passengers on the most frequent forward lines, the other player was Balanced.
- Of the 6 lines with 2 or more Passengers on the infrequent forward lines, the other player was Balanced 2 times, a Dump Truck 2 times, a Passenger 1 time, and an Opportunist 1 time.
Having a line with multiple Passengers on it intuitively seems like a bad idea, and the evidence would seem to suggest the same.
- There were no lines with multiple Opportunists among the most frequent forward lines.
- Of the 2 lines with 2 or more Opportunists on the infrequent forward lines, the other player was a Driver both times.
Having a line with multiple Opportunists on it might also be a bad idea, but this one needs to be looked at further.
- Of the 14 lines with 2 or more Drivers on the most frequent forward lines, the Drivers were on the center and wing 10 times, and both wings 4 times.
- Of the 9 lines with 2 or more Drivers on the infrequent forward lines, the Drivers were a winger and center 7 times, and a winger and winger 2 times.
It doesn't look like there's a whole lot going on here, which would suggest a successful Driver-Driver combination is more dependent on the playing style of the third player on that line, rather than which position each player is playing.
- Of the 5 lines with 2 or more Balanced players on the most frequent forward lines, the Balanced players were on center and wing 3 times, and on both wings 2 times.
- Of the 5 lines with two or more Balanced players on the infrequent forward lines, the Balanced players were a center and winger 4 times, and a winger and winger 1 time.
Balanced-Balanced combinations potentially work better when the players are at center and wing rather than both wings, but this more likely suggests yet again that the playing style of the third player on that line is more important than the positional composition of a line.
- On the line with 2 Passengers on the most frequent forward lines, both players were on the wing.
- Of the 6 lines with 2 or more Passengers on the infrequent forward lines, the Passengers were a center and winger 4 times, and a winger and winger 2 times.
Maybe Passenger-Passenger combinations work better on the wing.
- There were no lines with multiple Opportunists on the most frequent forward lines.
- Of the 2 lines with 2 Opportunists on the infrequent forward lines, the Opportunists were a center and winger 1 time, and on both wings 1 time.
Again, it might not be a good idea to have multiple Opportunists on the same line. Other than that more research would be needed to see if anything at all is going on here.
- Of the 60 wingers on the most frequent forward lines, 24 were Drivers, 21 were Balanced, 6 were Dump Trucks, 5 were Passengers, 3 were Opportunists, and 1 was not known.
- Of the 60 wingers on the infrequent forward lines, there were 20 Drivers, 14 Passengers, 13 Balanced players, 8 Dump Trucks, 4 Opportunists, and 1 player not known.
We see more Passengers and less Balanced players on the wing in the infrequent forward group, which might suggest Drivers and Balanced wingers are preferable. However, it might simply speak to the fact that good players are more likely to be Drivers or Balanced, and thus likely to get more ice-time.
We also get the suggestion that wingers seem less likely to be Opportunists or Dump Trucks.
- Of the 30 centers on the most frequent forward lines, 12 were Drivers, 8 were Opportunists, 6 were Balanced, 4 were Passengers, and 0 were Dump Trucks.
- Of the 30 centers on the infrequent forward lines, there were 10 Opportunists, 9 Drivers, 6 Passengers, 5 were Balanced players, and 0 were Dump Trucks.
It doesn't look like there's enough of a difference here to conclude anything chemistry-wise, but we do get a look at another suggestion, which would be that centers are more likely to be Drivers or Opportunists, and quite unlikely to be Dump Trucks. However, this would need to be looked at more.
- Of the 10 lines without any common player types on the most frequent forward lines, 9 of the 10 lines had a Balanced player, 6 of 10 lines had a Passenger, 6 of 10 lines had a Dump Truck, 5 of 10 lines had a Driver, and 4 of 10 lines had an Opportunist.
- Of the 8 forward lines with no common player types on the infrequent forward lines, 7 lines had a Driver, 6 lines had an Opportunist, 4 lines had a Dump Truck, 4 lines had a Passenger, and 2 lines had a Balanced player. One player in the infrequent forward lines group was not known.
Balanced players are on 90% of the most frequent forward lines that have no common player types, but just 25% of lines with no common player types among the infrequent forward lines. This suggests that Balanced players might be a bit of a Swiss Army knife chemistry-wise.
- Of the 10 lines without any common player type on the most frequent forward lines, 4 lines had a combination of a Driver, a Passenger, and a Balanced player type; 3 lines had a combination of a Dump Truck, Passenger, and a Balanced player type; 3 lines had a Dump Truck and an Opportunist, with the other player being Balanced twice and a Driver once.
- Of the 8 uncommon lines among the infrequent forward group, 3 of the lines had a combination of a Driver, Dump Truck, and Opportunist, 2 of the lines had a combination of a Driver, Passenger, and Opportunists, 1 line had a combination of a Driver, Balanced player, and Opportunists, 1 line had a combination of a Driver, Passenger, and Balanced player, and 1 line had a combination of a Dump Truck, a Passenger, and an unknown player.
Here we get a look at some potentially successful combinations. Lines with a Driver, Passenger, and Balanced player might be more likely to succeed, along with those that contain a Passenger, Balanced player, and a Dump Truck. Conversely, combinations such as Driver, Dump Truck and Opportunist might not work as well.
Now on to the defense pairings:
- 7 of the 30 most frequent pairs had the same player type on them (5 Balanced, 2 Driver).
- 7 of the 30 infrequent pairs had the same player type on them (2 Balanced, 2 Driver, 2 Passengers, 1 Dump Truck).
Common player types was on the majority of forward lines (both frequent and infrequent), but are on the minority of defense pairs (both frequent and infrequent). This suggests either that A) chemistry on defense works better with different player types, or B) that this phenomenon is largely a result of the fact that forward lines have 3 players on them and defense pairs have 2, meaning there's more of a possibility for common players to be on the same line than the same pair. Or it could be some combination of both A and B.
- Of the 60 defensemen on the most frequent pairs, there were 23 Balanced players, 16 Drivers, 14 Passengers, 4 Opportunists, and 3 Dump Trucks.
- Of the 60 defensemen on the infrequent pairs, there were 16 Drivers, 15 Balanced players, 13 Passengers, 7 Dump Trucks, 5 Opportunists, and 4 unknown players.
It's possible that Balanced defensemen work a little better for chemistry than Drivers, but this would need to be looked into more.
Here's the full composition rundown of both the most frequent, and the infrequent, defense pairs:
This would definitely need to be looked into more, but some guesses we can make are that Balanced-Balanced, Driver-Passenger, and Balanced-Passenger pairings might be a recipe for success, while pairings such as Passenger-Passenger and Dump Truck-Dump Truck might be a bad idea.
The Leafs' line of Van Riemsdyk-Bozak-Kessel was the second most frequent forward line in 2013-2014, which had a composition of Driver-Opportunist-Driver. I definitely think this gets at part of why they worked well together, with Van Riemsdyk and Kessel driving the offense, Bozak was good enough as a player to let them do the heavy lifting but still chip in when necessary.
The team also had a line on the infrequent forwards list, which was Raymond-Bolland-Clarkson, who had a composition of Driver-Passenger-Balanced. It's not surprising this line didn't work well together, given that Raymond was a lot more offensively apt than Bolland and Clarkson, yet not good enough to carry a line on his own.
Unsurprising to Leafs fans that remember this team well, Gunnarsson-Phaneuf were one of the most frequent defense pairings in the NHL that season, with a composition of Passenger-Opportunist. You'd need to dig into their underlying numbers a little more to explain why the pairing wasn't great, but certainly the fact that neither was able to drive offense well enough seems like it was part of the problem. Playing in Randy Carlyle's system probably didn't help either.
It's interesting to think about next year's team, even though we don't have clearly defined roles for them. The Leafs figure to have a lot of potential Drivers up front, with the likes of Van Riemsdyk, Kadri, Matthews, Marner, and Nylander potentially leading the offensive charge for many seasons.
On the back end, Rielly and Gardiner can help drive the offense, with Zaitsev and Maricin also potentially helping out. I do worry though that the Hunwick-Rielly pairing will be reunited, which struggled to drive play last season. Hopefully that's something Babcock avoids.
So, what does chemistry look like? My grand, sweeping conclusion here is that a lot more work needs to be done on defining player roles using numbers. Most especially, I'm only looking at player roles in the neutral zone, and it turns out there are two other zones in hockey as well. Things like tracking the location of shots, Ryan Stimson's passing project, and tracking where players are moving on the ice can all play a big part in defining player roles in both the defensive and offensive zone. I think once we start to paint a fuller picture of what roles certain players fill, we can really start to get a much more comprehensive view of what player chemistry really looks like. Which is exciting, because that could have particularly powerful applicability for coaches in terms of assembling an effective lineup.
Most Frequent Forward Lines, 2013-2014
Infrequent Forward Lines, 2013-2014
Most Frequent Defense Pairs, 2013-2014
Infrequent Defense Pairs, 2013-2014