Stat Nerdery

Last week our attention was piqued by the story coming out of Chicago that Marian Hossa and his agent, Rich Winter, had selected Chicago on the basis of a regression algorithm that had allowed them to winnow down the respective teams in the NHL that would attain 100 points in the regular season to a fairly high degree of accuracy.

This should not shock anyone that's a regular around these parts, as obviously the works of Gabriel Desjardins, Tom Awad, Vic Ferrari, and Alan Ryder are percolating through the NHL ranks these days.  Statistical analysis is making inroads into the game, and perhaps at the upcoming international hockey summit, greater efforts will be made to track various statistical information centrally in order to better develop and identify key aspects of the game.

The interest in these areas should increase drastically as it becomes more evident that the ability of teams, with easily available and cheap video and computer technology, to analyze every minute detail of their game play for the foreseeable future brings the ideas of Roger Neilson into the 21st century.

By way of example, in recent years we've heard tell of Ron Wilson, Rob Zettler, and Tim Hunter using video analysis on laptops behind the bench for immediate feedback to players. The Buffalo Sabres disposed of most of their scouting staff a couple of years ago and have shifted to a predominantly video scouting system.

Articles regularly crop up in media publications debating the value of statistical analyses vs. old school scouting methods, and frankly most of them are asking the wrong questions.  In a fashion similar to entrenched media vs. new media debates, the question should not be why one format of work is superior to the other, but how to integrate and allow the two to coexist optimally.

Stats like CORSI numbers (invented by Buffalo goaltending coach Jim Corsi), Zone Starts and Finishes, On Ice/Off Ice +/- ratings, team or opponent quality metrics, shot quality neutral save percentages, even strength save percentages, and GVT ratings all have their value in assessing the value of individual players.

Slowly hockey appears to be modernizing it's approach to player analysis, and eventually such statistical methods will tweak what people value in a player.  The arguments around the distinctions between NHL quality goaltenders have been raging for years, and they'll likely continue to.  Who knows how valuable a power forward is in comparison to a stud blue liner? Heck what is a power forward these days?  These are all debates that occur with regularity and interest many of us.

The applications of such methods of analysis are also far reaching in a cap controlled world.  If you can attach dollar and cents value to a players output, in terms of value to a team, then structuring a roster becomes much more of a turn-key process.

I'm curious about what other people perceive as the ideal team structure at this point.  How do you think a team needs to be built in the cap era for consistent long term competitiveness?  Recently it seems that teams that overload on young talent can get a Stanley Cup, but maintaining that for more than 1 year is a huge challenge.

Detroit has been the most consistent for years, and they seem to do well with cheap goaltending, amazingly talented top 4 forwards and top 3 defense, and then a solid mix of veterans and kids who are playing the supporting role.

If the Leafs want to follow the winning models of recent years I would argue they require the following:

1. Reliable, though not stellar goalkeeping. With Giguere, Gustavsson, Rynnas, Reimer - we should be ok here in the short and probably long term.

2. At least two all-star caliber forwards, 30+ goals, 80+ points. We'll call them forwards A and B and Phil Kessel should eventually qualify as A, so they need one more bonafide star for role B. Nazem Kadri? Tyler Bozak? Mikhail Stefanovich? Currently the Leafs have one of these but not two.

3. Two other forwards that are capable of scoring 20 to 25 goals in a season (forwards C and D).  One should be a power forward with size and decent hands who is hard to move in front of the net (C), while the other needs to be a relatively slick skill type with outside speed (D). Top 6 forward C might be Kulemin, Stålberg, Hayes, Hanson, or Caputi, while forward D might be Grabovski, Bozak, DiDomenico, or D'Amigo.  This is another situation where the Leafs do not currently have ANYONE to fill the role short term.  This season may show someone taking strides in that role.

4. The rest of the forward lines (all 8 of the players) need to be reliable defensively, and capable of playing with a physical edge. Speed is an asset, and an ability to play physically but cleanly is important. I'm thinking of players like Scott Nichol and Torrey Mitchell in San Jose, or Darren Helm in Detroit. They only need to combine for 75 goals or so (average roughly 8 to 10 goals per guy), but if they get more, the team's scoring will be more balanced and dangerous. This is where the younger players learn to play solid D while not worrying about offensive output. It's where Kulemin occasionally found himself for the past two seasons, and where Stålberg and Hanson resided for their visits last year. This is also where the veterans who are responsible defensively but don't put up stats fit in. Wayne Primeau and the like.

5. Top 2 D men that are offensively skilled but play solid positional D. Dion Phaneuf and Tomas Kaberle are the closest thing the Leafs have to that at this moment. Unfortunately, Kaberle is too soft defensively, while Phaneuf's positioning isn't always that great as he looks for huge hits far too often. Both players are also very predictable from an offensive perspective. This is where Keith and Campbell, or Boyle and Blake, or Pronger and Timmonen, or Lidstrom and Rafalski, Weber and Suter, or Doughty and Johnson will all fit in. The Leafs may not have this problem solved, which is a bit worrisome considering they spend more on their team's D than any club in the NHL.

6. Two second pairing D that are good in their own end, and chip in 15 to 20 points . The Leafs are actually set long term in this role with Schenn and Gunnarsson at this point. As far as I can tell, the two youngsters make very solid 3rd and 4th D men in the long term. Beauchemin is in this role currently, and he's decent at it, but he's a veteran and he'll be less important in 2 or 3 years when the team is more competitive. The likes of Juraj Mikus and Korbinian Holzer also likely fill this role for the Leafs longer term. Niklas Kronwall, Marc Edouard Vlasic, Matt Carle, Braydon Coburn, Brent Seabrook, Chris Phillips, Mattias Ohlund are all in this second tier.

7. Two imposing, physical shut down D men. Think along the lines of Hal Gill, Brooks Orpik, Rob Scuderi, Doug Murray, Anton Volchenkov etc. For the Leafs this would be Mike Komisarek, and eventually Keith Aulie will probably fit this role also. So again the Leafs aren't that poorly off in this set up, but they need some reinforcement longer term from a depth perspective. The next closest thing the Leafs have to an option in this role is Phil Oreskovic.

So realistically the Leafs need some more pop in their guns up front, and that may well come in time as players develop.  That being said, I think Burke and Wilson are well on their way to restructuring the Leafs into a winning organization.

Let me know your thoughts.

Do you feel that the increased statistical analysis in Hockey allows NHL teams to make better player decisions when it comes to roster management and the salary cap?