16 July 2012

The Pain Game 2011/12 - an attempt to use GVT

Throughout the regular season, my usual breakdown of the extent of injury/illness absences on NHL teams uses cap hit (CHIP) and TOI/G (AMIP) as weights to apply to each player, in an attempt to quantify the significance of any absences.  A fuller description is found with the end of year figures for 2011/12.

Both methods have more than a few imperfections (as noted) but offer a reasonable indication.  As a third - and probably final - weighting factor, I thought I'd try using GVT, as developed by Tom Awad and described at hockeyprospectus.com.  In disgracefully crude short form, this is a metric placing a relative value on a player's contribution to goals scored and conceded by his team.

So, rather than cap hit, for example, I've just used instead a player's average GVT per game (or for goalies, GVT per game available for) in the 2011/12 regular season (taken from www.hockeyabstract.com) as the weighting, producing a cumulative team figure I'll imaginatively call GVTIP.

  • For players that missed the whole or the vast majority of the season, I've used the equivalent 2010/11 (or 2009/10) GVT figure.
  • Unlike TOI/G used in deriving AMIP, the GVT figure does not account for any differences between the figures accrued with different teams during the season for players traded mid-season etc.
  • The main drawback is sample size, in that GVT in only a single year (or part year) is likely to be a pretty volatile indicator of a player's relative value, so there are more than a few examples where a player's contribution - perhaps based on a handful of games only - to team GVTIP skews the result completely.  So I'm not guaranteeing it is foolproof by any means...
The reworked 2011/12 table, ranked by GVTIP with the other metrics for comparison:

  •  Of the top three CHIP numbers, GVT is much more favourable towards the Penguins missing Crosby, Letang, Staal and Malkin (contributing about 41 goals of their deficit between them) than the Canadiens going without Gomez, Markov and Gionta and the Blue Jackets being deprived of Huselius, Martinek, Methot, Carter and Tyutin, for example
  • The fact that Gomez and Markov each managed to get a negative GVT in their short periods of active duty actually imply that the Canadiens were several goals better off by them not being in the line-up for most of the year (told you there were distortions...) - believable for Gomez, perhaps not for Markov
  • On a similar theme, most of the Islanders' absences are pretty insignificant in terms of GVT, except for Mr Anomaly himself, Rick DiPietro, whose stellar play in the three seconds of health he managed to muster skews the team number into strongly negative territory all by himself
  • Conversely, the Blackhawks have a low CHIP number but are way up on GVTIP, largely due to Toews (fair enough) and Carcillo (okaaaayyyyy...)
  • As I think would be expected - and partly expecting some of the relative movements above - the GVTIP table is more tilted towards playoff teams with higher team GVT at the top end. Perhaps an argument for using share of team GVT as the weighting of player value at a team level?

9 July 2012

The Pain Game - 4-year analysis (2008/09 - 2011/12)

Another brief data dump of figures aggregated over the last four NHL regular seasons, surprisingly enough updated to include 2011/12.

Analyses of individual seasons (including explanations of the figures and their limitations) plus my equivalent aggregate analysis (aggreganalysis?) are at the following links:

Firstly, an updated ranking of teams by aggregate CHIP over the last four seasons (playoff teams highlighted; click to enlarge image):

Still no obvious correlation between playoff qualification and the degree and quantity of injuries/illnesses suffered.  The Islanders, Oilers and Wild (all long-term playoff absentees) sat in the top three positions for the table covering 2008/09 to 2010/11, but have all slipped down slightly since, mostly due to Messrs Markov and Crosby, plus the Flyers' rotating cast of retired-but-not-really-retired veterans.

The same figures grouped by division:

A bit over-simplistic, but still fairly interesting that the supposedly travel-weary Pacific has suffered both the fewest man-games lost and the lowest CHIP in each of the last four seasons (OK, two man-games away in 2008/09 from doing so), while the Atlantic has routinely been hit the hardest.  Maybe 95% due to Rick DiPietro, but interesting anyway...

The largest ten and smallest ten CHIP figures by a team in a single season over the last four:

No changes at all in the bottom 10.  The Canadiens in the year just gone comfortably surpassed the Blues' previous high watermark, although worth noting that their CHIP number was still a lower proportion of the overall cap compared to the Blues' number from three years before.

The largest ten CHIP figures accumulated by a player in a single season over the last four:

None too surprisingly, Crosby and Markov each make their second appearance here, joined by Kristian Huselius who had the good fortune to only have to play in two games for the Blue Jackets in 2011/12.