Sep 132011

I have spent a lot of time and effort putting together player ratings so I decided it was time to finally put them to good use and attempt to use them to predict results for the upcoming season.  This is my attempt at the eastern conference and time permitting I’ll tackle the western conference in the future.

To accomplish this goal I used my 3-year (2008-11) offensive (HARO+) and defensive (HARD+) ratings at 5v5close, 5v4 powerplay and 4v5 penalty kill situations and combined all of the ratings for all of the players on each team and then converted them back to goals to come up with a predicted goals for and goals against for each team.  In doing this I estimated the ice time of every player so first and second line players will have a greater weight than third and fourth line players as well as reserve players and I also estimated how I believe each team will use their players on the power play and penalty kill.  I did this largely on each players PP and PK ice time last season.

I have made a few assumptions in these predictions.  First, teams will not suffer significant injuries.  Generally speaking, I cannot predict injuries so I have to go with this assumption.  There are a few exceptions though.  For example, I predicted that Sidney Crosby would miss the start of the season and miss about 1/4 of the season. There were one or two other players (Matthew Lombardi comes to mind) that I did this for but none were of the talent level of Crosby so the effects on the results will not be dramatic.  In any event, any significant injuries that occur will have an impact on results.

The second assumption I made was how to rate rookies and second year players that may not have a lot of ice time and thus not have reliable ratings.  For rookies, for the most part I rated them as slightly below average but it varied a bit depending on whether they are a big time prospect or not.  That said, if a team has a rookie or two that has an exceptional season it could affect the accuracy of my predictions.  For second year players or players without a significant history to develop ratings from I manually adjusted their ratings if they seemed to be out of whack (i.e. I manually regressed their ratings to the mean).  Some got their ratings bumped up, some bumped down.  For the most part these guys are not going to be key players to a team so errors in their guestimates are not likely to have a significant impact on overall team predictions.

The final assumption I made was that all teams will spend an equal amount of time on the power play and on the penalty kill.  This does not happen in reality and I am sure some teams are more prone to taking penalties (and drawing penalties) than other teams but I haven’t spent any time to attempt to predict that so for now I haven’t factored it in at all.

Oh, just remembered another assumption so this is the final final assumption I want to mention.  I have not factored in quality of competition.  If a team plays in an easier or more difficult division than another team this will affect their results somewhat.

So with all that said, here are the eastern conference predictions for the 2011-12 season.

Predicted 2011-12 Actual 2010-11
Washington Capitals 248.0 224.6 23.4 224 197 27
Boston Bruins 236.1 216.2 19.9 246 195 51
Pittsburgh Penguins 241.4 222.3 19.1 238 199 39
Buffalo Sabres 242.1 225.3 16.8 245 229 16
Philadelphia Flyers 242.7 226.7 16.0 259 223 36
Montreal Canadiens 229.2 225.1 4.1 216 209 7
Tampa Bay Lightning 230.7 231.6 -0.9 247 240 7
NY Rangers 219.4 223.0 -3.6 233 198 35
Toronto Maple Leafs 241.0 248.4 -7.4 218 251 -33
NJ Devils 214.2 225.4 -11.2 174 209 -35
Florida Panthers 216.2 227.5 -11.4 195 229 -34
NY Islanders 230.5 244.6 -14.1 229 264 -35
Carolina Hurricanes 220.4 242.1 -21.7 236 239 -3
Winnipeg Jets 212.7 244.4 -31.7 223 269 -46
Ottawa Senators 195.5 258.1 -62.6 192 250 -58

As you may have noticed, I haven’t predicted won-loss records, just goals for and against which correlates fairly well with won-loss records.  I have also included last years goals for and against for reference.  Generally speaking, the good teams are at the top and the bad teams are at the bottom.  If my predictions are reasonably accurate the Capitals, Bruins, Penguins, Sabres and Flyers look like they should make the playoffs fairly easily while the Hurricanes (a bit of a surprise maybe), Jets and Senators are likely on the outside looking in come playoff time.  That leaves Montreal, Tampa, NY Rangers, Maple Leafs, Devils, Panthers and maybe the Islanders fighting for the final 3 playoff spots.  Generally speaking, that makes sense to me.

Let’s take a look at this data in a slightly different way.  Lets look at who has the greatest improvement in goal differential (GF-GA) from last season to m predictions for this upcoming season.


2011-12 2010-11
Team GF-GA GF-GA Diff
Toronto Maple Leafs -7.4 -33 25.6
NJ Devils -11.2 -35 23.8
Florida Panthers -11.4 -34 22.6
NY Islanders -14.1 -35 20.9
Winnipeg Jets -31.7 -46 14.3
Buffalo Sabres 16.8 16 0.8
Montreal Canadiens 4.1 7 -2.9
Washington Capitals 23.4 27 -3.6
Ottawa Senators -62.6 -58 -4.6
Tampa Bay Lightning -0.9 7 -7.9
Carolina Hurricanes -21.7 -3 -18.7
Pittsburgh Penguins 19.1 39 -19.9
Philadelphia Flyers 16.0 36 -20.0
Boston Bruins 19.9 51 -31.1
NY Rangers -3.6 35 -38.6

Generally speaking the teams that have the highest predicted improvement were teams that had poor seasons last year and the teams with the greatest predicted fall back are teams that had good years last year.  There is probably a regression to the mean happening here.  The good teams last year probably had some luck going their way and the teams at the bottom of the standings probably had some bad luck.

For the gainers, the Devils potential gain is fully understandable.  They had a horrendous first half of last season but played much better in the second half.  They should be closer to their second half performance this upcoming year.  The Florida Panthers spent a lot of money on free agents and should have an improved team, but still may not make the playoffs.  The Maple Leafs, Islanders and Jets are probably more in the had some bad luck last season and will regress to the mean category though their young players should be a bit better too.

The Rangers predicted fall back is a bit of a surprise considering they signed Brad Richards but they lost Drury, Frolov, Gilroy, McCabe and Prospal.  Their projected defense looks potentially very weak.  After Staal and Girardi you have Sauer, McDonagh, Erixon, Del Zotto, and Eminger all of whom are very young with little or no experience or in the case of Eminger a one time quality prospect that never really established himself as an NHL regular.

The table below shows the predicted top offensive and defensive teams.

Team GF Team GA
Washington Capitals 248.0 Boston Bruins 216.2
Philadelphia Flyers 242.7 Pittsburgh Penguins 222.3
Buffalo Sabres 242.1 NY Rangers 223.0
Pittsburgh Penguins 241.4 Washington Capitals 224.6
Toronto Maple Leafs 241.0 Montreal Canadiens 225.1
Boston Bruins 236.1 Buffalo Sabres 225.3
Tampa Bay Lightning 230.7 NJ Devils 225.4
NY Islanders 230.5 Philadelphia Flyers 226.7
Montreal Canadiens 229.2 Florida Panthers 227.5
Carolina Hurricanes 220.4 Tampa Bay Lightning 231.6
NY Rangers 219.4 Carolina Hurricanes 242.1
Florida Panthers 216.2 Winnipeg Jets 244.4
NJ Devils 214.2 NY Islanders 244.6
Winnipeg Jets 212.7 Toronto Maple Leafs 248.4
Ottawa Senators 195.5 Ottawa Senators 258.1

It is probably not a surprise that the Capitals, Flyers, Sabres, Penguins, Bruins and Lightning are among the top offensive teams but it is interesting to see the Maple Leafs move up the offense list.   It is a common belief that the Leafs late season success last season was because of the play of goalie James Reimer and Reimer did play a part, but in reality, much of the reason for the success was actually due to the fact that the Leafs scored a lot of goals.  Add Connolly and Liles into the mix and the Leafs can put out three lines who can score so while they may not have the elite offensive players some of the other teams have, they have depth (not unlike the Bruins actually whose top point producer was Krejci with just 62 points – Kessel had 64 for the Leafs).  Defensively it seems the Leafs may continue to struggle.  They are not a good defensive team and they desperately need to figure out how to improve their penalty kill.  Defense could be a problem, even with improved goaltending (which may or may not be reality – Reimer had success over a somewhat small sample size and Gustavsson has never performed well).

It is probably worth saying a word or two about the Ottawa Senators.  It seems they will struggle to score and will struggle to keep the puck out of their own net.  The Senators may be in for a tough season but it will be a season of evaluation of young players and hopefully (for Sens fans) progress.  On any given night they will potentially have 6-8 rookies in the lineup.  Expect to see rookie forwards Bobby Butler, Mika Zibanejad, Erik Condra, Colin Greening, Zack Smith, Nikita Filatov and Stephane Da Costa in the line up through out the season as well as defensemen Jared Cowan, David Runblad, Patrick Wiercioch.  If some of these guys are truly ready to become solid NHL regulars they might not be as bad as the above tables suggest, but they will still likely be competing for the first overall draft pick (which is probably a good thing for them anyway)

Finally, let me suggest that you not all take these too seriously.  While I do think there is some merit to these predictions, if you think your team is ranked too low or another team is ranked too high, no need to have a fit over it.  I really don’t know how accurate they are and a lot can happen to alter what really happens anyway.  I wanted to post these in part to generate a discussion but also in part so I can track these predictions as the season progresses and come the end of the season look back see how well this unbiased, mostly mathematical prediction system performs.