Thursday, June 25, 2015

Why History Suggests Carey Price Deserved the Hart Trophy

I'm generally a fan of the Washington Post's Neil Greenberg, but in the wake of Carey Price sweeping the NHL awards last night I feel obligated to respond to an argument that he has championed this season.  It's been expressed in a few different ways (and by other people as well, of course), but generally has come down to some version of the claim that since goalies have rarely won the Hart Trophy historically, Alex Ovechkin should have been the expected and possibly more deserving Hart winner.

To quote Greenberg from yesterday:
"But would the voters name Price as the league's most valuable player over Ovechkin?  History shows they haven't in the past."
Obviously the voters did overwhelmingly name Price as the league's MVP (and so did the players).  But Greenberg was not so much predicting the result as making a claim from historical evidence, which means that it is not the presence of hindsight that allows me to challenge his assertion.  I'm not interested in criticizing anybody because their guy didn't win, but I am quite interested in what history says about the relative comparison of forwards vs. goalies, since that is an interesting problem of player valuation.  My contention is that history actually shows that the voters were completely consistent in backing Price, and would in fact have been just as likely in years past to choose a goalie season equivalent to Price's over a scoring season like Ovechkin's.

It is true that goalies have rarely won the Hart (Price's win was just the fourth time a goalie has won in the past 50 years).  However, I wouldn't necessarily say that Hart voters are hugely biased against goaltenders.  It is pretty clear that there is a bias towards forwards, and a strong bias against defencemen.  If someone wanted to argue that an elite defenceman will lose the Hart to a top forward, that would be a very safe prediction because defencemen typically never finish anywhere near the top in the Hart standings (a pattern that continued last night with Norris winner and most valuable defenceman Erik Karlsson coming in a lowly ninth behind five forwards and three goalies).  In contrast, goalies may not win very often but they do routinely get significant Hart support.  Since Ray Bourque narrowly lost the 1990 Hart Trophy to Mark Messier, Chris Pronger in 2000 is the only defenceman to finish in the top three in Hart voting.  In that time, 17 goalies have been similarly nominated.

Looking at past years where a goalie won the Hart, it is notable that it was usually coupled with the lack of an elite scoring season by a forward.  In 1961-62, Bobby Hull and Andy Bathgate tied for the scoring lead at 84 points, the second-lowest Art Ross winning total of the decade, surely a factor in Jacques Plante's MVP selection.  When Dominik Hasek won his first Hart in 1996-97, Mario Lemieux's 122 points was the lowest Art Ross total in 27 years, even though it was actually still a fairly strong season in context (that season marked the start of the so-called "Dead Puck Era", and voters may not yet have fully adjusted to the new scoring environment).  In 1997-98, Jaromir Jagr missed 19 games yet still won the Art Ross with 102 points, an even lower total that was no match for Hasek's peak season.  And in 2001-02, Jose Theodore snuck off with the Hart when Jarome Iginla led the league with 96 points, a mark that ranks as the third-lowest Art Ross winning total in a full season since expansion.

This is what the "goalies rarely win the Hart" argument is missing.  The reason that goalies rarely win the Hart is that elite forward seasons will almost always beat elite goalie seasons, and most of the time there is at least one forward out there having a really strong year.  The reason that Roberto Luongo and Martin Brodeur lost out on the Hart in 2006-07 was not that they were undeserving or that the writers weren't willing to vote for them, it was that Sidney Crosby had a 120 point season.  Luongo (who finished 2nd) and Brodeur (3rd) both vastly outpointed the rest of the field in Hart balloting, including high-scorers such as Joe Thornton (114 points), Vinny Lecavalier (108 points) and Dany Heatley (105 points).  Similarly, Ryan Miller's '09-10 campaign could have easily been a Hart threat in other seasons, it just happened to coincide with three forwards all scoring 109 or more (even though Miller did beat out a number of strong forwards seasons in Hart voting, including Stamkos' 51 goals and Kane's 88 points for one of the league's top teams in the stronger Western Conference).  An elite high-scoring forward will almost always win the Hart (pretty much barring Hasek-level goalie dominance), but in the absence of such a forward season (if, for example, the Art Ross winner scores a mere 87 points), a goalie having an elite season will absolutely become an MVP favourite.

It could be argued that voters have a pattern of tokenism for goalies, and while some might be willing to list a goalie in 2nd or 3rd place on their ballots the majority have been consistent in reserving the top spot for the Art Ross or Rocket Richard winner.  I don't think there is much evidence that this is the case.  To me, it looks like the most valuable goalies have consistently finished high in voting, and the determining factor of whether they come close to winning or not is the level of production of the league's best forwards.

Adjusted points, as calculated by Hockey Reference among other places, is not a perfect stat.  It is a good first step to normalizing for era, but has some limitations, including not taking into account the relative usage of top line players which can mean that some eras have disproportionately high or low scores.  For example, there's a solid case to be made that today's players are somewhat underrated by adjusted points relative to scorers from the late '90s because of reduced power play levels and the increasing tendency for coaches to keep shifts short and use all four lines.  That said, any comparison across eras does need to take into account the difference in scoring context, and adjusted points are certainly better than raw stats in that regard.

During the 2014-15 season, the league leaders in adjusted points ended up in the mid-90s, led by Jamie Benn's 97 and John Tavares' 96.  I chose to look at Hart voting and adjusted scoring totals for every season since 1996-97, a period that corresponds to an increasingly defensive style of play in the NHL.  Here's the entire list of players from 1996-97 to 2013-14 who recorded a final adjusted points total below 105 while also at the same time finishing ahead of every goalie in the league in Hart Trophy voting:

Considering that only three Harts were awarded to goalies during this period while dozens of players recorded adjusted point totals in the 85-104 range, that is a surprisingly short list.

The presence of four players from the past two seasons seems to reinforce what I said earlier about adjusted stats shortchanging current stars.  I'll get to them in a minute, but first let's dispense with a couple of players who are not particularly relevant to the current Hart Trophy conversation.  Mario Lemieux is only on this list because of games missed due to injury (43 games played).  Nobody in 2014-15 is even remotely close to his off-the-charts per-game scoring rate.  Pavel Datsyuk is also a poor comp for any of this year's top forwards, given that he was an outstanding two-way player that not only managed that level of offensive production, but did it while winning the 2008-09 Selke,  Ovechkin received plaudits this year for his improved play away from the puck, but he's still a long way from throwing his hat into the Selke discussion.

Bure and Iginla seem to have both mainly picked up Hart votes because they were elite goalscorers on fairly mediocre teams, which certainly brings to mind Ovechkin.  However, it should be noted that 1999-00 was a very weak season for top-end goalies.  Hasek missed much of the season due to injury, and Olaf Kolzig was awarded the Vezina almost by default as no starter in the league managed to even hit .920 (the only time that has happened since 1996-97).  In contrast, 2003-04 was a much stronger year for goaltending with goalies finishing 3rd, 4th, and 6th in Hart balloting.  However, all three goalies (Martin Brodeur, Miikka Kiprusoff and Roberto Luongo) had very different cases as the most valuable at their position, and it looks like Iginla benefitted from vote-splitting behind the nearly unanimous top-ranked skater Martin St. Louis.  The three goalies combined for 4 first place votes, 43 second place votes and 26 third place votes compared to Iginla's 2, 20, and 15 respectively.  If there was a clear goalie front-runner in the Hart debate, odds are they would have surpassed Iginla.

As for the past two seasons, 2012-13 was a lockout shortened season where few of the league's star goalies had top-drawer campaigns.  The top goalie (Sergei Bobrovsky) also played on a team that missed the playoffs, something that surely cost him points among Hart voters.  Last year Semyon Varlamov did have a strong season for a team that relied heavily on goaltending, but he didn't even win the Vezina and statistically can't be placed quite on the same level as Carey Price in 2014-15 (.927 vs .933).

The results from the last two seasons seem to suggest that voters are adjusting at least somewhat for the lower-scoring climate among the league's top point producers today.  On the other hand, even very strong scoring seasons aren't always guaranteed to beat out goalies.  Since 1997-98, 50 players have scored 105 or more adjusted points.  Of that group, 23 still ended up finishing behind at least one netminder in Hart balloting.  Not only is the top goalie very likely to beat out any scoring seasons in the 90-104 adjusted point range, but it's quite possible to surpass even better scoring years if the goalie has a strong enough case themselves.

Here are some of the scoring totals that were required to beat out some of the best recent goalie seasons (GSAA is Goals Saved Above Average, save percentages are adjusted to the 2014-15 league average of .915):

Miikka Kiprusoff, 2005-06:
Seasonal stats:  .934 adjusted save %, 41.6 GSAA
Finished behind:  121 and 120 adjusted points

Roberto Luongo, 2006-07:
Seasonal stats:  .929 adjusted save %, 34.5 GSAA
Finished behind:  122 adjusted points

Ryan Miller, 2009-10:
Seasonal stats:  .932 adjusted save %, 36.2 GSAA
Finished behind:  119, 117 and 117 adjusted points

Tim Thomas, 2010-11:
Seasonal stats:  .939 adjusted save %, 45.8 GSAA
Finished behind:  106, 113, and 107 adjusted points

In that context, it becomes pretty clear that Carey Price's .933 adjusted save % and 36.7 GSAA is historically on a higher level than Ovechkin's 91 adjusted points or John Tavares' 96.  Elite scoring may trump elite goaltending more often than not in Hart voting, but elite goaltending has routinely beaten everything else.  The only real exception in the above group is Corey Perry beating Tim Thomas in 2010-11.  I'm pretty sure a few of the above netminders would have won the Hart if they were fortunate enough to peak in a year where the scoring leaders struggled to crack 85 points.  It may be harder than ever for today's top scorers to rack up points, but is it 25% harder than it was just a few seasons ago?  That doesn't seem likely, and that's the kind of adjustment needed to close the gap between today's top scorers and Price, based on past voting.

The only potential saving grace for those who want to claim Ovechkin got robbed is that elite goal scorers do generally get an extra push in Hart voting compared to scorers with similar overall point totals (as we've already seen in the cases of Bure and Iginla).  Of the 26 snipers that managed 55 or more adjusted goals between 1996-97 and 2013-14, exactly half finished ahead of all goalies in Hart voting.  On its own, that would suggest that Ovechkin and his 60 adjusted goals might be deserving after all.  However, that is again comparing against any given goalie season, not necessarily a season on the level of Price's.  The other problem is that most of those goalscorers also managed to collect a healthy number of assists, something that Ovechkin was unable to do in 2014-15.  As a result, the lowest adjusted points total of any of the 13 was Pavel Bure's 103 in 1999-00, which as already mentioned was one of the very worst seasons for elite goaltending.  Next lowest was Perry's 2010-11 at 106 adjusted points.

Eight of the 55+ adjusted goals group scored fewer adjusted points than Bure's 103, and seven of those eight actually finished behind multiple goaltenders in Hart voting.  That's why Alex Ovechkin and his 91 adjusted points losing to a goalie is not at all unprecedented in a historical perspective.  It would actually require the voters to give him an extraordinary boost because of the declining numbers among top forwards, something on the level of 15-20 adjusted points (perhaps even more, considering how good Price was) compared to prior benchmarks to get him to the point that he could be considered likely to win.

This entire discussion is based on historical precedent, not necessarily who was more deserving.  Although I'm personally fine with Price being showered with hardware, I think reasonable cases have been made for Ovechkin based on relative performance.  If you want to suggest that Ovechkin's goalscoring is relatively more valuable than even Price's terrific puckstopping, by all means go ahead and present the evidence, I'm not sure there are many clearly right or wrong answers when it comes to a difficult problem like positional comparisons.  I just don't think it's correct to argue that Ovi had the weight of history on his side.  In actual fact, the clear and oft-repeated historical Hart Trophy benchmarks show pretty clearly that Ovechkin simply didn't score nearly enough points to be able to challenge a .933 in 66 games played, which is why it should not be considered at all surprising that the media and players cast their ballots the way they did.