Hockey Prospectus 2010-11 Review

Ah, October. The days are getting shorter, the winds are crisp and hockey is right around the corner. Who among us isn’t happy to know that?

However, with the coming off the new NHL season comes the inevitable previews and prognostications from all of our favorite (and not-so-favorite) websites and print publications.

I understand the reasoning behind publishing these collections of guesswork. They definitely do draw readership and attention. But let’s face it; nearly every preview (including mine) will look foolish come the end of the 2010-11 regular season.

The NHL, like all pro sports leagues, is too difficult to predict with any high level of accuracy. Who can correctly anticipate injuries? What about the premature regression of star players (Vinny Lecavalier, anyone?)? For that matter, we know there will be young players who develop faster than expected (ala Steven Stamkos last season); how do we account for that?

Heading into last season, who figured Phoenix or Colorado would be playoff contenders? How many thought the Eastern Conference Finals would pit the seventh and eighth seeded teams against each other? Wasn’t Washington and/or Pittsburgh supposed to represent the East in the Stanley Cup Finals?

Predictions can be impacted by random events and that makes making predictions very unscientific. However, there are some who are beginning to make headway into injecting some science and objective analysis into the practice of prediction.

At the forefront of this movement is Hockey Prospectus; formerly know as Puck Prospectus. Launched in February of 2009 as the new child of the Prospectus Entertainment Ventures group, Hockey Prospectus is the pre-eminent source for new analysis into the sport.

They’ve introduced new metrics used to evaluate players like Goals Versus Threshold (GVT) and the Vukota projection system. GVT isn’t the first attempt to develop an all-encompassing player evaluation statistic but it probably is the best going right now. For a refresher on GVT, here you go.

Vukota is hockey’s equivalent of Baseball Prospectus’ Pecota projection system. The team at Hockey Prospectus has spent countless hours researching hockey history, identifying statistical trends in the performance of players. They incorporate concepts like regression and progression due to the aging of players while also trying to anticipate something which seems to occur randomly, injuries. Vukota projects the number of games played in addition to goals, assists and points.

So is it any surprise that Hockey Prospectus has finally put all of this data into their first annual year book? Drum roll please; let’s welcome “Hockey Prospectus 2010-11: The Essential Guide To The 2010-11 Hockey Season,” to the hockey world.

Not only does Hockey Prospectus include traditional data like player stats and team records but they also apply their new metrics to project individual and team results for the new season.

Of course it isn’t all about the new season either. Hockey Prospectus analyzes a number of other issues including: whether the NHL should go to a balanced schedule; can we predict how well a player making the jump from another league will transition to the NHL; and introductions to GVT as well as to Corsi.

But isn’t it still mostly about the numbers for Hockey Prospectus; the numbers and the way they are analyzed?

One thing that did surprise me in the Vukota projection results was the uncanny similarities between this season’s projected numbers and the player’s production in recent seasons. From experience I would expect more fluctuation in production.

For an example of what I mean, let’s look at the Phoenix Coyotes Vukota projections (point per game) and their production in recent seasons (also point per game). I also include the number of games played, both in 2009-10 and projected for 2010-11, and the player’s age this season to give the numbers some context. Let’s see if my interpretation of the Vukota rankings proves true when held up to some further analysis.

Player Age GP PPG (’09-’10) Proj. GP Vukota PPG (’10-’11) Variance % Variance
Mikkel Boedker* 21 92 0.34 44.0 0.40 0.07 20.1%
Shane Doan 34 82 0.67 74.6 0.74 0.07 10.3%
Vernon Fiddler 30 76 0.39 67.9 0.38 -0.02 -4.5%
Martin Hanzal 23 81 0.41 73.6 0.50 0.09 22.1%
Lauri Korpikoski 24 71 0.15 62.5 0.21 0.06 37.4%
Petr Prucha 28 79 0.28 63.5 0.37 0.09 32.9%
Taylor Pyatt 29 74 0.31 64.1 0.35 0.04 13.4%
Lee Stempniak 27 80 0.60 66.9 0.59 -0.01 -1.1%
Scottie Upshall 27 49 0.65 56.2 0.61 -0.04 -6.0%
Radim Vrbata 29 82 0.52 70.2 0.60 0.08 14.4%
Wojtek Wolski 24 80 0.81 72.8 0.78 -0.03 -4.1%
Ray Whitney 38 80 0.73 69.4 0.74 0.02 2.2%
Eric Belanger 33 77 0.53 66.5 0.48 -0.06 -10.5%
Adrian Aucoin 37 82 0.34 54.7 0.39 0.05 14.0%
Ed Jovanovski 34 66 0.52 60.9 0.50 -0.01 -2.1%
Sami Lepisto 25 66 0.17 62.5 0.20 0.04 22.9%
Derek Morris 32 76 0.38 63.6 0.33 -0.06 -14.7%
Keith Yandle 24 82 0.50 73.5 0.53 0.03 6.7%

*With regards to Boedker, I incorporated both his 2008-09 and 2009-10 seasons together since he only appeared in 4 contests last year.

When you take it all in together, these Coyotes will increase their scoring collectively by a modest 0.03 PPG, or 6.7%. That doesn’t seem like a lot given the Coyotes have several young players who can reasonably be expected to improve their production this year.

I would have expected some of the younger vets like Wolski, Yandle and Hanzal to perhaps have had steeper increases in their Vukota projections since they are just entering their prime years. Wolski specifically seemed to blossom upon his arrival in the desert yet Vukota actually sees a modest drop-off in his production.

Something else I want to learn more about is what factors into Vukota’s projections of games played. Let’s take a guy like Adrian Aucoin. Last season he appeared in all 82 games for Phoenix. The two years prior to that, Aucoin played in a total of 157 games; an average of 78.5. Yet Vukota only projects 54.7 games for the veteran blueliner. What does Vukota see that leads it to conclude Aucoin is in for an injury-filled campaign?

Apparently Vukota doesn’t put much credence in Lee Stempniak’s 14 goal, 18 point performance in just 18 games in the desert. They have him pegged much closer to his career numbers which makes sense to me.

Vukota seems to think Ray Whitney, even at age 38, can maintain his pace for another season. It projects the veteran winger to hold his .73 PPG production in 2009 – 2010 with a similar .74 PPG in 2010-11.

This is the type of information Hockey Prospectus gives the reader for each team in the NHL. How close the Vukota projections approximate the actual results from the season remains to be seen but it will certainly be interesting to find out.

Regardless, this is the most comprehensive NHL year book or preview than any I’ve ever seen. Anyone can publish last year’s stats and records along with their guesses on how this season shapes up but Hockey Prospectus actually uses a different approach that utilizes sophisticated and well-thought out analysis to accompany the stats.

As mentioned before, “Hockey Prospectus: The Essential Guide To The 2010-11 Hockey Season,” includes several thought provoking articles within its pages as well. Tom Awad offers his introduction to GVT and Vukota. That follows a foreword from respected hockey blogger Christopher Botta. Botta is a former New York Islanders front office employee whose writing now appears on several hockey websites.

One of the founding fathers of statistical analysis in hockey, Iain Fyffe, documents the history of statistical analysis in hockey. In this piece, Fyffe mentions one of the earliest examples of hockey analysis and one of the works which inspired him: The Klein and Reif Hockey Compendium. The Compendium was written by two American sportswriters; Jeff Klein and Karl-Eric Reif. In their book, Klein and Reif take some of the first shots at traditional stats like plus/minus and a goalie’s save percentage. They even made an attempt to find an alternative to save percentage: Goaltender Perseverance.

Awad returns to discuss the advances made in the understanding of shot quality in hockey. The NHL’s introduction of Real Time Scoring System (RTSS) has given us a better ability to understand what shots are more and less dangerous than others.

It’s pretty obvious a shot from in tight has a greater likelihood of going in than one taken from the blue line but there are still some surprises in Awad’s analysis. One consideration he mentions is the significant differences in how scorers judge distances from arena to arena. Alan Ryder did some research into this and discovered the scorers in Madison Square Garden tend to underestimate the actual distance a shot comes from. This tends to inflate Ranger goalie Henrik Lundqvist’s performance while making the Rangers forwards look bad.

If, as Ryder argues, the scorer at MSG marks the distance of a shot against The King as 15 feet when it was actually 20 feet that would artificially inflate the degree of difficulty in stopping the shot and thus makes Lundqvist look better. The opposite is true if the same scenario occurs in regards to a Rangers sniper. If Marian Gaborik doesn’t convert a shot from 15 feet (even though it may have been from further back) that reflects badly on Gaborik.

Awad also looks at what percentage of rebound shots are converted into goals and how the score at the time in the game impacts shooting percentage. Also, to no one’s shock or surprise, shots taken immediately following a turnover result in a higher percentage of goals than the norm.

Next up is Robert Vollmer’s dissertation on the improvements made to the Corsi Rating. I’ve introduced you to this metric before but since then analysts are beginning to use it in different ways. What started out as a simple way to gauge which team is controlling the play and which players play a big role in determining that is near to becoming a more useful stat because of the different applications it appears.

Richard Pollock next makes a compelling argument for the NHL adopting a balanced schedule. Simply put, instead of putting a higher emphasis on playing more intra-divisional games, the NHL should come up with a formula where every team plays a more balanced schedule.

Pollock does a division-by-division breakdown analyzing goal differential. His rational is divisions that finish with better goal differentials (i.e. more combined goals scored than combined goals allowed) then those divisions are stronger as a whole. If the division is stronger as a whole the teams within that division that have to face each of their rivals 6 times in a season may be at a disadvantage relative to teams in weaker divisions.

Pollock charts last season after adjusting for divisional strength. The Pacific and Atlantic divisions were the most severely affected by Pollock’s adjustments. Pollock’s calculations award each team in the Pacific a minimum of 14.27 additional points in the standings due to his adjustments.

In fact, his standings would see every team in the Pacific in the postseason with the exception of the Dallas Stars; who by virtue of the NHL’s insistence that each division be represented in the playoffs were out despite having 3.77 more points than Vancouver in these adjusted standings. The Atlantic Division would have sent each of their teams to the playoffs if the standings were actually calculated the way Pollock theorizes. The Rangers and the Islanders moved up while the Bruins and the Canadiens would have moved out of the playoff mix.

Pollock emphasizes his point by asking this question; “If teams in three different divisions in the Eastern Conference are competing for the same fourth or fifth seed in the conference, should the teams not have an almost identical schedule?” This rationale makes sense to me but I’m sure the NHL doesn’t care much what I think about this. Thus I would expect the NHL schedule to keep the same format for the foreseeable future.

Next up, Vollman introduces us to the concept of league equivalencies. How many of us wonder how well our new foreign import (Mats Zuccarello-Aasen of the Rangers for example) or this year’s supremely talented No. 1 draft pick (Taylor Hall) will transition to the NHL? Vollman explains how years of historical data can be used to calculate a newcomer’s NHL production based on his production in other leagues.

A basic approach to this was introduced in 2004 by Gabriel Desjardins. Desjardins simply analyzed years of historical data to determine basic translation factors for each league. Not surprisingly, players coming over from the KHL produce at a rate of 83% their KHL production in the NHL. Thus if a player was a 50 point scorer in the KHL, he can be expected to score 41.5 points in the NHL.

A more advanced approach, one factoring ice time, quality of said ice time and age, can also be used. Using the more advanced method, fans of the Rangers (Zuccarello), Red Wings (Jiri Hudler), and Flyers (Nikolay Zherdev) have reason to look forward to their new arrivals.

Will Carroll chimes in with a piece questioning whether the vagueness used when reporting injuries is really necessary. There is a prevailing fear that if a club is too specific reporting the type of injury to a a player or the body part affected, opposing teams will focus on that specific injury or area upon the injured player’s return.

He looks into a case last year where Alex Ovechkin was hurt and it was reported simply as an “upper body injury.” After a couple of weeks, the diagnosis was more specific; it was a shoulder strain. Upon his return NHL executives analyzed game film and tracked how many hits were attempted on Ovechkin to see if there was an increased focus on him. Carroll reasoned if opposing players were really focused on “taking Ovechkin out,” then there would be more hits attempted on him. Carroll’s study showed no significant change in the volume of physicality directed at the Capitals star.

To further back this conclusion up, Carroll spoke with an unidentified NHL defenseman often asked to play a physical style. When asked if he would specifically target a player returning from injury more than an uninjured player, the NHL defenseman replied, “No, I guess not.”

It would seem Carroll has isolated a myth rather than a fact when confronting the idea that players target other players with a reported injury. This revelation would seem to render teams’ unwillingness to report the type and severity of their player’s injuries as unnecessary.

Finally, Corey Pronman closes it out with his top 50 NHL prospects. Columbus winger Nikita Filatov comes in No. 1 on Pronman’s list. Edmonton’s prize prospect (Hall) and Bruins center Tyler Seguin, the first and second choices in this past entry draft, placed No. 2 and No. 4 respectively on the list.

In addition to Hall, the Oilers placed 2 other names in Pronamn’s top 11; Magnus Paajarvi-Svensson (5th) and Jordan Eberle (11th). It would seem the future looks bright in the great white north.

The most controversial pick of Pronman’s is probably his choice of Islanders’ LW Kirill Kabanov as the 8th best prospect in the NHL. Kabanov was bypassed by every team in the NHL at least once this past June at the entry draft due to concerns over his maturity and character. The Islanders finally selected the talented forward in the third round (65th overall).

To say Kabanov is the prototypical boom or bust prospect is an understatement. For Pronman to list him so high on his list is a testament to his belief in Kabanov. We’ll see who is right ultimately; Pronman or the 29 other teams who chose not to select Kabanov before the Islanders.

At the end of the day, Hockey Prospectus has produced a 349 page NHL season preview of epic proportions. Whether you are a believer in statistical analysis or not Hockey Prospectus offers you enough numbers and analysis to make your head spin.

This is the perfect NHL annual for any hockey junkie. A person can practically waste days thumbing through the pages of this annual and still not see everything it has to offer. Go to:, to find out how to get your own copy of the best hockey annual in publication today.


Tags: , , , , , , , , , , , , , , , , , , , , , , , ,

Comments are closed.