SB Nation User Blog: Big Picture Guy

Goalies are Better Looking?

Big Picture Guy — Sun, 22 Nov 2009 21:27:17 -0000

A new study published in New Scientist contains this sentence "In previous work, they found that hockey goalies and forwards, and soccer strikers and goalkeepers, tended to be more attractive than their teammate". Here's a link to the original study (pdf) called " Does the face reveal athletic flair? ".Positions in team sports and facial attractiveness. Personality and Individual Differences. I suspect that this study was completed long

after Gump retired.

via ourhistory.canadiens.com

The Surplus Value of Draft Picks

Big Picture Guy — Mon, 22 Jun 2009 17:31:24 -0000

Draft picks are coin of the realm among NHL general managers, hoarded, traded, exchanged. Given that they represent pure probability, trading them is like exchanging win/place/show tickets on a horse race that is far off in the future. Their value seems to change from trade to trade and is often insanely difficult to calculate as part of individual trades.

One approach developed by Victor Wang (in Hardball Times) to quantify the economic value of baseball prospects seems promising. Wang's methods are described below but basically he compares the salary savings of a drafted player during the years in which that player is controlled by an organization (six years in MLB, until age 27 in the NHL) with the cost of a obtaining a comparable player in free agency. The product is a weighted result that Wang calls surplus value* which he then uses to evaluate trades involving prospects and which others have used to rank teams' farm systems (See this example from SBN's Beyond the Boxscore).

It seems to me that this "surplus value" approach may have even more relevance to the NHL than to MLB. MLB teams can't trade draft picks: MLB teams don't operate under a "Hard" salary cap like NHL teams. Both rich and poor teams in the NHL need to maximize the "surplus value" of draft picks to create an efficient team salary structure.

I've developed two approaches based on Wang's work to quantify the value of NHL draft picks and prospects. The first approach, which I call Optimal Payroll Surplus (OPS-a nod to the SABR origins of Wang's work), is for prospects. The second, Probable Payroll Surplus (PPS), is intended to apply to draft picks. It discounts OPS by the percentage of draft picks that never play in the NHL.

Numbers and comparisons are in the table below. For those unmoved by stats, here are three narrative examples of how these theories might be applied:

The Chicago Blackhawks had 12 players 23 years or younger on their 2008-9 roster, resulting in a savings versus expected payroll of about 17 million. Fourteen million of that came from five drafted forwards and two prospects acquired in trade (Versteeg and Fraser). Based on draft position, these seven forwards have a projected PPS of 37 million in their cost-controlled years. The Blackhawks are well on their way to fully realizing those savings. In the days before the salary cap, that would have meant more money in the late William Wirtz's piggybank: instead the payroll savings allowed Dale Tallon to overpay for Brian Campbell and snafu his way into a ridiculously expensive goalie tandem. But Tallon's profligacy is not the point here, rather, the payroll savings gave Tallon the flexibility to make those moves and now give him the flexibility to recover from his mistakes. Which is not the case in my next example
The noose of unrealized PPS is tightening around Glen Sather and the New York Rangers. In 2004, Sather, realizing he needed forwards, used 11 of his 13 picks on them. Add in the ill-fated first round pick of Hugh Jessiman in 2003 and you get a PPS of $38 million. So far the Rangers have realized only about 8 million in savings (Dubinsky, Callahan). The core of the problem is the three risky first round picks (Al Montoya, Lauri Korpikoski and Jessiman). Had just one of those picks been a forward who met expectations (Zajac, Parise, Getzlaf, Richards etc ) Sather might well have avoided tying up 15 million in Drury and Gomez and would have the flexibility to make trades and sign free agents this summer possibly vaulting the Rangers into a genuine Cup contender.
One of the many trades at the 2008 NHL Draft was this one: the Ducks traded pick # 28 to the Coyotes in exchange for 2 picks, #s 25 and 39. The Coyotes obviously had their eye on Victor Tikhonov and he has made a good start towards providing that cash-poor franchise with significant OPS, playing 61 games at age 20. Whether he ever reaches his full OPS potential will depend on whether he raises his game to the level of a second or at least third line forward. On the other hand, if Brian Burke was thinking that two picks were better than one ,he forgot to do his math homework: The Coyotes one pick, with a 52% probability of success, was worth more than the two picks received, with a 50% chance of one pick succeeding (general rule of addition) Burke did acquire an 8% probability that both would be successful . And, because second round picks have a significantly lower probability of becoming stars or first line forwards, the PPS Burke traded away- 4.6 million- is actually greater than the 2.9 million PPS of the two forwards Burke drafted.(All of this is academic since Burke’s successor, Bob Murray, traded Eric O’Dell (who was pick # 39) to the Thrashers for Eric Christensen. And the trade on draft day actually had its genesis in February 2007 when the Kings traded Mattias Norstrom to Dallas:try flow charting this series of trades (5 teams are involved so far) and you’ll understand the what I mean when I say trades can be “insanely difficult” to value.

Finally, for those who see this approach as bean-counterish and reductive, I tend to agree. It's not for everyone. Islanders fans won’t be thinking of the more than $30 million in PPS Garth Snow brings to the draft on Friday (although Charles Wang probably should be). They’ll be cheering the actual hockey players who might someday help their team, they'll be debating the skills those prospects display on the ice, not their possible contributions to the bottom line.

I plan on using these theories as a performance measure to evaluate GMs, sort of like ROI in finance. And maybe quantitative values will help shed some light on the often bizarre trades of draft picks that take place every year.This post at Under Review "Moving Up in the Draft" is a sample (hint-don't play this game with Doug Wilson).

I don't plan on rating the economic value of each organization's prospects but I think that might be a good thing to do and I would be interested in the results. Too many prospect lists read like Garrison Keillor's description of Lake Wobegon-"all the children are above average"-and there's a need for some kind of objective standard.

* "Surplus value" should have a familiar ring to students of history. The term is central to Karl Marx's theories on the exploitation of labor. Is Victor Wang suggesting prospects are exploited? I don't know but that term's historic associations make me nervous so I've substituted my own.

** Notes-Wang began by using WARP (wins over replacement player) to categorize prospects and comparable free agents. He has since switched to an even more exotic SABR formula. I grouped free agents/prospects into 6 groups(Star-Ist Line through 4th-marginal) based on TOI, PPG, historical depth charts. Projecting players salaries over their cost-controlled years is probably the weakest point here, not enough data and recent trends which give young players long term contracts affect this fairly dramatically. The OPS for picks 31 through the end are so similar because the distribution of types of forwards (mostly 3rd and 4th line) picked in those rounds is so similar. The PPS is so low because it reflects the much lower probability of success. I used a constant entry age of 20 years: I fully realize that lower round picks enter the NHL later (less salary savings) but more research is needed to find the appropriate break points. All my calculations are based on salary not cap hit and are for forwards only.

Draft Picks (Forwards)	Optimal Payroll Surplus	Probable Payroll Surplus
Picks 1-10	$ 12.3 million	$ 10.8 million
Picks 11-30	$7.8 million	$ 4.8 million
Picks 31-60	$6.6 million	$2.7 million
Picks 61-90	$6.6 million	$2.5 million
Picks 91-180	$ 6.1 million	$ 934 k
Picks 180 +	$ 6.4 million	$ 667 k

GMS and the Draft: Trends,Tendencies, and a Scorecard

Big Picture Guy — Tue, 09 Jun 2009 15:47:34 -0000

Team selections in the NHL Draft are usually a product of a consensus decision within each organization, but when the general manager of each team steps to the podium in Montreal this year to announce his team's selections, those selections will be indelibly linked to his name and his record, his success or his failure.

This report is the first I know of to systematically track and evaluate draft success or failure by general managers rather than by teams. Sixteen current general managers have records that are substantial enough to be evaluated: each draft pick they have made, as GM for any team, since 1990 has been recorded ,analyzed for tendencies (position,nationality, league source), and their overall record has been ranked by comparing their records of success with the average expected success rate by draft position over fifteen years of draft history (1990-2004). Success is measured by the number of players selected who play 200+ NHL games.* Each general manager has then been assigned a +/- rating versus the average expected success rate by draft position. Full reports (22 general managers in all) are at Under Review. Charts like these are included:

via 2.bp.blogspot.com

SCORECARD

1.Lou Lamoriello +12 2.Darcy Regier +8 3. Bob Gainey +5 4. Don Maloney +4 5. Dean Lombardi +3 6. Ken Holland +2, Bryan Murray +2 Jim Rutherford +2 9. Larry Pleau +0, Doug Wilson +0 11. Brian Burke -1, Darryl Sutter -1 13. Don Waddell -2 14. George McPhee -3 15. Glen Sather -6 16. David Poile -10

Note-Both Poile and Sather had successful drafts prior to 1990. So this survey is somewhat unfair to them:in fact, the year this survey begins (1990) Sather had perhaps the worst draft in history at Edmonton:none of his 11 picks ever played a single NHL game. At least we now know the price the Hockey Gods extract for beginners’ luck (Sather took Coffey and Kurri with two of his first three picks ever in 1980): it’s 15 years of sub-par drafts.
Lamoriello’s success would seem to refute theories about the randomness of the draft: in 15 years he has drafted 22 more players with 200+ games than David Poile (one less draft). That’s essentially a complete NHL roster and a pretty good one. On the other hand, Lou’s more recent picks are suspect (HF ranks the Devils 28th in organizational strength) suggesting possible regression toward the mean.

The following ratings and tendencies may be of interest to those compiling Mock Drafts (as should the full reports):

Best and Worst %-Forwards-Best-Wilson 33% Worst-Sutter-8% Defensemen-Best-Lombardi-40% Worst-Sather-8% Goalies- Best-Gainey-30% Worst-Several, 0% Highest # of Picks By Position-F-Sutter 69% D-Waddell-35% G-Wilson-20% Best Success Above Pick #180-Pleau-21% Worst-Sutter-0%(note-Holland is at 4%)

Tendencies- Europhile-Holland-52% Europhobe-Sutter 13% Highest Pct by Country and League- Canada-Sutter-69% US-Wilson-38% OHL-Poile-38% WHL-Sutter-40% QMJHL-Burke,Regier-15% NCAA-Tallon-20% US High School-Wilson-18%

* Other ranking systems use 100 NHL games as a threshold. That would add 137 players (over 15 drafts) to the list and would not change the rankings very much (except to boost Lou L.'s lead even more). My judgement is that the vast majority of players with between 100-200 games played are below replacement value statistically. It's worth noting that at least 137 undrafted free agents have entered the NHL during the 15 years surveyed and many of those non-draftees have had careers exceeding 200 games. Using 200 games as a standard, in short, is an attempt to ascertain the quality as well as quantity of a GM's drafting.

Season Series Records Not Predictive in Playoffs

Big Picture Guy — Mon, 13 Apr 2009 20:26:00 -0000

With the NHL playoffs set to begin on Wednesday, sports writers, and bloggers are focusing on the regular season series between the teams that are matched up in the first round,. While analyzing personnel, coaching style, and statistical data from the season series may provide pundits with a basis for plausible predictions, head to head records in a regular season series have proven to have low predictive power, especially in determining a winner in the first round of the playoffs.

In the last 15 seasons, there have been 210 matchups in the first three rounds* of the playoffs. Teams that "won" the season series have prevailed in 80 of those matchups, teams that "lost" the season series have won 80 of the playoff matchups and the remaining 50 matchups have been between teams which split the season series.

Results have been evenly distributed throughout the three rounds: in 1st round matchups, the winner of the regular season series has won 45 times, the loser has won 46 times. Second round results are 22(regular season winner) vs. 23 (regular season loser): third round results 13 (winner) to 11(loser).

What if a team has dominated in the regular season series against its' playoff opponent? The results there should be a little more encouraging to fans of Chicago (swept all four games vs. Calgary) and Boston (won five of six vs. Montreal). Using a margin of at least three more victories in the regular season series as a definition of dominance, "dominant" teams have won 11 of 19 such matchups in the first round. Oddly enough both "dominant " and "dominated" teams (teams which lost the season series by a wide margin but won the playoff series in round one) have fared badly in round two, compiling a 4-15 record.

Interestingly, regular season dominance seems to take on more significance in the later rounds.Teams that dominated the regular season series have won 11 of 15 matchups in the second round and all five such matchups in the third round. Perhaps, if the first round of the playoffs is the cliched "whole new season" then teams return to their regular season performance patterns in later rounds.

* The Championship finals match teams which have played very few games vs. one another or, as has been the case in the last three finals, teams that did not face each other in the regular season.

Data Mining the European Attendance Figures

Big Picture Guy — Sat, 14 Mar 2009 15:45:16 -0000

In a press release last week, IIHF president René Fasel trumpeted a "substantial increase" in attendance at European club hockey games. What Fasel didn't say and a closer look at the attendance figures reveals is that attendance actually declined for 37 of the 64 teams for which year over year comparisons are possible.

Most of the declines were small, to be sure, and seem related to poor on-ice performance by the individual clubs. Still, the Finnish Sm-Liga must be concerned about declines at 12 of its 14 clubs, the Swedish Elite League by declines at 9 of the 11 clubs for which year over year comparisons are available. The Russian KHL which blustered on to the scene last summer was a lot like its predecessor, the Russian Super League, except that its average per game attendance was actually lower for reasons explained below.

Fasel points to growth in attendance and the opening of larger arenas over the nine years that the IIHF has surveyed European attendance. That is certainly true and it's true that overall attendance at the five leagues for which strict comparisons can be made increased by 2.8%. But that increase can be more than accounted for by changes in 3 cities: in Berlin where the 14,200 seat O2 Arena replaced a much smaller arena and in two promotions, EHC Biel in the Swiss NLA and Rogle in the Swedish Elite League.

A fairer statement about European hockey attendance would go something like this:Attendance at European club hockey has plateaued at slightly over 70% of arena capacity and is at risk of trending downward .

Here are the numbers for the six major European leagues

League                                     Avg Pg 08-09 % Capacity Avg Pg 07-08 Total 08-09
Swedish Elite (330 games)                6260        83.00%               6288      2065800
NLA (Swiss, 300 games)                    6073        74.00%               5872      1821900
DEL(Germany,416 games)                 5867        65.00%               5504      2440672
KHL(Russia, 696 games)                   5097        75.00%               5325      3571872
Sm-Liga (Finland,406 games)           4919        70.00%               5012      1997114
Extraliga (Czech, 364 games)            4902         62.00%              4784      1784328

Attendance figures come from IIHF releases and the web site Hockeyarenas.net which deserves much credit for maintaining a vast historical database on teams,arenas , and attendance worldwide. Here are complete league by league and team by team attendance figures.

The KHL's overall average attendance per game fell below that of the 07-08 RSL (5325) because the four added teams were below the league average. As with everything involving the KHL, there is some confusion as to attendance figures. The IHHF reports a lower average per game than Hockeyarenas.net (5142) but the IIHF's numbers for the top nine teams in the league are significantly greater than those shown by Hockeyarenas.net.

Comparisons with the NHL are difficult since there are so few NHL-size arenas in Europe. However, a macro comparison of attendance on the two continents may provide some perspective:

The NHL reports attendance of about 21 million for 1230 games involving 30 teams in metro areas with a population of about 117 million. The six major European leagues report combined attendance of about 13.7 million for 2512 games involving 92 teams in metro areas with an estimated population of about 96 million.

And for those who like really big numbers, try this: North American hockey attendance at 11 leagues (including the three major junior leagues) with 201 teams is approximately 44 million per year. European (including Russia and Great Britain) attendance for the 301 teams listed at Hockeyarenas.net is about 18 million .