June 10, 2009
Mailbag
June

Puck Prospectus

Alex O. (York, PA): What exactly is GVT?

Tom Awad: Thanks for the question Alex. GVT is "Goals Versus Threshold" and means the value, in terms of goals, a player contributes above what the marginal player would contribute. For example, a + 2.0 GVT means player A has contributed 2 goals above what the marginal player would contribute. GVT has four components: offense, defense, goaltending and shootouts. By definition, player GVTs on a team sum to that team's goal differential, so GVT allows us to compare players at different positions and quantify how much they have contributed to their team's success. This summer I plan on running a GVT tutorial so that all of our new readers can catch up on Puck Prospectus terminology.

Dan S. (Yonkers, NY): I am anxious to see more about GVT as I think it will prove what I have postulated for many, many years: That Martin Brodeur is nothing more than a slightly above average goaltender that was surrounded by an exceptional defensive team, badly inflating his statistics.

Timo Seppa: A few months ago, Tom Awad wrote a short series ranking all-time goalie performance for longevity, peak and playoff performance, using GVT. By this methodology, Brodeur came in tied for 5th with Ken Dryden, Ed Belfour and Glenn Hall, with considerable work to do to catch Tony Esposito and especially Jacques Plante, Dominik Hasek and Patrick Roy before he calls it a career. I’m betting that Marty ranks 4th at the end of the day, which sounds pretty fair to me.

As for this past season, ranking goalies with at least 20 GP by GVT per game played (a measure of how many goals per game better or worse each was than a marginal goaltender), Brodeur came in 17th out of 50 eligible netminders, or roughly in the top third:

Rank	Name			Team	GP	Val OG	GVT/GP
1	Tim Thomas		BOS	54.3	+36.3	+0.67
2	Tomas Vokoun		FLA	55.4	+29.4	+0.53
3	Craig Anderson		FLA	27.3	+14.3	+0.52
4	Niklas Backstrom	MIN	68.1	+27.6	+0.41
5	Roberto Luongo		VAN	53.0	+19.6	+0.37
6	Jonas Hiller		ANA	41.4	+14.8	+0.36
7	Nikolai Khabibulin	CHI	41.1	+13.9	+0.34
8	Ryan Miller		BUF	57.4	+18.7	+0.33
9	Mike Smith		TBL	41.2	+12.5	+0.30
10	Martin Biron		PHI	52.9	+15.9	+0.30
11	Pekka Rinne		NSH	50.0	+14.5	+0.29
12	Chris Mason		STL	53.6	+15.4	+0.29
13	Scott Clemmensen	NJD	39.3	+11.2	+0.28
14	Jaroslav Halak		MTL	32.2	 +8.9	+0.28
15	Henrik Lundqvist	NYR	69.2	+18.3	+0.26
16	Dwayne Roloson	        EDM	59.9	+15.6	+0.26
17	Martin Brodeur		NJD	30.2	 +7.6	+0.25
18	Cam Ward		CAR	65.5	+15.8	+0.24
19	Steve Mason		CLB	61.1	+14.7	+0.24
20	Brian Boucher		SJS	21.5	 +5.1	+0.24
21	Jonathan Quick		LAK	41.6	 +9.6	+0.23
22	Antero Niittymaki	PHI	30.1	 +6.8	+0.23
23	Marc-Andre Fleury	PIT	60.7	+11.6	+0.19
24	Kari Lehtonen		ATL	43.7	 +6.9	+0.16
25	Alex Auld		OTT	40.8	 +6.4	+0.16
26	Yann Danis		NYI	29.3	 +3.9	+0.13
27	Manny Fernandez	        BOS	27.4	 +2.9	+0.11
28	Evgeni Nabokov	        SJS	61.4	 +6.0	+0.10
29	Ty Conklin		DET	37.4	 +3.6	+0.10
30	Cristobal Huet		CHI	39.2	 +3.7	+0.09
31	Carey Price		MTL	50.6	 +2.0	+0.04
32	Ilya Bryzgalov		PHX	62.7	 -0.1	+0.00
33	Miikka Kiprusoff        CGY	73.6	 -2.1	-0.03
34	Brian Elliott		OTT	27.8	 -1.5	-0.05
35	Jose Theodore		WAS	54.8	 -3.6	-0.07
36	Martin Gerber		TOR	25.7	 -1.8	-0.07
37	Jason LaBarbera	        VAN	24.1	 -1.8	-0.07
38	Erik Ersberg		LAK	24.6	 -1.9	-0.08
39	Jean-Sebastien Giguere	ANA	41.0	 -3.4	-0.08
40	Dan Ellis		NSH	32.7	 -3.2	-0.10
41	Joey MacDonald	        NYI	46.5	 -5.2	-0.11
42	Patrick Lalime		BUF	21.6	 -2.8	-0.13
43	Peter Budaj		COL	53.8	 -7.2	-0.13
44	Marty Turco		DAL	72.1	-10.8	-0.15
45	Karri Ramo		TBL	21.9	 -5.9	-0.27
46	Andrew Raycroft	        COL	28.7	 -8.7	-0.30
47	Vesa Toskala		TOR	50.9	-18.1	-0.36
48	Chris Osgood		DET	44.4	-17.4	-0.39
49	Manny Legace		STL	24.2	-10.1	-0.42
50	Johan Hedberg		ATL     28.6	-13.1	-0.46

Don C. (Melville, NY): Who should the NY Islanders take with the first overall pick? Who will be better: John Tavares or Victor Hedman?

Andrew Rothstein: Don, that is a very good question. The best answer I can give you is that they're both likely to be superstars in the NHL. Iain Fyffe wrote a great amount on both of the draft-eligible prospects, and came to the conclusion that each will likely have a great impact at the professional level. Tavares's numbers have dropped off recently in the OHL, from 72 goals and 134 points, second in the OHL, in his age 15 season. Still, the 17 year olds numbers are very impressive in comparison to the rest of the juniors that are draft-eligible this year. Hedman scores fewer points than Tavares, but he's at a premium position and has played in the Swedish Elite League. The SEL is one of the most difficult leagues to play in, making Hedman's adjusted numbers look a lot more favorable than his statistics would previously suggest.

Jason A. (Palo Alto, CA): If I remember correctly, three points (Goal, Assist, 2nd Assist) are awarded on a very large proportion of NHL goals scored. If this is true and one of the three forwards fails to receive a G or A--then by default that 'missing point' is most likely going to the defense. Is it possible that individual players with a low OC% (Offensive Contribution Percentage) might just be benefiting from a good offensive defenseman? On the other hand, if it was a aggressive offensive defensemen play creating low OC% numbers for forwards--I would expect the effect to be largely random from one season to the nest--but I think you indicated that OC% were correlated from one season to the next--which lends strength to the view that it is a player attribute and not random in origin.

Robert Vollman: That's an excellent question, and worthy of future study. For those of you new to Puck Prospectus, you can find my column on Offensive Contribution Percentage here. If it were true that high-scoring defensemen would result in forwards with lower OC%, then we'd expect to see some Red Wings among those with the lowest OC%, after all they have Lidstrom, Rafalski and Kronwall. However there were no Red Wings in the bottom 10. Instead we had two Thrashers, and a collection of players on teams that, for the most part, aren't known for having an abundance of scoring from the point. Still, your reasoning makes sense, and the sort of thing we should study more closely.

Matt G. (Atlanta, GA): Bryan Little and Todd White both played about half of the season on Ilya Kovalchuk's line. Kovalchuk has a tendency to carry the puck the length of the ice and score. Perhaps, both Thrasher forwards have low contribution numbers because of Kovalchuk's tendencies? Is it possible that the low OC% (Offensive Contribution Percentage) could tell us more about the player's linemate than it is about their own role/contribution to the team? (I'm wondering how Ovechkin's linemates came out in this study? To move away from an Atlanta example, I notice both Sedin's rank very high in OC%--does this mean that 1) they pass to each other a lot or 2) the third member of their line isn't helping much or 3) The Sedins tend pass to prefer pass to the D who are getting that third point.

Robert Vollman: This is another great question regarding OC%. Virtually every statistic tells us at least something about a player's linemates, and OC% moreso than most. Of course, it also tells us something about Little and White if they're put on the ice with a player like Kovalchuk. And the low OC% makes us wonder, would Little and White score as much if they weren't with Kovalchuk? Would Kovalchuk score more if he lined up with players who (directly) contributed more offense than Little and White? As for Ovechkin's linemates, though they changed throughout the season, one of them, Nicklas Backstrom, made the bottom 10 in lowest OC%. One of the others, Victor Kozlov, would have been 6th in the bottom 10 had he scored enough even-strength points to qualify for the list, so you may be on to something. As for the Sedins, whenever I watch them play I'm struck by the fact that they only seem to work with each other, as opposed to the other players on the ice. Of course, that's a purely subjective observation on my part, but that's my take on it. Thanks a lot for your questions, I'm glad this statistic gave us so much to discuss.

Michael I. (Pittsburgh, PA): Sykora's not playing because he's hurt. It's probably his shoulder and happened at the end of the regular season, which was why he was totally ineffective in the first round of the playoffs before being replaced in the lineup. Your article about his regular season productivity vs. Ray Shero's ego/waiver wire pickups is missing the boat completely.

Timo Seppa: I don't know if that jives with the information out there. Petr Sykora was quoted as saying "I'm surprised" when he was first held out of the lineup against Washington.

I'll give you that he was ineffective against Philadelphia (and again, in Game 6 against Detroit last night), but my question was what Bylsma has to lose playing a forward with a positive GVT over the likes of AHL level fodder like Craig Adams. In my mind, both Sykora and Satan should be out on the ice instead.

Isaac L. (Ann Arbor, MI): Regarding ignoring major penalties, do you mean just coincidental majors? Drawing a major for high-sticking, for example, has an even higher cost than 2.5 minor penalties, and dropping it would seem in appropriate.

On a side note, though I understand why you are not counting misconducts or coincidental majors, as in today's NHL, they are primarily taken by specific role players for which the same standard of discipline may not apply, I would not say they cause no harm, as this assumes the player has little value taking a regular shift. Even a fringe player can be missed: consider the case of a defenseman missing for half a period, throwing the line changes out of alignment and requiring an average of 20% more ice time for each defender.

Robert Vollman: In this study I disqualified all major penalties, although your approach would certainly be more correct. And as someone who regularly sees Jarome Iginla get pulled off the ice for 5 minutes after tangling with a 4th-line nobody, I can certainly appreciate your 2nd point! I strive for statistics that strike the right balance between insight and complexity, and I'm concerned that it would be too difficult for the average fan to get data on which of their majors were coincidental compared to the added value it would bring them. I appreciate your writing in with these points, keep it coming.

Eric K. (Toronto, ON): Great article, great analysis - may I suggest using penalty minutes as your divider instead of minor penalties, because not all majors are fighting majors. It should be just as easy to simply remove fighting minutes and misconducts from a player's total penalty minutes, no? Also, have considered lowering your weighting on last minute penalties taken in non-close games?

Robert Vollman: Yes, you certainly may suggest. When attempting to measure something new it's important to get other perspectives. For example, between you and Isaac the general consensus seems to be that we definitely want to include the non-coincidental majors in our measurement. I love your idea of further refining the statistic by examining when a penalty takes place, and penalizing a player less when they're not taken during key moments of a game, but again there's that balance between complexity and insight that we want to hit. I'd love to hear more opinions on this, and the other statistics we've introduced, so please keep the ideas coming and we'll reach our targets quicker. Thanks for writing!

This article was authored by the staff of Puck Prospectus. You can contact the Puck Prospectus team by clicking here or click here to see the Puck Prospectus team's other articles.