Tuesday, November 27, 2007

Consistency

In his new book, radio host Bob McCown rates Martin Brodeur as the third best goalie of all-time. One of his arguments is Brodeur's consistency, and he cites as evidence the fact that Brodeur has only once had a GAA higher than 2.50. This, of course, is a ridiculous argument. The reason Brodeur has had low GAAs is because he has faced few shots throughout his career. In most years, he would have had to have been terrible to have a GAA even close to 2.50. Here are the save percentages Brodeur would have needed to post in each year in his career for his GAA to be 2.50:

.912, .900, .905, .902, .890, .898, .900, .898, .891, .893, .897, .914, .910, .902

For seven years in a row, Brodeur only needed to hit .900 or better to have a sub-2.50 GAA. So giving him credit for achieving that arbitrary mark is misguided.

I raise a skeptical eyebrow any time someone uses the word "consistency" (could be the Fire Joe Morgan influence). This is another example of consistency inappropriately being used to praise a player, as his performance actually shows a reasonably high degree of variance. From 1994-95 to 2003-04, the average number of shots he faced per game ranged from a low of 22.8 to a high of 26.4. That is a pretty narrow range, and was well below league average every single season. His save percentages in that span ranged from .902 to .927, i.e. from below average to excellent. Four times he was at .906 or lower, three times .917 or higher, and three times in between, showing a fairly broad range of success, and it was the stingy defence that always kept him at 2.4 goals against or lower.

Brodeur has consistently played a lot of games, but that is about the only thing he has done consistently (unless you count being a member of a great defensive team). He has had a couple of outstanding seasons, as well as several mediocre ones, which is fairly typical of goalies. His low shot totals tend to hide his off-years, and make commentators gush about his consistency, but they are wrong: The consistency truly lies in New Jersey shot prevention, not in New Jersey goaltending success.

Thursday, November 22, 2007

Count the Rings...Or Not

Here are 12 random facts to illustrate the fallacy of using Cup wins and playoff team success to evaluate individual goaltenders:

1. The last 11 Stanley Cup champions have all finished in the top 5 in the league in regular season points. None of them had the regular season save percentage leader on their team, and only one (Martin Brodeur in 2003) won the Vezina Trophy and the Stanley Cup in the same season.

2. From 1976 to 1988, only three teams won the Stanley Cup (Montreal, Edmonton, and N.Y. Islanders). Three goalies combined either a Vezina or First All-Star season with a Cup win in the same season (Fuhr, Smith, Dryden), but all three teams had regular season and playoff success with several different goalies in the net.

3. Patrick Roy is widely considered to be the greatest ever playoff goaltender who stole many games, led many upset wins, and took two mediocre Montreal teams to the Stanley Cup. In his 19-year career, however, Patrick Roy only won five playoff series in which his team did not have home ice advantage. In his four Stanley Cup wins combined, Roy's team only beat one opponent that had more wins in the regular season that year. Roy was also a member of 2 President's Trophy winners and 10 division champions.

4. Between 1993-94 and 2003-04, the New Jersey Devils outshot their opponents in 20 out of 22 playoff series. They won 14 of those 22 series. Throughout his playoff career, Martin Brodeur has faced an average of nearly 4 fewer shots per game than his goaltending opponent.

5. In Edmonton, Grant Fuhr was 226-117-54 with 5 Stanley Cups. Outside of Edmonton, he was 177-178-60 with 0 Cups. In the playoffs with Wayne Gretzky on his team, Grant Fuhr was 63-21. Without Wayne Gretzky, he was 26-29. In international competitions with Gretzky: 8-2-2. Without the Great One, Fuhr was just 1-3-1.

6. Gilles Meloche played in 788 games over 18 NHL seasons, making the All-Star Game twice. Because of the poor quality of his teams, he played in only 45 playoff games, despite having a winning record in postseason play.

7. Every playoff game in Jeff Hackett's career was against a 100+ point team with either Patrick Roy or Martin Brodeur in the other net.

8. Chico Resch's career playoff GAA with the New York Islanders was 2.49, better than Billy Smith's 2.73 mark. Resch also had much better regular season statistics than Smith. Resch was traded to the Colorado Rockies on March 10, 1981, meaning he only got one Stanley Cup ring to Smith's four.

9. Michel "Bunny" Larocque went 111-26-20 from 1974-75 to 1978-79, playing with Ken Dryden and the Montreal Canadiens, and was on four Cup-winning teams. In his post-Montreal career with Toronto, Philadelphia and St. Louis, Larocque's record was 16-41-14 with a 4.87 GAA.

10. The 2003-04 Florida Panthers scored 188 goals and allowed 221, despite Roberto Luongo's .931 save percentage. Given their scoring level, the Panthers would have needed to allow 178 goals or fewer for an expected number of points that would put them in a playoff position. To accomplish this, Luongo would have had to save an additional 43 goals over the 2,475 shots he faced that season. That would have required at least a .948 seasonal save percentage just to give his team a chance at earning the 8th playoff seed in the East and a probable first round loss against the eventual Stanley Cup winners Tampa Bay.

11. In Dominik Hasek's record-breaking 1998-99 season, when he set the single-season save percentage record of .937, his team was the #7 seed in the Eastern Conference. His team did not have home ice advantage in any rounds of the playoffs, and ended up losing in the Final to Dallas, a team that was 23 points better than Buffalo during the regular season.

12. The 1981 New York Islanders finished first during the regular season with 110 points. In the playoffs, they did not play any of the next 7 highest ranked teams. Not surprisingly, they won the Stanley Cup, outscoring their opponents 97-48 in the process. Similarly, the 2004 Tampa Bay Lightning finished second overall during the season, and then avoided meeting any of the teams ranked 1st through 7th on their way to the Stanley Cup.

Two other fortunate teams were the 1991 Pittsburgh Penguins (avoided the league's 4 best teams, 82 point average opponent), and the 1986 Montreal Canadiens (avoided the league's 5 best, opponents averaged 84 points).

An example of an unfortunate team was the 2004 Calgary Flames, who faced the top 3 seeds in the Western Conference, as well as the top seed in the Eastern Conference, before losing the Stanley Cup Final in 7 games. The average regular season record of their opponents was 105 points.

Monday, November 19, 2007

"I don't see myself as that big of a deal"

Congratulations to Martin Brodeur on achieving his latest arbitrary milestone of heavily team-influenced statistics, win #500.

I'm sure Marty was just being modest about his accomplishments in the above quote, but if he is going to say things that support my thesis then I'm probably going to end up posting them.

Friday, November 16, 2007

Marty Turco is No Longer an Elite Goaltender

One of the problems with goalie evaluation is that reputations tend to linger far too long. Because of the inherent variability with the goaltending position, as well as the changing dynamics of teams, we need to be careful to jump to conclusions. Good goalies will eventually right the ship, just as the luck will run out for those who are temporarily playing over their heads. However, when presented with sufficient evidence of an improvement or decline in a goalie's game that is independent of team factors, his ranking should be changed, regardless of how many Cups, Vezinas, or anything else he has won or done.

I think the majority of hockey fans would rate Marty Turco as an above average goaltender. The Hockey News put him in the top 5 in the league at his position. In reality, though, he is not. There is some pretty good evidence that something negative happened to Turco's game over the lockout, as the splits are quite startling:

Pre-lockout: 185 GP, 1.91 GAA, .922 save%, 24.4 SA/60

Post-lockout: 148 GP, 2.42 GAA, .904 save%, 25.3 SA/60

A lot of that is the game opening up, and some of it is Dallas getting weaker, but the reality still is that right now Marty Turco is a mediocre goalie. He ranks 23rd in the league this season according to Hockey Numbers' shot-quality neutral save percentage. However, the reputation is still there, bolstered in part by his strong playoff performance last season. It is interesting that Turco was largely underrated earlier in his career because of his playoff struggles, but now is likely overrated because of one good stretch of playoff games last year against a weak offensive team. In addition, even though he is probably the league's best puckhandling goalie, he tends to receive too much credit for that ability, much like Martin Brodeur.

Monday, November 5, 2007

Exhibit 1,567 on Why Goalie Wins Are a Poor Statistic

This season, in the games Martin Brodeur has "won", he has given up 5, 4, 1, and 2 goals. In the games that he has "lost", he gave up 2, 3, 3, 4, 2, 3, and 1. Which leads to this unusual split:

In wins: 3.00 GAA, .876 save %
In losses: 2.56 GAA, .906 save %

The other 18 skaters on the team have a very large impact on whether the team wins or loses. Usually much more so than the goaltender. This has certainly been the case this year in New Jersey, and the reason has nothing to do with goal prevention but rather what is going on at the other end of the ice:

Brodeur's goal support in wins: 5.00 goals per game
Brodeur's goal support in losses: 0.71 goals per game

Goalie wins ignore half of the inputs that determine wins and losses (goals for), and leave the other half (goals against) unadjusted for number of shots against and shot quality. With all that noise clouding the data, it is impossible to tell from his win totals if a goalie is very good or just playing for a high-scoring or great defensive team (or even both). Brodeur's season to this point is a great example that goalie wins are not very meaningful at all in terms of evaluating goalies.