Tim Thomas: The Best Goalie in the Last Half Decade

To evaluate the performance of goaltenders over the last 5 years, I use as a point of comparison in each year the average save percentage of goalies who were considered #1 (starter level) and the average save percentage of goalies who were considered backups (replacement level) in each year. The goalie composition of the two peer groups can change from season to season considering whether they qualify or not the group criteria (a starter needs to have played at least 49 games and a backup less than 25 games in a particular season).

Data from the above table taken from here



Relative performance of goaltenders

The following ranking is based on the cumulative even strength save percentage relative to the peer group of #1 goaltenders over the last 5 years (highlighted in blue). We use the even strength save percentage for the ranking, because team effects can have an important effect on the save percentage of goaltenders on the penalty kill. Only seasons were goaltenders played more than 45 games are considered and factored in the statistics. To be on the list, a goaltender needs to have either 1 season with 60+ games or 3 seasons with at least 45+ games played: 32 goalies qualified. (Click the image to enlarge)

 Remarks:

• The Nb of Seasons column counts the number of seasons the goaltender was considered the starting #1 goalie (criteria used: more than 45 games played).
• Highlighted in orange: the number of games played by goaltenders who have played more than 300 games over the last 5 years (only 7) and the number of seasons of goaltenders who were considered a #1 goalie 5 times in the last 5 seasons (only 8).
• Highlighted in red: Best year statistics after which a long contract (4 years or more) was signed less than an half-season after the best performing year of 45+ games in the observed period. Note: Fleury's big contract was given after a strong year of less than 45 games played
• Highlighted in purple: Best performing year statistics under coach Ken Hitchcock
• Highlighted in green: Worst performing year statistics which have a positive Rel EV Sv% (only goaltenders with at least 3 seasons of 45+ games are considered). In other words, the goaltender never underperformed the peer group of #1 goalies, a feat accomplished by only 4 goaltenders.

Interpreting the data from the table

During the 4 seasons Tim Thomas was considered a starter, he faced 6895 shots of which he stopped 6401 for an overall save percentage of about .928. During that period, his relative cumulative out-performance relative to starters was .054 or 5.4% for an average "added value" of .0135 or 1.35% (5.4% / 4) per season over the "average #1 goaltender". He thus prevented about 93 more goals or 23 more goals per season than the average starter. Compared to the "average" backup goalie, he prevented 170 more goals or 42 more goals per season (this however accounts even less for the quality of competition and is understated).

Consistency of goaltenders

Next, we look at the consistency of goaltenders over the last 5 years as measured by the concept of Target Semi-Variance (all games during the last 5 years are considered).

http://www.investopedia.com/terms/s/semivariance.asp#axzz1yrZfLHq9

This measure is particularly sensible to bad games and games where the goalie is pulled out (or comes in as substitute) as the save percentage of each game consists in an observation.

Remarks:

• Highlighted in orange: the save percentage of goalies who achieved a mean save percentage above .918 during the last 5 years (only 6)