Visualizing goaltender performance: 1982-2014

Comparing goaltending between eras

This analysis was inspired by OilFaninYYC’s recent article, which posited that the Oilers have had some bad goaltending throughout their history. Many Oilers fans are concerned about this season’s performance of Ben Scrivens and Viktor Fasth, which has been very-much below average and has led to questions of how our current duo compares to goaltenders from past eras.

League-wide all-situations Save Percentage (SV%) has generally risen since we’ve been keeping track of it, from 0.876 in '82-'83 to 0.914 at present. Exceptions to this bullish performance include the '92-'93 post-expansion and the '05-'06 post-lockout seasons when percentages dropped 0.003% and 0.010% respectively. All sorts of factors have led to the rise of SV% including the defensive techniques of the dead-puck era, goaltending equipment, advances in statistical analysis, and technological advances such as video review.

The number of goals scored per game in the NHL also varies by season and era, and has generally declined since the 1980s. An exception to this bearish tendency occurred post-lockout when goals per game slightly increased in '05, due in part to more frequent powerplay opportunities.

All of this makes accurately comparing goaltending performance between eras almost impossible.  However, we can compare a goaltender’s performance relative to that of his colleagues for a given season.  Previously on Copper & Blue I visualized Save Percentage Plus, which is a statistic that measures SV% relative to the league-average and provides this type of measure. In this analysis I did something similar with three different kinds of data: SV%, Goals Against Average (GAA), Saves (SV), and Shots-Against per 60 minutes (SA60). I compared each goaltender's data to the league average for each season by fitting it to a curve.

Process used for visualization

1. Compiled data from Hockey Reference and Hockey DB.
2. Removed goaltending performances less than 600 minutes from each season.
3. Normalized and standardized SV%.
4. Normalized, standardized and inverted GAA.
5. Calculated, normalized & standardized, then classified SA60 data into categories (SA60CAT).
6. Visualized data.
   

How to "read" the visualizations

The higher the bubble, the greater the SV%. The further the bubble is to the right, the better the GAA. The bigger the bubble, the higher the SV. The darker the bubble, the higher the shot volume (SA60). Zeros and accompanying lines indicate league averages, all other lines indicate standard deviations. Positive values for SV% and GAA indicate above average-performances.

I standardized SA60 data and classified it into the following categories (SA60CAT):
•    One: highest shot volume, 3 standard deviations above the mean (black).
•    Two: higher shot volume, 2 standard deviations above the mean (gray 3).
•    Three: average shot volume (gray 1).
•    Four: weaker shot volume, 1 standard deviation below (light gray 1).
•    Five: weak shot volume, 2 standard deviations below (light gray 3).
•    Six: weakest shot volume, 3 standard deviations below (white).

League-wide visualizations

Interactive chart and table viewable here.

Figure One displays goaltending performances greater than 600 minutes since the ‘82-‘83 season (n=1753). Mousing-over bubbles reveals player name, team, season, and associated data.

There is an inverse relation between SV% and GAA. As SV% increases GAA decreases (visualized GAA data is inverted, and the trend is from the bottom left-hand corner to the top-right).

A positive relation between shot-volume and GAA is present. The darkest bubbles tend to be associated with performances involving below-average GAA, as the left side of the chart is darker than the right. Ostensibly, darker bubbles are associated with performances on defensively weak or non-possession teams. Large, dark bubbles in the upper-right quadrant are performances where goaltenders "beat the odds" and produced average or above-average results while facing very high shot-volume (e.g. Luongo '03-'04). Goalies on strong defensive teams tend to have white, smaller bubbles (e.g. Brodeur most seasons with NJD, Elliot '11-'12).

Bubble size corresponds to raw SV. Larger bubbles tend to be higher, as goalies who stop pucks tend to get more time-on-ice.

Exceptional performances include: Roland Melanson (NYI, '83-'84), Curtis Joseph (STL, '92-'93), Dominik Hasek (BUF, '94-'95, '96-'97, '97-'98, '98-'99), Jose Theodore (MTL, '01-'02), Dwayne Roloson (MIN, ‘03-'04), Cristobal Huet (MTL, ‘05-'06), and Tim Thomas (BOS, '08-'09, '10-'11). These bubbles are two standard deviations above the mean for SV%, one standard deviation above the mean for GAA, involve more than 1000 SV, and incorporate average or  greater-than average shot volume. Only Hasek and Thomas have performed at this level more than once, and Hasek did it four times. Interestingly, two of the goaltenders on this highly selective list played for the Oilers but only after their elite seasons.

Some of the "greatest" goalies in Figure One are also the "worst". Among the worst performances visualized are Melanson’s '82-'83 season with the Islanders and Roloson’s ‘11-'12 season with Tampa Bay.

Interactive chart and table viewable here.

Figure Two is a visualization of the current NHL season. Bubble colour corresponds to SA60, with the exception of Edmonton players who are shown in orange. In order to build Figure One I first visualized each season separately. Interactive visualizations and tables for previous seasons may be viewed at the following links: '82-'83, '83-'84, '84-'85, '85-'86, '86-'87, '87-'88, '88-'89, '89-'90, '90-'91, '91-'92, '92-'93, '93-'94, '94-'95, '95-'96, '96-'97, '97-'98, '98-'99,'99-'00, '00-'01, '01-'02, '02-'03, '03-'04, '05-'06, '06-'07, '07-'08, '08-'09, '09-'10, '10-'11, '11-'12, '12-'13,'13-'14. Players who played for more than one team in a given season are displayed once as a total value (e.g. Scrivens & Fasth '13-'14).

"Ladies and gentlemen your Edmonton Oilers..."

Interactive chart and table viewable here.

Figure Three presents the same data as Figure One, however, Oilers bubbles are orange while bubbles from other teams are clear. Only two performances are more than a standard deviation above-average: Moog’s '84-'85 season (939 saves) and Markkanen’s '01-'02 season (312 saves).

Interactive chart and table viewable here.

Oilers performances are displayed on their own in Figure Four.

Surprisingly the Oilers have generally not allowed their goalies to be completely bombarded by shots relative to the rest of the league. There are no very dark bubbles (SA60CAT One). Bill Ranford faced above-average shot volume in '93-'94, playing more than 4,000 minutes and making more than 2,000 saves. Although perhaps not a "Hasek quality" season, by most measures it’s still pretty darn good. His '92-'93 season was similar in terms of performance, with slightly less shot volume. Dubnyk has faced more shot volume than some other goalies. His '09-'10 and '13-'14 seasons are close to this season’s performance by Fasth. Brathwaite’s '94-'95 season is also in this ballpark. These performances are approximately 2 standard deviations below average with regards to SV% and GAA. Of these terrible seasons, only Dubnyk’s '09-'10 season involved greater than average shot-volume (darker bubble).

Three performances from the '05-'06 season (Markkanen, Morrison and Conklin) show below average SV% paired with average or above average GAA. A strange combination, until one remembers that Chris Pronger played on this version of the Edmonton Oilers and likely buoyed their GAA.

Interactive chart and table viewable here.

Figure Five shows performances pooled across seasons. These bubbles are the averages of the bubbles from Figure Three, excluding goalies who only played one season for our beloved team. Generally speaking, Roloson, Fuhr, Moog and Salo were above average during the course or their time with Edmonton. But most of our goalies are in "C-" territory relative to their colleagues in the NHL: below average but above negative-one for both SV% and GAA. Concerning number of saves, Bill Ranford has made the most at 12,286. Grant Fuhr is second in this respect with 9,678.

In some ways Ben Scrivens is typical of the kind of goalie that has played for the Oilers. Clearly Scrivens is capable of fantastic play in some situations. But to date he is in solid "C-" territory. Keep in mind his numbers in this analysis include his performance with LA. When those numbers are removed Scrivens sinks significantly.

Viktor Fasth, on the other hand, has numbers to date that are some of the worst in our history and are similar to Freddy Brathwaite’s. But maybe if he gets the chance to play more games those numbers will rise.