When two heads are better than one
Throughout the course of this season Dave Tippett has split the duties in net between Mike Smith and Mikko Koskinen. Smith has played 39 games and 2157 minutes, while Koskinen has played 38 games and 2096 minutes. By keeping the competition open for the starting spot Tippett has squeezed the best from both players and created a rare opportunity to accurately compare their performance.
Some journalists and analysts have reported that Smith and Koskinen have performed equally well, with some going so far as to claim it doesn’t matter which goalie starts in the playoffs. We believe that Tippett slightly favours Smith and that Smith is objectively the better of the two goaltenders. Smith and Koskinen are extremely different characters and goaltenders, and it does matter who starts. Like all good analysts we will bend the numbers to prove our point, and while doing so show that score sheet analytics are somewhat inadequate.
Mikko Koskinen is a 32 year old with a relatively small number of NHL games beneath his belt. He was signed to a one-year, 2.5 million dollar contract by Peter Chiarelli after finishing the 2016-2017 season in the KHL with a 0.937 save percentage. Chiarelli granted the 6’8” giant, who had only played four NHL games, a no-movement clause in order to secure his services. After 27 games of the audition Chiarelli upped the ante by extending Koskinen to a three-year 13.5 million dollar contract on January 21st, 2019. This somewhat bizarre move signalled the end of ever-talented Cam Talbot’s time with Edmonton. Koskinen had been fine but not great during his four month stint with the club and Chiarelli awarded a somewhat average performance with a more than average contract. It was Chiarelli’s last move as the Oilers GM and he was fired two days later.
Smith, a 38 year old veteran with 13 seasons of NHL experience, signed a one-year free agent deal with the Oilers July 1st, 2019. Ken Holland awarded Smith a two million dollar contract with up to $750,000 in performance bonuses. To date, Smith has won $600,000 in bonus money and if he backstops the Oilers past Chicago he will earn the maximum.
When Smith was acquired there were a number of free agents available including premium starters Semyon Varlamov, Sergei Bobrovsky and Robin Lehner. Ken Holland was in competition for UFA starters in a lower band after having been saddled with Koskinen’s hefty salary and no-movement clause. Such goalies possibly included Cam Talbot and Petr Mrazek. Of the three (Talbot, Mrazek and Smith), Talbot is a notch-above but most likely was not an option after Chiarelli had poisoned the well. Tippett and Smith have a long history and one wonders to what extent Holland consulted Tippett when he brought the veteran on board.
Smith started his career with Tippett as his head coach in 2006 and played 42 games over two seasons with the Stars as the back-up to Marty Turco. In 2008, Smith was sent to Tampa as part of the Brad Richards trade and in 2011 backed-up 42 year old Dwayne Roloson as the Bolts made their way to the conference finals. In the summer of 2011 Smith signed with Tippett’s Coyotes and became the defacto starter when he was healthy, playing more than 50 games four of his six seasons with the club. In 2012, Smith lead the Coyotes to their first playoff series victory in 25 years resulting in a six-year 34 million dollar contract. He completed the final two seasons of the deal in Calgary after being traded for non-roster players. When Smith signed with the Oilers he had played 571 NHL games but was coming-off of one of the worst seasons of his career. His 2018-2019 season saw him finish with a 0.898 Save Percentage and 2.72 Goals-Against Average while playing behind some of the better defensive pairings in the league. Ultimately he lost his job to Czech stick-flipper David Rittich.
Tippett opted to start Smith slightly more often than Koskinen before the season was clipped short. Smith was on the ice for the national anthems 37 times to Koskinen’s 34. But these starts did not always come at low risk. Smith struggled at times before the new year and had a rough November. Smith started four games that he did finish, while Koskinen was pulled twice.
- October 27, 2019, Panthers 6, Oilers 2: Smith pulled after giving-up three goals to start the 2nd period. Was solid in the first small injury sustained in the first period.
- November 21st, 2019, Oilers 1, Kings 5: Smith pulled after struggling in the first period. Smith struggled in four of last seven starts.
- December 4th, 2019, Senators 5, Oilers 2: Koskinen pulled after giving-up three unanswered goals to start the second period.
- December 28th, 2019, Oilers 1, Flames 5: Koskinen pulled late in the second after the Oilers give up their 4th goal.
- February 1st, 2020, Oilers 8, Flames 3: Smith ejected after a fight with Cam Talbot with 24 seconds left in the second period. Smith had some awkward moments handling the puck. Gave Gaudreau a wide open net in the first 90 seconds of the game and was forced to deny him with a big save.
- March 5th, 2020, Oilers 3, Blackhawks 4: Smith pulled mid-way through the second period, beaten on two longer shots and a wrap-around.
Possession is nine-tenths of the law
Although Smith and Koskinen are both tall goaltenders, at 6’5” and 6’7” respectively, their similarities stop there.
Koskinen is a textbook butterfly style goalie who tracks the puck well, plays the angles and is able to use his size and his enormous pads to his advantage. He has a good glove hand and has decent lateral movement for a goalie of his size. Koskinen is a master of positioning and tracking and is somewhat conservative in his movement and approach, staying low to the ice in order to drop into the butterfly position. He uses his enormous length in order to make-up for any limitations in lateral movement. Koskinen’s play creates a higher rate of rebounds, 4.62 per 60 minutes of 5v5 play this season, ranking him 3rd highest in the league.
Smith, who also uses the butterfly style, is extremely active in playing the puck and is influenced by Marty Turco whom he supported the first two years of his career (Smith uses the Turco grip). Smith is extremely athletic and active and is an expert at reading the play. His ability to shoot and pass is exceptional. October 19, 2013 he became the eleventh NHL goaltender to have scored a goal in competition. Smith’s reading of the play, puck handling skills, and rebound control creates fewer shots against. This season Smith ranked 49th highest in rebound attempts against at 2.85 per 60 minutes of 5v5 play.
The following charts Shots Against per 60 minutes of 5v5 play (SA/60), Goals Against per 60 minutes of 5v5 play (GA/60) and 5v5 Save Percentage for all 71 games of the 2019-2020 season.
Smith’s games are in blue while Koskinen’s are in orange. Games shared between the goalies are shown in grey. Games involving higher SA/60 and SV% belong to Koskinen, while games with lower SA/60 belong to Smith. The right of the chart is orange, while the left of the chart is blue, and the middle of contains blue, orange and grey bubbles. The higher the bubble, the higher the GA/60. The bigger the bubble, the higher the SV%.
It makes sense that Smith, an expert puck handler, creates games where the Oilers face fewer shots: he is able to foil a higher percentage of zone entries using a dumped puck than Koskinen. Smith causes the puck to exit the Oilers’ zone faster than it would otherwise and as a result the opposition receives fewer attempts. In other words, Mike Smith is a possession style goaltender who gets the puck moving up ice quickly. His possession style of play also may also contribute to a slightly lower save percentage than goalies who do not move the puck in this manner, possibly because his SV% is not bolstered by shots in low danger areas. This is, of course, when everything is going well. When things are going poorly for Smith and he does not read the play properly or is not able to communicate and connect with his defenders things can get... exciting.
If what we say is true, then Mike Smith should have better possession stats than Mikko Koskinen. The table below shows all situations possession, shot, goal and save percentage for Smith and Koskinen this season.
Across the entire season Smith’s play results in better shot, shot-attempt, and Scoring Chances data. For the most part his better for-percentage scores (CF%, FF%, SF%, SCF%) are a product of fewer shot-attempts and chances against (CA/60, FA/60, SCA/60). Although Smith’s play has resulted in slightly worse GA/60 and SV% than Koskinen, he more than makes-up for these slight negatives with a superior GF%. If one looks at the Oilers’ goaltending in the new year, Smith is shown to be significantly better than Koskinen, posting a SCF% of 52.24 versus Koskinen’s 40.24 and a GF% of 58.2 versus Koskinen’s 50.91.
There is more to goaltending than SV% and stopping the puck
Ultimately we are criticizing analytical approaches for goaltenders that are limited to examining goals-against and save percentage. Smith’s slightly-worse stats in this regard appear to be out-weighed by his on-ice possession and GF%. We do not necessarily suggest here that all goalies be measured only by GF%, but rather that in Smith’s case we do believe what we see both on spreadsheet and on the ice. Mike Smith might not stop the puck quite as often as Mikko Koskinen, but to the kean observer this is actually a good thing. Objectively Smith should be the Oilers starting goaltender, and unless he sees something vastly different in the next few days than he’s seen in the new year, Tippett will likely start Mike Smith in net against Chicago.
All data from Natural Stat Trick.