**Updated to Include the 2016-2017 Season through March 18th with additional tracked statistics. Blog on these statistics coming before the end of the month.
The shootout is a major part of today’s game. Of the past three seasons, exactly 14% of all NHL regular season games go to shootouts. Even though the league may be looking at ways to extend overtime thus putting less emphasis on the shootout, it is here for the immediate future. With the shootout being such a big part of the game and a crucial second point at stake (see the 2013-2014 season where NJD went 0-13 and NSH went 2-9 in shootouts where both teams missed the playoffs by 5 and 3 points respectively), it is amazing it isn’t studied more in depth. Individual teams definitely track shootout data privately but nothing even remotely advanced is available publicly besides the NHL website that splits up how player do at home vs. on the road which isn’t all that useful.
The beginning of the hockey analytics movement was met with concern as the game is continuously flowing and and can’t be easily tracked like baseball. In baseball, you have a repeated battle between a pitcher and batter where gameplay doesn’t flow from play to play. The pitcher stands on the mound and batter sets up in the batters box then a pitch is delivered from the mount to home plate. After each pitch, both the pitcher and batter reset positions and the battle ensues. The sport of hockey is very different from that description of baseball. Hockey’s shootout is the one aspect of the game that is very similar: the puck always starts at center ice, the goalie must start in the crease, and after the shot attempt, the puck is picked back up and placed at center ice for the next shooter to pick up. This creates a situation where data collection is very feasible and trackable.
Players must have tendencies to repeat shootout moves. Goalies must be stronger against different handed shots, different shot types, or even location of the shot. Teams must have tendencies to send out certain players and goalies, but are they statistically making the right choice of who to send out? With no data to back up my thoughts, I set out to answer my own questions.
I have completely compiled shootouts from the 2012-2013 through the 2014-2015 season. The data mainly came from watching each shootout attempt and noting what happened within my spreadsheet. The data from the NHL is often wrong (How is it possible to have a deflected shot or a tipped shot in the shootout? There were no slapshots taken in Washington/Florida’s 20 round shootout even though the shootout report says differently). With 445 total shootouts tracked to date, I thought now would be a good time to publish my findings so far and to give me a break from going crazy after watching over 3100 shootout shot attempts.
I was going to extend this project back one more season to 2011-2012 but am not going to at the moment as the percentages haven’t changed much over the years and the team’s personnel has changed significantly since then. It is also very likely that shooters and goalie’s tendencies have changed over the years.
I’ve broken up the data into different categories that I’ll explain in the following segments. Without further adieu, here are my shootout statistics that can be viewed and downloaded in the links:
5 YEARS COMPILED (2012-2013 to 2016-2017) via DROPBOX or GOOGLE DOCS
2016-2017 SEASON (THROUGH MARCH 18) via DROPBOX or GOOGLE DOCS
2015-2016 SEASON via DROPBOX or GOOGLE DOCS
2014-2015 SEASON via DROPBOX or GOOGLE DOCS
2013-2014 SEASON via DROPBOX or GOOGLE DOCS
2012-2013 SEASON via DROPBOX or GOOGLE DOCS
Overall the shooting percentage on 3190 shootout attempts was 31.94%. 2012-2013 recorded the highest scoring percentage of any post-2005 lockout season with a 35.82% success rate. This occurred in the shortened lockout season and likely would’ve regressed down closer to the 3 year average with more attempts. The 2013-2014 season saw the second most shootouts in league history and shooters scored on 31.65% of all attempts which is below the three year average.
In 170 shootouts in the 2014-2015 season, shooters are only shooting 30.30%, which is on pace for a record low. No season in history has the shooting percentage been less than 30.50% before this. One explanation for a lower shooting percentage this season is a rule change that removed the dry scrape before the start of overtime and no cleaning of the ice prior to the shootout. This change has led to worse ice conditions for the shooters which in result, has favored the goaltenders.
Looking at the shooting percentage for the entire season doesn’t tell us much so the shooting data was broken up in smaller categories that could be analyzed better. This included shot type, handedness of the shooter, distance the shooter shot from, the physical location where the puck was shot in respects to the goaltender, and whether the shot attempt was a shot or deke.
The most successful shot types were backhand and snap shots which were both well above the league average. Wrist shots accounted for a majority of total shots taken but resulted in a below average scoring percentage. Slap shots were few and far between and resulted in the lowest scoring percentage.
Over 57% of all shootout attempts over the past three seasons were taken by lefties. Approximately 60% of all NHL players are left handed so there doesn’t appear to be any lefty biased in the shootout. It is interesting to find that even though lefties had more shot attempts, right handed players found more success in the shootout by almost 2.5%.
Looking at where the shooters shot the puck, there was no clear favorite. Shots were spread around relatively evenly. Overall, goalies are stronger down low on both their glove and blocker sides where shooters have a below average shooting percentage. On shots towards the high glove, high blocker, and fivehole, shooters are highly favorable above the league average. One number that really stands out is that almost 50% of all shots taken towards the goaltender’s high blocker went in the net. Breaking it down even further, 72% of all backhand shots taken towards the high blocker scored. Breaking that figure down even further, backhand shots taken from the crease and shot towards the high blocker scored on 75% of the opportunities.
Looking at the distance from the goal line that the shooters shot the puck, it was clear that to be successful, the shooter had to back the goalie up into the net. To do this, they attack with speed and hold the puck until they are close to the goal line. Almost 34% of shots from inside the crease (distance of six feet) were converted into goals. More impressively, 66% of shots from in the crease that were shot high either to the glove or blocker side were put in the back of the net. Though most of the shots taken were from the crease to the hashmarks (total distance of 14 feet), attempts were converted into goals at a below average rate. Shots from outside the hashmarks (over 20 feet from the goal line) rarely went into the net and should be avoided.
Shooting vs. Deking
Looking at specific attempts to whether the shooter shot or deked, the results are right on point with our earlier findings. A majority of attempts were shots but deking had a higher scoring percentage. We’ve already established that the closer a shooter gets to the net, the higher scoring chance he has and dekes tend to be executed closer to the goal line than a normal shot,
It is known that home ice advantage exists in hockey and in recent times, home teams have won 55.7% off all NHL regular season games. But when looking specifically at the shootout, having home ice is actually a disadvantage. In the past three seasons, the home team shot and saved about 1.1% less than the away team. These two statistics combined for the home team to win only 48.99% of shootouts which is both less than .500% and considerably below the home ice advantage winning percentage of 55.7%.
While overall the road team shot higher, on most cases, teams shot consistently in the same range whether they were home or on the road.
The main advantage the home team has in the shootout is having the choice whether they want to shoot or defend first. In over 83% of all shootouts, the home team decided to shoot first but is that the right call? Looking at the past few years, teams have a losing record no matter of the coach’s decision to shoot or defend first even though shooting first results in a higher winning percentage. The sample size of defending first is so small compared to shooting first that there is a good chance that the data is skewed. With an equal data size for each category, is is very possible that defending first is a better option. With more research, I hope to find a definitive answer for this subject.
Goal Scorers in Game
One interesting question I had when starting with this project was would a player who scored in the game prior to the shootout have more success than a player who hadn’t? I assumed so as the shooter who already had success on the goalie could build off of it and score again. My hypothesis was confirmed as shooters who already scored at least one goal in the game (they accounted for about 22% of all attempts) did shoot better than shooters who hadn’t scored in the game. With prior goal scorers accounting for one attempt out of every 4-5 attempts, it is clear coaches are sending them out in the shootouts and they are continuing to have success.
Going deeper with this topic, I looked into shooters scoring two or more goals in the game vs. shooters who had only scored one goal in the game. Two plus goal scorers had a slightly worse scoring percentage than one goal scorers (31.42% compared to 33.64% respectively). Since the difference was small and there were almost ten times as many attempts by one goal scorers, I’ve reached the conclusion that the amount of goals scored in regulation doesn’t effect the shooter’s shootout shooting percentage.
Alex Ovechkin had the only attempt where a shooter scored a hat trick in the game (4 goals on 12/10/13 but was stopped in the shootout).
A number of players shot above the league average and found major success in the shootout. Though many other players shot above average or had more goals, I’ve highlighted my top 17 shooters that were key to their teams obtaining shootout success. Jonathan Toews led the league with 18 shootout goals and was followed by TJ Oshie and Frans Nielson at 15 shootout goals then Jakob Silvferberg and Logan Couture with 14 shootout goals. Behind them, Patrick Kane scored 13 shootout goals, Corey Perry and Mikko Koivu scored 12 shootout goals and Matt Duchene, Alexander Steen, and Nicklas Backstrom each put 11 goals in the net. Rounding out the double digit totals, Sidney Crosby scored 10 shootout goals.
Some players are known as one trick ponies which can lower their effectiveness in the shootout. For example, Mats Zuccarello (7 for 16 in shootout attempts) is known for picking up speed from center ice, gliding through the hashmarks, and picking a corner to shoot towards.
A common thread between the success of these top players is that they have an arsenal of moves and are always trying something new. The goalie is left guessing on where the shot is coming from or going and often guesses wrong. While the attempt may be completely different, they are all of high quality (statistically having a higher percentage of going in). This included driving the puck to the net, shooting high, and effectively picking a shot type. For example, Logan Couture didn’t take more than 32% of all attempts towards any one location on the net or 46% from one specific shot type.
You’d expect the highest goal scorers to be the team’s most clutch snipers in the second or third round, but it was the exact opposite. A majority of the attempts by these top scorers were from the first round. For example, 25 of TJ Oshie’s 26 attempts over the past three seasons have been in the shootout’s opening round. It is obvious that the team’s strategy is to get an early lead and defend it with solid goaltending and secondary scoring.
This list is what I consider the top tier shootout goalies. They’ve had high success saving shots which has led them to be successful in winning games.
A number of goalies on this list have extremely high save percentages compared to the league average of 68.12%. A strong goaltender can make a bad team good and a good team great. This rationale applies to the shootout as well. Saving the puck well over 70% and in some case 75% of the time is phenomenal and has a huge impact on the team’s overall success. Their stellar play is key to their respective teams winning shootouts and are usually relied upon in winning shootouts by a score of 1-0. Sergei Bobrovsky and Ryan Miller lead this category of elite goaltenders. Of these goaltender’s teams, only Sergei Bobrovsky’s Blue Jackets have topped the league average of 34 shootout goals scored per team over the past three seasons thus confirming how essential these goaltender’s play is to their team’s shootout success.
Other goalies such as Braden Holtby, Brian Elliot, and Antti Niemi had major success in the shootout while having a save percentage that is only slightly above the league average. These goalies aren’t counted on as much to steal a shootout and even with them playing below their best, their team can pick up the slack to win the games. Not surprisingly, Washington (48) , St. Louis (51) , and San Jose (46) all scored well over the league average of goals per team in the past three seasons. A good goaltender combined with a strong offense often leads to success.
Over the past three seasons, over half the league achieved winning records in the shootout while two teams were .500 and eleven teams had losing records. I’ve highlighted the eight teams with the highest winning percentage. It’s important to look at how they got to the winning records.
Seven of the eight teams shot and saved better than the league averages. Having a strong offense and a strong goalie in the shootouts is key to long term success. Most of these teams had a shooter or goaltender if not both near the league’s top. St. Louis was overly impressive shooting at 49% (51 for 104 attempts) and led the league easily in shooting percentage. Having two top shooters (Oshie & Steen) along with a top goaltender in Elliot contributed to their success. Columbus had the league lead in save percentage by almost 4% and was almost 10% over the league average.
The one exception to this trend was Colorado who had a slightly below average save percentage but boasted an exceptional shooting percentage (2nd highest in league) which saved their poor goaltending. A majority of their wins were by the score of 3-2 or 4-3.
Looking at the team’s PDO (shooting percentage added to their saving percentage) is very interesting as well. Nearly every team with a winning record in the shootouts had a PDO of over 100 (Winnipeg was the one exception going 17-15 in shootouts and having a PDO of 99.96). Every team with an even or losing record had a PDO of below 100. St. Louis, who had the most success in the shootout, had the league’s highest PDO at 120.47 while New Jersey (4-26 record) by far had the league’s worst record and lowest PDO at 76.64.
While most teams with a losing record in shootouts over the past three seasons were only a few games under .500, a handful of teams really struggled. This includes New Jersey (who at one point lost 18 straight shootout games), Philadelphia, Detroit, and Nashville. They shot and saved well below the league average which directly led to their lack of success in the shootout. Most teams that had losing records in the shootouts were around average at either shooting or saving but were well below average in the other category.
Looking at these team’s PDO rating, their winning percentage ranking correlated perfectly with their rank of their PDO. New Jersey was far below everyone else at 76.64 and the other three struggling teams were in the 80’s. Every other team with a losing record still recorded a PDO above 90.
These teams could use these statistics to show their players what they should be doing versus what they are doing. A majority of their attempts are of low percentage. It doesn’t help when New Jersey is missing the net completely on over 28% of their attempts (league average is 16% of shots miss the net). They need to look towards St. Louis, Pittsburgh, and Columbus and copy what they are doing in the shootout because it is working effectively. These teams may not have the same personnel that St. Louis can put out in the shootout but it is definitely something to strive for in the long term.
How Teams Could Benefit from These Statistics
These statistics are interesting to look at, but how can individual teams use them and benefit from them? The debate on analytics is whether the data being analyzed is theoretical or actionable. They are only effective if applied within the team’s system.
This information can help out a coach wondering who to send out in the shootout. In the fourth round of a shootout, should Washington Capitals head coach Barry Trotz send out right-handed shot Alex Ovechkin (26% shooter over the past three years)? Most people would say yes because he’s an elite goal scorer in regulation, but the answer depends on many other variables. Say they are playing Edmonton with Devan Dubnyk in net. Dubnyk is much stronger against righties (67% SV% compared to 58% against lefties). Add in left-handed shot Brooks Laich (3 for 3 in shootout attempts over the same period as Ovechkin) who has scored earlier in the game. Now the statistics highly favor Laich having a higher chance of scoring than the superstar Ovechkin. This one switch could be the difference between winning and losing the shootout and over the course of the season, could add a good 5 points towards the team’s place in the standings.
More importantly, these statistics give important information about both your team’s and your opponent’s goaltenders. The goaltender is the most important player in the shootout and his play often determines who wins and loses the game.
It is important for a team to determine what goalie they want in net for the shootout. The NHL permits teams to change goalies in the period between overtime ending and the shootout beginning. If your backup goaltender is stronger in shootouts than your starter would you want him in net instead? One big concern about this is the backup goalie coming in cold but if this decision is made early enough, the backup goalie could get the entire overtime period to stretch and get loose in expectation that the game will go to a shootout.
A perfect example of this is the Detroit Red Wings. Jimmy Howard is an all-star goalie but is among the league’s worst in shootouts (9-18 record, 57% SV% over the past three seasons). With Peter Mrazek (2-2 record over that time span but a 67% SV%) coming off the bench in a role similar to a closer in baseball, the Red Wings would have a much higher chance of winning the game.
Teams dedicate valuable time to the shootout every week. These statistics can help a coach determine what areas of a player’s game need the most work. Colorado goaltender Semyon Varlomov is very weak facing shots up high compared to shots down low. Colorado’s goaltending coach, Francois Allaire, would want to know this so he can focus more time in practice working on positioning and technique for saving shots up high. This may include being more aggressive coming out challenging skaters so the number of high shots is cut down. This work would make him a better all-around goaltender.
On the flip side, if you’re a team facing Varlamov, it is important to know the goaltender’s strengths and weaknesses. To be successful, it is key to target and attack a goaltender’s weaknesses instead of playing into their strengths. Varlamov has saved 79% of shot attempts shot towards his low blocker side but only 30% shot towards his high blocker over the past three seasons. A coach knowing this can relay the message to his players to shoot high rather than low if the opportunity presents itself.
If anyone wants to discuss shootouts further or has any suggestions, feel free to reach me via Twitter at @QuickkNess over via email at SNessHockey@gmail.com. The spreadsheets include only the raw numbers as it took up much less data. If you’re interested in playing around with the numbers and want my original formulas, please email me as I’d be happy to share it with you as long as I get credit for the data collection.