Hockey fans are used to instantaneous analysis. We like to hear about the key plays between periods, identifying the game breaking hits, goals, and saves. From a statistical standpoint we’re used to tracking shot counts, goals, assists, saves, hits, blocked shots, faceoff wins, etc. Unfortunately for the average fan, statistical analysis in hockey has made it increasingly clear that the chaotic nature of the sport makes drawing conclusions based on small samples extremely unwise. The amount of variation in scoring rates, save percentages, and even possession results, year to year – let alone within a single season – makes it a very difficult proposition to evaluate players with limited data sets.
More NHL on Sportsnet:
Subscribe: Rogers GameCentre Live
Rogers Hometown Hockey | Broadcast Schedule
Sportsnet Fantasy Hockey Pool
Early on in the NHL season there are numerous challenges to identifying which teams and players are playing at the highest level. The schedule is quite unbalanced as some teams have travelled more than others. Similarly, the strength of opponents some teams have faced is disproportionately difficult or easy.
Additionally, comparison is challenging due to the small amount of ice time each team has played in the wide array of game states that are possible. There isn’t much point in comparing performance between players at 5v5 when the game is tied if the time they’ve played in that situation is miniscule.
Despite the fact that we are around the 10th game of the season for the vast majority of NHL clubs, no team had played more than 300 minutes of hockey with the score tied at 5v5 going into Wednesday. In fact the average team had only played 191 minutes in that situation. At the rarer end of things, the average NHL team still has less than 150 minutes of 5v5 hockey while leading (or trailing).
This variety in exposure for each team is significant because they will play differently depending on the score at the time – a phenomenon referred to in analytics as Score Effects. Simply put, teams with a lead tend to sit back and play more defensively in order to protect their lead, while teams that are trailing tend to push the pace in order to catch up. Thus, players who find themselves playing predominantly with a lead will often end up with a reduced percentage of shot attempts, while players who have largely been trailing tend to post an inflated percentage.
Ideally we can account for the different situations players find themselves in by normalizing for the sake of comparing across game states. The method we use to accomplish this is based upon the logic of Team Score-Adjusted Fenwick – only applied to individual skaters.
First, we account for the league-wide spread in shot attempt differentials in each game state – rarer game states will show a wider spread in results. We then determine a weighted average for every NHL player’s percentage of shot attempts, weighting their results in different situations by the amount of TOI they have played at 5v5 in each game score state (trailing by 2+, trailing by 1, tied, leading by 1, leading by 2+).
The resulting statistic, Z-Score Adjusted Fenwick For Percentage (Z-Score Adj. FF%), gives us a reasonable tool for comparing the results of individual NHL skaters relatively early in the season. When we combine this with individual skater PDO (Sum of On-Ice SH% and SV% at 5v5) we can form a clearer picture of who is driving the successes and failures around the NHL so far this season.
Some of the more interesting highlights for Z-Score Adjusted FF% so far would be as follows:
WESTERN CONFERENCE
• Gabriel Landeskog (51.97%) is the only member of the Colorado Avalanche with a Z-Score Adjusted FF% above 50% in six-plus games played. This goes a long way in explaining their crash back to Earth early this season.
• Minnesota’s excellent start is largely being driven by their top five skaters at 5v5. Parise (64.48%), Pominville (63.99%), Granlund (62.16%), Brodin (61.37%) and Suter (60.18%) all highlight amazing possession results for the Wild so far.
• It also doesn’t hurt that Minnesota currently boasts the highest PDO in the league at 1042. The team’s elevated 5v5 SV% has begun to dip (13th in the NHL at .923), but their SH% is otherworldly right now (11.88%). That won’t last, and the goals will likely dry up a tad eventually.
• The only other team in the NHL with five skaters above 60% in Z-Score Adjusted FF% is Chicago, but they lag Minnesota slightly as only two of them are top-line players – Patrick Sharp (62.65%) and Marian Hossa (62.55%). The other three are Jeremy Morin (67.02%), Marcus Kruger (62.14%) and… Dan Carcillo (60.76%)? Yes you read that right.
• If we’re looking for players riding high on PDO – look no further than “that ‘70s line” in Los Angeles, where all three of Jeff Carter, Tyler Toffoli and Tanner Pearson boast an insanely high PDO of 1141 or higher (1000 is normal – anything above 1030 is unlikely to last long term). The line is seeing On-Ice SH% above 16%, and is playing in front of .980 SV% goaltending at 5v5.
• In Vancouver the Canucks have to be pleased with the early-season form of Henrik (56.03%) and Daniel (55.13%) Sedin. Radim Vrbata (54.61%) and Nick Bonino (51.65%) are also fitting in nicely.
• The Edmonton Oilers’ recent string of success is largely being driven by contributions from the likes of Jesse Joensuu (56.73%), Boyd Gordon (52.82%) and Jeff Petry (52.71%). Off-season additions Benoit Pouliot (45.89%) and Mark Fayne (43.99%) have struggled though, and Nail Yakupov (41.04%) is having a rough start to his season.
• Calgary’s early-season success is almost entirely PDO driven. Jonas Hiller’s 5v5 SV% of .960 ranks second in the NHL behind only Roberto Luongo’s .963. The only skaters for the Flames with a Z-Score Adjusted FF% above 50% are Mark Giordano (51.64%) and Jiri Hudler (50.63%). Expect a regression.
• Winnipeg has a number of players with half decent possession numbers, including Mathieu Perreault (55.38%), Blake Wheeler (54.89%) and Mark Scheifele (53.95%). That being said, they’ve missed Evander Kane’s goal scoring with a team that’s shooting 6.91% at 5v5.
• The Jets could also use some decent goaltending – the team is sitting at a .903 5v5 SV% so far.
EASTERN CONFERENCE
• The Avalanche might be missing the departed Pierre-Alexandre Parenteau (61.97%), who has the highest Z-Score Adjusted FF% on the Montreal Canadiens.
• A lot of the Habs are riding high on silly-good PDO results so far. Max Pacioretty and David Desharnais, in particular, are both sitting at 1076 thanks to .971 and .980 On-Ice SV% paired with 10.53% and 9.09% On-Ice SH%, respectively. Takes a lot of luck to be good this early.
• Buffalo is giving new meaning to bad hockey. No skater on the team has a Z-Score Adjusted FF% above 45%. The team leader is Chris Stewart (44.65%) and Drew Stafford is having an absolutely abysmal year so far (31.80%). Stafford is dead last amongst NHL skaters with six or more games played. The bottom-nine skaters in the NHL all play for Buffalo.
• The Sabres should probably send Sam Reinhart (32.71%) back to Kootenay. Not just because they’ll waste a year of his entry-level contract, but because his Z-Score Adjusted FF% is the second-lowest in the NHL.
• Ottawa has a problem with the play of Bobby Ryan (38.94%) at the moment. After eight games he ranks 607th out of 636 skaters across the league in Z-Score Adjusted FF%, and that just isn’t good enough for a guy making $5.1 million against the cap.
• Jack Johnson (41.79%), Andrew MacDonald (41.52%) and Brad Stuart (41.06%) are all still posting atrocious possession numbers. They rank in the bottom 11 defenders across the NHL – but six of those play for Buffalo – so let’s just say they’re in the bottom-five out of everyone not playing for the Sabres.
• The Leafs are hoping Stephane Robidas (42.31%) can round into form and find his game as he slowly recovers from breaking the same leg twice last season. So far his results are not good, as he ranks 14th from the bottom amongst defensemen in the NHL.
• PDO problems largely explain the trials for the two players in Toronto who were voted “most likely to come up in a trade proposal” early on this season. Jake Gardiner (933) and Nazem Kadri (938) are suffering as a result of unbelievably bad goaltending (.864 and .895 SV%) AND shooting (6.98% and 4.35% SH%) while they are on the ice at 5v5. Those numbers should improve.
• Kadri has already shown glimpses of improved results in terms of Z-Score Adjusted FF% sitting at 57.94% – currently tops on the Leafs. At the opposite end of the spectrum we have Tyler Bozak (43.54%) who sits third from the bottom on the team, ahead of only Richard Panik and the aforementioned Robidas.
The story painted with these numbers is far from complete, but hopefully this helps to illustrate how you can get a deeper insight into what’s happening across the NHL so far at this early point in the season.
