The stats revolution in hockey is over. Teams are hiring bloggers left and right to acquire their information and talents. The advanced stats sites are popping up constantly (and some are subsequently being bought out by teams). Coaches are using them for real-time decisions, like St. Louis Blues coach Ken Hitchock.
For those looking for the advanced numbers, I recommend Behind The Net, Hockey Analysis, War On Ice, Progressive Hockey, and Hockey Stats. Each is unique in the information they present, and each brings value. Spending a night figuring out the use, and information, on each of the sites is a night very well spent.
Of course, just getting a grasp of what these things mean is the launch point. I use analytics all the time in fantasy hockey, and they can help owners gain an edge over the casual fan. Here are quick definitions for the important ones.
Puck Possession
Corsi
Puck possession is not calculated in time of possession, but rather in shot attempts. That’s what Corsi is: shot attempts. Any shot attempt, even if it was blocked or missed the net. Most of the time, these are measured while teams are at 5-on-5 (where most of the game is played).
Fenwick
Fenwick is just a cousin of Corsi. It refers to all shot attempts, without including blocked shots. Blocking shots can be considered a skill, which is why it’s not included in Fenwick analysis. In a small sample of games, Corsi has more predictive value. In a larger sample, it’s Fenwick. Again, it’s usually measured in terms of 5-on-5 play.
CorsiFor Percentage / FenwickFor Percentage
While the basic Corsi (and Fenwick) definitions describe a shot attempt, they need context. CorsiFor percentage (and FenwickFor percentage) is the percentage of shot attempts that occurred that belonged to a particular team or player. For example, if Los Angeles is playing Chicago, and Los Angeles has 60 shot attempts while Chicago has 40, Los Angeles would have a 60-percent CorsiFor (60 of 100 total shot attempts that occurred). They can be expressed at either the team or player level, and anything above 55-percent is elite.
ScoreClose
For predictive purposes, CorsiFor percentage and FenwickFor percentage are best used in a ScoreClose situation (also sometimes just called “Close”). This describes a game state where the two teams playing are within a goal of each other in either of the first two periods, or tied in the third. The reason for this is that teams down a couple of goals or up a couple of goals will change their strategy. That doesn’t give an accurate reflection of talent – most teams with a lead go into a defensive shell, which drives down their shot attempt rates.
Relative
Possession is measured with shot attempts, and being average isn’t necessarily bad. But a 50-percent CorsiFor player on the Chicago Blackhawks isn’t nearly as impressive as a 50-percent CorsiFor player on the Buffalo Sabres. Thus, CorsiRel (and FenwickRel) measures a player against his own team’s numbers. It’s one way to find good players on bad teams.
Using puck possession
Many fantasy hockey leagues still use plus/minus as a category. Players that can dominate puck possession tend to be good plus/minus players, while the reverse is true: the top-5 skaters in FenwickFor Close last year combined to be plus-130; the bottom-5 in FenwickFor Close last year combined to be minus-103. At the team-level, teams with good possession marks tend to suppress opposing shots. That is good news for goaltenders, as it can drive down their Goals Against Average (GAA). GAA is still popular in fantasy leagues, so selecting goalies from teams with good possession numbers is a shortcut to a good number.
PDO
PDO is a quick way to measure luck in hockey. Individual shooting percentages can vary wildly year to year, as can individual goalie save percentages. Numbers-wise, it can take 73 regular season NHL games to parse out luck from skill. A bounce off the glass, redirections off skates in front, there are a number of ways for teams and players to get “lucky.” PDO helps quantify that, and it is composed of two numbers:
On-Ice Shooting Percentage
On-Ice Shooting Percentage measures the rate at which a team scores with a given player is on the ice. This is something that players can establish over the course of their careers: Sidney Crosby hasn’t been below 10.6-percent on-ice shooting (not his personal rate, the rate at which Pittsburgh scores with Crosby on the ice at 5-on-5); Crosby’s teammate Craig Adams, meanwhile, has never been above 8-percent in a regular season. It’s intuitive – good players make others around them better – but this number quantifies that. A high on-ice shooting percentage can lead to a higher frequency of goals, thus more chances for points.
On-Ice Save Percentage
The second component to PDO is on-ice save percentage. This is the rate that the team (namely, goalies) save shots while a given player is on the ice. This has a big impact on plus/minus, a fantasy hockey roto category. Unlike on-ice shooting percentage, though, on-ice save percentage can vary wildly year to year. That’s one luck factor of fantasy hockey.
Together…
A player with an on-ice shooting percentage of 10-percent and on-ice save percentage of 90-percent will have a PDO of 100 (10+90). A mark of 100 is considered average. Too far above, a player should slow down. Too far below, a player should improve. It may take a season to happen, but it usually does. Better teams like Boston and Los Angeles will produce better PDOs for their players, while Edmonton and Carolina will not. This has to be taken into account.
PDO can be used at the player-level, and at the team-level. For fantasy hockey, using it at the player-level is helpful. Consider this: in the 2011-2012 season, Tyler Seguin had a 108.7 PDO on November 26th. He had 12 goals, 11 asssists, and a plus-18 rating through 21 games. Seguin played 81 games that year, thus:
- Seguin was on pace for 46 goals, 43 assists, and a plus-69 rating in 81 games.
- Seguin finished with 29 goals, 38 assists, and a plus-34 rating.
PDO said Seguin’s pace wasn’t sustainable, and it wasn’t. In that sense, PDO can be used to identify good buy low/sell high candidates in fantasy.
The Rest
Zone Starts
Zone starts just measure where a player starts a shift: more defensive zone starts can suppress offense, while more offensive zone starts can boost offense. That’s part of the context needed for CorsiFor and FenwickFor percentages. In fantasy hockey, it helps indicate players that are in positions to succeed, which is helpful in head-to-head and daily fantasy leagues.
QualComp
Quality of competition can be measured by different things, but usually with CorsiForRelative. That measures how good (or bad) the quality of the opponent a player faced. Shutdown defensemen and top forwards (Crosby, Bergeron, Kopitar, Toews) will have a high QualComp, meaning they play tough competition. Looking for players with low QualComps, maybe third line forwards or second pairing defensemen, can help plus/minus in deeper leagues.
There are other stats used (zone entries, for one) but these are the basics. Grasping these, and how they can help in fantasy, will give the savvy owner a leg up on the competition. Any small edge is worth exploring.
