Welcome to our Toronto Maple Leafs‘ Advanced Statistics page. This page is new, and to start we will be adding some interactive charts or “viz” and a glossary. We will be adding more content here over time.
Here is what you can find right now:
1. Maple Leafs Report – a summary of each Leaf skater’s key numbers, both individual and on-ice at 5v5.
2. Rolling Trends – examine the current trends at 5v5 for the Leafs and other teams on over 25 different statistics.
3. GameShots – explore shots on a rink map and see how effective they were based on the stats provided
4. Glossary – a quick summary of common hockey statistics and terminology
1. Maple Leafs Report
This viz lets you rank the Leafs on a hockey statistic by position (all, forwards, or defensemen) and then pick a player to spotlight — to be shown on the right side. Note that most statistics are individual and cover all game strengths, but four are 5v5, adjusted for score and venue, and report how the five-man unit does when a selected player is on the ice.
The additional tabs at the top of the viz contain ‘quadrant’ charts for 5v5 on-ice statistics. These let you see a plot of the Leafs players by expected goals, scoring chances, and shot rates for and against.
Players in the upper right have per-60 rates better than the team average. Players in the bottom left have per-60 rates worse than the team average. Note that the vertical axis is reversed, so that lower rates against are at the top.
Maple Leafs Report usually updates the day after Leaf games.
2. Rolling Trends
Rolling Trends lets you select from a wide variety of 5v5 team statistics to see how the Leafs are trending (it lets you check on other teams as well). To smooth out some of the noise in game-to-game data, a 10-game moving average is used. This averages the last 10 games — that’s the trend you see plotted.
You can pick a statistic, a division, and then click a team to highlight.
There is a date range slider to let you zoom in on a certain range of game dates. Clicking or hovering on a trend point will also pop up the statistic as of that game. If you would rather select a set of teams and focus on them, set the division to “All” then use the “Pick Teams” menu at the bottom to check as many teams as you like. The chart will then include them.
If you would rather look at individual games without a moving average, use the second tab at the top of the viz.
GameShots is updated each Monday and lets you explore the location and danger of shots. You can look at a team, a shooter, a goalie, or a defending team. It is preset to Toronto but you can look at any team or player. The instructions are on the left and the Viewing options are at the top. Viewing options include heat maps, shot maps and maps showing the different shot types.
All data comes from MoneyPuck. Note that I chose to eliminate empty net goals and shots from outside the blue line. I consider them as noise that has little to do with the art of shooting.
MoneyPuck explains their expected goal model here: http://moneypuck.com/about.htm.
xG models are at the forefront of modern hockey analytics, they allow us to examine teams, players, and goalies more closely, considering the danger associated with each shot. The MoneyPuck model is state of the art and uses a machine learning approach called ‘Gradient Boosting’. This allows it to “learn” based on the patterns of over one million unblocked shots.
“Rate Statistics” – usually expressed on per 60 minutes basis, e.g. Points per 60 (P/TOI*60). These help aid comparisons between players but can be unstable in small sample sizes, e.g. under 100 minutes.
“Game Strength” – Hockey results vary greatly depending on whether teams are playing at full strength (5 skaters + 1 goalie also known as “5v5”) , or are playing at a special strength, such as a power play (5v4, 5v3, 4v3) or penalty kill (4v5,3v5, 3v4).
Overtime (3v3) and 4v4 are are included with 5v5 when referring to “even strength”. Many reports also include pulling the goalie as even strength (6vs5). For this reason, you should be careful in understanding which game strength is being used.
“On-Ice Statistics” – refers to statistics that are tied to the results for all skaters on the ice with a given player, usually at 5v5. On-ice CF% (shot share) and on-ice shooting percentage are examples.
“Adjusted” – this most often refers to the adjustment of on-ice statistics such as CF% for score and venue effects. (Individual statistics like goals are not adjusted.) Teams who are leading or who play on the road tend to take fewer shots. These are adjusted upwards to try and account for these effects. In the same way, teams who are trailing or who play at home will see their statistics adjusted down.
“Expected Goals” – (xG) each unblocked shot has its location recorded by the NHL. That location plus other information such as shot angle, rebound, time between shots, type of shot, and game strength go into a statistical or deep learning model, which then predicts the probability of a typical shooter scoring on a typical goalie in that situation. These models are built on years of historical data involving over a million shots. The xG of a shot is the probability of scoring and ranges in value from 0 to 1.
Individual Player Statistics
Assists (primary) – (A1) the 1st assist on a goal. More predictive of playmaking than secondary assists.
Assists (secondary) – (A2) the 2nd assist on a goal. More important with defencemen.
Assists – (A) Total Assists = primary assists + secondary assists
Face Off % – (FO%) face off wins/total face offs taken x 100.
Goals – (G) pucks that entirely cross the goal line of the net during legal play.
Goals/60 – (G/60)goals per 60 minutes at the strength stated (All, 5v5, PP, etc)
GP – Games Played
High Danger Scoring Chances – (iHDCF) Any shot attempts with a scoring chance score of three or more. See below for the scoring chance definition. This is the most common definition. Some hockey sites such as MoneyPuck use different definitions, eg. shots with an xG > .2. The main idea is to capture the most dangerous chances which are generally in the “home plate” area in front of the net.
Individual Expected Goals – (ixG). This is the expected goals for an individual player in a game or season. See more about Expected Goals in the General section above.
Points – (P) Goals + primary assists + secondary assists
Points/60 – (P/60) points per 60 minutes.
Penalty Differential – (P+/-) the difference between penalties drawn and penalties taken
Save Percentage – (Sv%) Saves divided by Shots on Goal x 100
Shooting Percentage – (iSh%) Goals divided by shots on goal(SOG) x 100
Shots – also known as “shot attempts” or “Corsi” (iCF). Includes blocked, misses, saves, and goals. Not to be confused with shots on goal which is often abbreviated to “shots” by some.
Scoring Chance – (iSCF) Scoring chances are unblocked shots that meet a certain criteria based on location and other factors. They include the more dangerous “High Danger” category.
Here is the most common scoring chance definition (from Natural Stat Trick Glossary/War-on-Ice):
TOI – Time on Ice, or minutes played.
TOI/GP – the average minutes a player plays per game. It varies depending on how a player is used at even strength and on special teams.
CF – “Corsi For”. Better known as on-ice Shot Attempts For or just Shot Attempts.
CA – “Corsi Against”. Better known as Shot Attempts Against.
CF% – “Corsi For” percentage = CF/(CF+CA) x 100. Known as shot attempt share or shot attempt differential. This measures the percentage of total shot attempts that are made by a player’s team, usually at 5v5 or even-strength. When used as a player statistic, it summarizes the CF% when a player is on the ice.
CF% Above 50% = CF% – 50. This highlights players who are above or below 50%. Players above 50% are helping teammates drive play more than their opponents.
SCA – Scoring Chances Against
SCF – Scoring Chances For
SCF% Above 50% – SCF/(SCF+SCA) x 100 – 50. Scoring Chance share above 50%. Players above 50% are on the ice for more scoring chances than their opponent. See Scoring Chance definition in the individual section above.
xGF – Expected Goals For. Includes the expected goals for all teammates on the ice with the player. See more on expected goals in the General section above.
XGA – Expected Goals Against. Includes the expected goals for all teammates on the ice with the player.
XGF% – Expected Goal Share = xGF/(XGF+xGA) x 100. Includes all players on the ice with the player. Very similar to the CF% explanation above except that each shot’s xG is used instead of a “1” weight for each shot. Players with an xGF% above 50% are on the ice for a greater share of expected goals than their opponents.
xGF% Above 50% = xGF/(xGF+xGA) x 100 – 50. Expected Goal share above 50%. Players above 50% are on the ice for more expected goals than their opponent.