Not-So-Advanced Stats: An Introduction To “Fancystats” For Wild Fans
If you read my articles here on the regular, you may notice I use a wide range of stats in my analysis. Some of these are often referred to as “advanced stats” or “fancystats” and have a lot of taboo and mystery around them. With the new season just around the corner, I figured now would be the best time to provide a beginner’s introduction to these statistics so that anyone curious can get to grips with them for the coming season. So, make sure you bookmark this in your browser for a quick reference when needed.
The first thing you need to remember is that “fancystats” are actually extremely simple. Tremendous amounts of hard work have been done already by the people who developed these metrics, the people who track and post the numbers on their websites and the people who break this stuff down and find new perspectives in their article writing. Thanks to their hard work, all that is left for the average fan to do is usually just read the number charts and articles, or, if you want to get a little bit deeper, do some basic adding and division to work stuff out for yourself. Simple as.
-Possession Metrics-
Corsi
The word “Corsi” often seems to put people off, but it is a very simple concept (“Corsi” is just the surname of the Buffalo Sabres goaltending coach who came up with the metric) which basically means “shot attempt differential”. Shot attempts are: goals, shots on goal, shots that miss the net, and shots that are blocked. It is usually represented as either a percentage, or as a plus/minus rating for either a team as a whole, or for individual players.
Unlike regular shot statistics, a player’s Corsi number refers to the amount of shot attempts by his team, minus the number of shot attempts by the other team while he is on the ice.
The theory among “stats heads” is that the difference between good teams and bad teams is 5v5 puck possession. Successful teams dominate in this respect while unsuccessful teams don’t. The best way to measure puck possession is by seeing whether teams direct more shots at the opposition’s net than the opposition direct at their net during a game, the theory being that you must have control of the puck to take a shot, and the more shots you direct at goal, the more goals you’re gonna score over the course of a season. So, Corsi is a proxy for puck possession.
This also works the same way for individual players. Some players drive puck possession, some get dominated. Theoretically, assembling a team of strong puck possession/Corsi players is likely to result it a high level of success.
For example, say Zach Parise is +7 in Corsi over the course of the game, that means 7 more shots were directed towards the opponents net than were directed towards the Wild net while he was on the ice.
-Here are the different types of Corsi numbers you will see:
-Corsi ON (or “Corsi/60”) = (Total Corsi +/-) x (60 mins) · (The player’s total ice time).
-Corsi OFF = Team’s overall corsi number while a player is off the ice, tabulated in a similar fashion to Corsi ON.
-Corsi REL = Indicates how solid a player’s possession statistics are relative to those of their team-mates. (Corsi ON – Corsi OFF = Corsi REL)
For more on Corsi:
–http://www.habseyesontheprize.com/2013/7/16/4376378/fancy-stat-summer-school-corsi
Fenwick
Fenwick is almost the same as Corsi, the only difference being it does not factor-in blocked shots. It is named after Matt Fenwick, the Calgary Flames blogger who invented it.
Whether you choose to use Fenwick or Corsi is up to you. Both metrics have a very high correlation with team success.
The stat which is considered the have the highest correlation with team success is “Fenwick Close” which is defined as the Fenwick number when the game is within one goal in the first or second period and tied in the third period and overtime. This number is usually reported as a percentage of Fenwick evens that a particular team gets. Thus an average team has a 50% close Fenwick. The reason for this being, when one team has a comfortable lead they often “take their foot off the gas” and play a more defensive system and allow their opponents to control the puck, while attempting to prevent high quality shooting opportunities. When a team is behind in a game they are “playing desperate” in an attempt to tie up the game. They play a much more offensive system where they control the puck and may be susceptible to high quality scoring chances on the counter-attack.
For example, the charts in this article on Eyes On The Prize last season show exactly how strong Fenwick Close teams tend to make the playoffs and the Stanley Cup more often than not. That article is a must-read.
For more on Fenwick:
–http://www.habseyesontheprize.com/2013/7/17/4520794/fancy-stat-summer-school-fenwick
–http://www.pensionplanpuppets.com/2012/7/25/3184137/intro-to-advanced-hockey-statistics-fenwick
–http://www.habseyesontheprize.com/2013/4/4/4178716/why-possession-matters-a-visual-guide-to-fenwick
-Useful Tools-
A problem I often find with these “stats-primer” articles is that they explain how the metrics work, but they don’t really show the reader how to adapt and use the stats in their day-to-day interaction with hockey.
So, with that in mind, I’m gonna preview a couple of websites that have really made interacting with stats very simple for me and that I hope will do the same for you.
Player Usage Charts
I would say I reference Player Usage Charts more frequently than any other type of statistic. They are a simple, visual, informative way to get some quick information players or a team. I use Player Usage Charts from two different sites; SomeKindOfNinja and Hockey Abstract.
In these charts, the Y-Axis is for “Quality Of Competition” and the X-Axis is “Percentage Of Shifts Starting In The Offensive Zone”.
On SomeKindOfNinja’s charts, the format for the charts is:
-Blue Bubble=Positive Corsi
-Red Bubble=Negative Corsi
-Bubble Size=Corsi Number
Hold your cursor over the bubble to get the exact numbers.
(Click to Enlarge)
You can change the settings on the site to show different numbers and different configurations of players. You can change the “minimum games played”, you can look at different seasons, you can choose one player and look at his entire career, you can look at Corsi Rel or Corsi On. There is so much data to be found on this website, but it is extremely easy to use.
On Hockey Abstract, the format for the charts is:
-Bubble size indicates “Total TOI” (this can be changed to “Average TOI/Game”.
-Bubble Colour indicates whether Corsi is positive or negative, and the Corsi number, which is colour-coded, with brown being the extreme negative, and deep blue being the extreme positive.
(Click To Enlarge)
Once again, you can change the settings to look at a bunch of different configurations. What I find useful about this site is that you can select certain players from a drop-down list, which is useful when comparing several players from different teams all at once.
What do these charts tell me?
The things you’re looking at in these charts are:
-How was a player deployed during the season.
-What kind of role did he play.
-How did he perform?
By the “Offensive Zone Start %” you can see if a coach generally used a player in goal-scoring situations, or was he used in a defensive role. By the “Quality Of Competition” you can see if a player faced the best players the other team had, or if he was “sheltered” and faced-off against the other team’s weaker players.
Players in the top-left of the graph played very tough minutes. Players in the bottom-right played very soft minutes. The Corsi On/Corsi Rel number tells you whether they succeeded or failed in their role. A player might have a slightly negative Corsi, but if he played tough minutes, then this needs to be taken into account in your evaluation of him.
Click on both of these sites and mess around with the settings until you feel like it starts makes sense. These are really simple, but informative tools and worth getting to grips with.
NiceTimeOnIce.com
This website is an absolute game-changer in terms of watching and analysing hockey. I make it a habit to always begin a game night by opening this in a tab on my computer. It gives you instant links to every stat or piece of information you could possibly want to access while before, during and after a game.
What it allows you to do is select a team and the game you are about to watch, and then it gives you a series of links, which are divided between ‘Season Links’ and ‘Game Links’.
As you can see, there are a whole bunch of different areas of information you can access, all neatly stored in one place. It’s got links to all the regular and advanced stats sites you could want.
All this is very useful, but the real game-changer for me is the “TimeOnIce” section under ‘Game Links’. You can access shift charts, head-to-head data to see who players are getting matched-up against, exact player-by-player zone start stats and, most importantly, detailed Corsi and Fenwick numbers.
I highly recommend, before every game, open the “Corsi/Fenwick” chart in a new tab on your computer (if you have it near you) and refresh the page whenever there’s a stoppage in play. The chart automatically updates and gives you up-to-the-minute shot attempt numbers that are easy to understand and extremely helpful in analysing what is happening on the ice.
TimeonIce.com Corsi/Fenwick chart from the Game 3 of the playoffs, the Wild’s OT win vs Chicago:
The format of the chart is:
“Game No. – Team Name – Player No./Pos/Name – Goals For While On Ice – Goals Against While On Ice – Saved Shots For While On Ice – Saved Shots Against While On Ice – Missed Shots For While On Ice – Missed Shots Against While On Ice – Player’s Fenwick No. (A total of all the goals, missed and saved shots for and against while that player has been on the ice) – Blocked Shots For While On Ice – Blocked Shots Against While On Ice – Corsi No. (Player’s Fenwick, plus Blocked Shots For, minus Blocked Shots Against).”
Pretty simple, right?
From the chart above, you can see that Jared Spurgeon, Marco Scandella and Jason Zucker performed very well and managed to drive possession all through the game. It’s no coincidence that Zucker scored the game-winner in OT that night.
Also, you can see that Brodin and Suter had rough nights (if you remember, they had just come off crazy minutes in Games 1 and 2 in Chicago) by their huge Corsi and Fenwick deficits.
I only screen-capped the top half of the page, so if you scroll down, you also get the Blackhawks numbers.
I highly recommend that you use all the features on NiceTimeOnIce, but if you only use one, make sure it’s the “Corsi/Fenwick” chart. The numbers in this chart, combined with the “eye-test” gives you the ultimate perspective of who is driving a game and who is struggling.
Try it right now and get familiar with it for the new season!
-PDO-
First off, don’t be scared by the name. It sounds fancy, but it was just the name of a blog commenter who came-up with the concept. PDO, if you want to break it right down, measures “luck”.
According to Eyes On The Prize:
“PDO is a statistic that’s used to determine whether a player or team’s performance over a set sample is sustainable. The way that this is accomplished is by taking two relatively stable statistics, even strength on-ice shooting percentage and even strength on-ice save percentage, and adding them together.”
It can be used at a team level or for individual players. It generally works better for analysing a team, as there are more variables and smaller sample sizes for individuals.
The average PDO in the NHL is 1000, so a team/player with a PDO lower then 1000 can be expected to improve their/his EVSV% and EVSH% over time and a team/player with a PDO higher then 1000 can expect their/his EVSV% and EVSH% to get worse over time (this is a concept called progression/regression to the mean, depending on what side of “even” you’re on). The use of the number 1000 can be a bit confusing as we usually see SV% as a three number decimal and SH% as a three digit percentage. For PDO, we just use a three digit decimal for both.
The best example of PDO at work that I can think of is, in 11-12, during the first half of the season, the Wild (with a PDO far North of 1000) shot to the top of the Western Conference, while the LA Kings (with a PDO far South of 1000) languished at the bottom. Then the Wild regressed towards the mean, while the Kings progressed towards it, and, lo and behold, the Kings made the playoffs and won the Stanley Cup, while the Wild plummeted to the nether-regions of the West.
Here’s a good quote on PDO from Fear The Fin last August:
Even a team that manages a massive 1040 PDO over a 1500-shot sample should expect to see that total regress to the mean by about 90% over the rest of their schedule, meaning they’d most likely post a 1004 PDO over their next 1500 shots. Similarly, a team with a ridiculously low 960 PDO through 1500 shots would likely be a 996 PDO team over the remainder of their season. These are extreme examples to be sure, as teams rarely post PDOs that high or low over any significant portion of the year, but they help to illustrate that regardless of how confident you are in a team’s ability to sustain a PDO substantially higher or lower than 1000, it’s usually a bad idea to bet on it. With the level of cap-induced parity in the NHL, the spread in shooting and saving talent between teams is small and vastly overstated by what we observe in the short run. Some teams with legitimately great (Boston, Vancouver) or legitimately awful (Columbus, Toronto) goaltending can be expected to sustain slightly above- or below-average PDOs but the majority of teams, as seen above, regress sharply to the mean.
PDO isn’t always going to work perfectly, but its main job is to highlight outliers, teams/players who are at the extreme end of the scale in having either a very high or very low PDO, and show that they won’t last.
Cam Charron ranks the PDO numbers for each NHL team each month on Nation Network, so keep an eye on that and, around December, see which teams are riding a high PDO, and which ones are riding a low number and you can probably figure out how things will go for them in the 2nd half of the season.
For more on PDO:
–http://blogs.thescore.com/nhl/2013/01/21/pdo-explained/
–http://www.habseyesontheprize.com/2013/7/29/4566716/fancy-stat-summer-school-pdo
-WOWY-
WOWY (With Or Without You) stats are basically used to show how players perform with and without each other, to figure-out if one is acting as a crutch for the other or vice versa. WOWY tries to separate out the effects of playing with one particular teammate on the Corsi/Fenwick numbers of both players. They are tracked at Stats.HockeyAnalysis. To view a player’s WOWY numbers, select “Players” from the tab at the top of the site, choose a player, then you can choose a season, and either “5v5”, “5v5 ZS Adj” or “5v5 Close ZS Adj”. Each of these will provide his WOWY numbers in a different situation.
5v5 = All even-strength play.
5v5 ZS Adj = ZS Adj (Zone-start adjusted) is something that is done by the guy that runs Stats.HockeyAnlysis.com. He says “I adjust for zone starts by ignoring the first 10 seconds after a face off in either the offensive or defensive zone. I do this because it has been shown by both myself and others that the benefit of a zone start is almost completely negated after 10 seconds of play.”
5v5 Close ZS Adj = This is the same as previous, only now it filters the data further to only include situations when the score is considered “Close†(the game is tied or a one-goal lead in the first two periods, and tied in the third period). People have analyzed this and noticed that in games where the scoreline favours one team, the trailing team usually tends to start overpowering the possession metrics; because they take shots from just about everywhere to press for a tying goal.
The last category would be considered the most representative of true possession value, with the qualifier that by adding the most filters of data it’s also the smallest category and could be subject to noise from the small sample size.
Here’s an example of a WOWY page from Stats.HockeyAnalysis.com. It is Tom Gilbert -> 2012-2013 -> 5v5 Close ZS Adj:
(Click To Enlarge)
The players listed under Gilbert are organised by the amount of time they spent on the ice with him. The first non-goaltender is Clayton Stoner, who was Gilbert’s most common defensive partner last year (207:53 minutes together). The chart shows that Gilbert and Stoner’s Corsi For (CF) % together was 42.6. It shows that Gilbert’s when he was away from Stoner (136:05 minutes) was 53.6, and Stoner’s (195:35 minutes) when he was away from Gilbert was 47.9.
So both players did better away from each other, but Gilbert significantly more so. These sample sizes are a bit too small to draw any significant conclusions, due to the lockout-shortened season.
Something that needs to be considered when looking at WOWY numbers is the possibility that players will be taking on vastly different roles when not together. For example, two 3rd line forwards could spend most of the year together, apart from 10 games where 1 moved to the top line and the other to the 4th. They are going to be facing different competition and be playing with players of highly differing skill. This is why context is important when examining WOWY numbers.
A good example of WOWY numbers being put in to action is in this Cam Charron article, where he basically dispels the much-hyped myth that there is good chemistry between Phil Kessel and Tyler Bozak.
-Further Reading-
I’m relatively new to this stuff, but I feel like I’ve got a decent understanding now by reading the countless helpful articles located around the web. While I tried keep things fairly simple in this article, I may have missed something , so I highly recommend that you do some further reading around the topics and keep-up with the various blogs that regularly post about advanced stats.
Remember, an interest in advanced stats doesn’t mean that you’ll be reading spreadsheets instead of watching games. All these numbers are just a companion for what you see on the ice.
No one can follow everything that is happening on the ice at all times, not to mention no one has time to watch every other team’s games and analyse them. It’s simply impossible. These stats give a fairly clear picture of what is happening on the ice for every player and team over the course of a season. Not to mention they give fans another aspect of hockey to discuss when there are no games on, which, for me, is a good thing.
Here are some articles you should read, and blogs you should bookmark for future reading:
Articles:
–The Faker’s Guide To Advanced Stats In The NHL
–Factoring Regression Into Analysis
–Shooting Percentage Regression
Blogs:
If you have any further questions or suggestions, hit me up on Twitter, or drop me an e-mail at “gerrydevine2006@yahoo.co.uk”.