Before we return to our regularly scheduled dissection of the 2021-22 Chicago Blackhawks season, we thought this would be an ideal time to offer a walkthrough of the statistics, charts, graphs and other pieces of information that frequent the site, taken from some of the best advanced hockey statistics sources on the internet.
And because we, the SCH staff know that you, the SCH reader, are already a wonderful, delightful, amicable, endearing, marvelous and outstanding group of people, we wanted to add “knowledgeable” to that list of dazzling character traits.
So let’s get into it, starting with what’s probably the most popular source for advanced hockey statistics available:
Natural Stat Trick
Frequent visitors of the site will recognize Natural Stat Trick as the source for the game charts below that appear in game recaps:
Along with that game-by-game information, Natural Stat Trick also offers a wealth of individual player and team statistical data. Here are some of the lesser-known stats track beyond, simply shots and goals:
Corsi events for or against (CF or CA): A “Corsi event” is more simply known as an attempted shot on goal, whether that shot is on net, gets blocked, or misses the net completely.
Fenwick events for or against (FF or FA) — The difference between Fenwick and Corsi is that Fenwick events exclude blocked shots. Only shots net on goal and shots that miss the net are counted as Fenwick.
Scoring chance for or against (SCF or SCA) — shot attempt inside the “home plate” area (see the figure below). Scoring chances are any shot attempts that are taken from areas of the ice where goals are more likely to be scored defined by War-on-Ice’s value assignment.
High danger scoring chance for or against (HDCF or HDCA) — shot attempt in the mid or low slot, as depicted by the red area in the figure above
All of the figures above can be expressed as whole numbers, as a percentage share of a data sample (CF%, HDCF%, etc.), or as a rate (CF/60, HDCF/60, etc.)
By counting up all of those data points, among others, the Natural Stat Trick data arrives at a statistic known as expected goals.
Expected goals for or against (xGF or xGA): This stat is based on shot attempts that have been weighted for shot quality by factoring in things such as shot type, shot location, shot angle, and whether a shot was a rebound or rush shot. This does not tell if a shot actually resulted in a goal but rather the probability of scoring a goal based on the factors listed above. Expected goals models have been even better at predicting future performance than shot attempts alone.
PDO — A combination of on-ice save percentage and shooting percentage that shows a team’s puck luck or how fortunate a team’s bounces have been. Over larger samples, this number always trends towards 100, which means any outliers can be perceived as statistical anomalies due for potential progression or regression to the mean.
Evolving Hockey’s RAPM charts
Understanding those numbers above will be handy when these charts are involved, as Evolving-Hockey’s “Regularized Adjusted Plus-Minus” (RAPM) charts use those stats for a visual expression of how good — or bad — players are in multiple categories.
Here’s what 2009-10 Jonathan Toews looked like using these models. As the headers at the top indicated by “EV TOI” and “PP TOI,” the left chart is for even-strength time and the right chart is for power-play time.
The stat categories are listed along the X-axis, while the “Z-score” is the Y-axis. The “Z-Score” is based on the standard deviation of each statistic. To keep this from turning into math class, just know that a “Z-score” of 0 is league average.
From a starting point of 0, the numbers go something like this:
- From 0 to 1: Call this “slightly above league average.” About one-third (34.1%) of the league’s players are in this category for that statistics.
- From 1 to 2: Players in this range are approaching all-stars levels in this category. A touch below one-seventh (13.6%) of the NHL occupies this space.
- 2 or above: This is elite territory, where barely over 2 percent of the league’s players reside
The negative portion of the ledger follows a similar trend in the opposite direction:
- From 0 to minus-1: Call this “slightly below league average.” Again, roughly one-third (the same 34.1% as above) of NHLers are in this range.
- From minus-1 to minus-2: Possible descriptions for this range are “pretty bad” or “definitely not good”
- Minus-2 or below: Among the league’s worst
So, going back to the Toews charts above, you’ll see Toews was up near 3 for every power-play statistics. He was above league average in all five categories and at or near elite levels in three of them. Toews was an all-star that season. The Blackhawks team that season was ... pretty good.
Want to see what the other end of that spectrum looks like?
Sorry in advance:
During the ‘21-22 regular season, Riley Stillman had the third-worst CA/60 in the league at 67.01, which is why he hits minus-3 in that category. He was also well below league average in CF/60 and xGA/60.
JFresh Hockey’s player cards
In the last few years, the player cards seen below have become staples of the advanced hockey statistics world, as created by former SCH podcast guest JFresh Hockey.
The data in these player cards is compiled by Patrick Bacon (aka @TopDownHockey on Twitter, full website available here). All of the numbers are expressed as percentiles. For example, in the card below, 2010 Marian Hossa was in the 96th percentile for even-strength offense, which means that the data they’ve compiled said that Marian Hossa was better than 96 percent of the NHL at even-strength offense. Shocking, I know.
A few other stats which may be unfamiliar:
- A1/60: Primary assists per 60 minutes
- Quality of competition (QoC): A statistic which estimates the level of opponents that were on the ice at the same time as that player
- Quality of teammates (QoT): relative to the rest of the league in terms how good their most frequent linemates or D partners were
The other stat referenced on these cards is Wins Above Replacement, or WAR. It’s a metric that attempts to express how much value (or wins) each player contributed to their team’s success, expressed as a single number. This single number is comprised of multiple components that are ratings for each area of play within a given sport. The scope of this data goes well beyond this article. For those interested, here’s the model that Top Down Hockey uses. Want more? The EvolvingHockey model has three parts: Part 1, Part 2, Part 3
The player cards from JFresh have recently been updated to include microstats which are tracked by Corey Sznajder (aka @ShutdownLine on Twitter), who was recently profiled by The Athletic.
Microstats are not some elaborate calculation: it’s merely the more detailed aspects of the game that were not as easily tracked until technology evolved. Through Sznajder’s work (found at this website), we can now quantify how successful a hockey player is at aspects of the game such as:
- skating the puck into the offensive zone
- passing the puck out of the defensive zone
- passing into the “high-danger” areas of the ice
- recovering the puck after it’s dumped into the offensive zone
There are a ton of categories tracked by Sznajder, although he doesn’t track every game for every team. As indicated on the bottom right corner of Alex DeBrincat’s card below, Sznajder’s data covers 23 of DeBrincat’s games from the prior season (Sznajder updates that number, along with the rest, as more games are tracked):
Side note: DeBrincat’s WAR of 4.99 was No. 13 in the NHL, which gives an idea of just how good of an all-around player DeBrincat has become in his career — as discussed last week.
There are also team-based cards available, although the numbers on the card are not expressed as a percentile, they’re expressed as the team’s ranking in the league in that category:
A lot of red on that card, which would line up pretty well with a team that finished 27th overall in a 32-team league.
A final word
None of the above resources are presented as definitive truth regarding whether or not a player is good: they’re better interpreted as pieces of the overall puzzle that is player analysis.
But the benefits of all this data is that it comes from unbiased, objective sources. Put all of it together and it should paint a pretty good picture of what’s happening with a player individually and a team collectively. It’s not without its faults. Just last month, there was a Twitter discussion regarding Patrick Kane, who’s always seemed to outperform some of the statistical models:
There are two things I can say with confidence: Patrick Kane is way better offensively than WAR can capture, and Patrick Kane is just as bad defensively as WAR says.— JFresh (@JFreshHockey) April 17, 2022
But situations like that are more of the exception than the rule. And this data can help explain things that can go unnoticed by the bests of eye tests: like a defenseman who consistently needs his goalie to bail out his defensive miscues or an offensive player who’s generating countless scoring chances without being rewarded by goals and assists.
We’re not saying this data should replace the eye test. But if the numbers above are telling a different story than your own eye test is, perhaps one of those sources requires a deeper level of evaluation. We can’t help with the eye test portion of it, but hopefully this offers some assistance for the other part of this hockey analysis equation.