Unconventional Wisdom - Progressive Hockey

When Darryl Metcalf was hired by the Toronto Maple Leafs and his site ExtraSkater went offline, many people online were concerned, wondering what they are going to do now without their favorite resource. My response was that ExtraSkater only existed for like a year, so the obvious answer was to go back to what you did before it existed. However, the void left behind and the fan outrage it created have spawned some new websites that attempt to present some of the features that were lost. So we have one more resource to work with for looking at our own Pittsburgh Penguins this season.

On of the newest is Progressive Hockey, a site that provides the ability to view 5-on-5 data for many of the things we are used to. He has Goals, Assists, Points, Sh%, PDO, and ZS%, which are all pretty standard by now. However, we also see a few interesting features that are borrowed from and expanded upon that which came from some of our other favorite stat sites. In keeping with the popular ExtraSkater format, he present TOI% for tracking player usage, certainly one of the features that will be missed, although as of now he does not have QoT and QoC based on TOI% like ExtraSkater used.

He also borrows rather heavily from the work of David Johnson at HockeyAnalysis. They just recently added a WOWY page, a feature that made HockeyAnalysis popular even amongst detractors that disdain anything just based on the name of the author. And one very nice feature that I really hope to see implemented in HockeyAnalysis in the future is that they provide WOWY for FF% in addition to the usual GF% and CF%. He also gives us the ability to add in some other situations, all in all not very helpful, but it is nice to be able to see the numbers for PP and PK.

Of course it is not a carbon copy, we do see that Progressive Hockey doesn't have the numbers broken down into For and Against, rather he just provides us with the overall percentage. He also doesn't give the option to look at Close data or that adjusted by zone starts. However, those are features that we see when looking at the main page of data.

Progressive Hockey allows you to check a box to "Adjust for zone starts and score state," which is actually quite similar to using HockeyAnalysis' Close Zone-Start Adjusted data. However, unlike HockeyAnalysis the Close data only applies to the GF%, FF%, and CF% numbers, so we can't see how that changes the effect of the other statistics he provides.

He also provides relative numbers for GF%, FF%, and CF%, which likewise you can see as either standard 5-on-5 or using the "Adjust for zone starts and score state" option. Unlike the sites like BehindTheNet in which relative means the player's stats minus his teammates stats when he is not on the ice, Progressive Hockey uses both the player's teammates and his opponents performance, adjusted for how much time he spent with/against each of those players, in order to create a relative value. This may sound familiar, and in fact it should, because this is precisely how HockeyAnalysis creates his HARO/HARD/HART metrics.

Once again though, while it may be based off Johnson's work, it isn't just a carbon copy. ProgressiveHockey doesn't break it down into For and Against the way HockeyAnalysis does, instead just providing us with the total GF%/FF%/CF% Rel, which would be more or less the equivalent of Johnson's HART statistic. There also appears to be some slight differences, most noticeably that when looking at just the Penguins we see Stempniak and Goc last year come out as quite different when looking at HART vs Progressive Hockey's relative values. That suggests to me that his data is actually broken down by team, whereas HockeyAnalysis only allows us to see the entire season for players that get traded midway through.

So when we compare the data, we don't get a perfect comparison, but it is quite close. HART vs GF% Rel on the Penguins forwards last season gives us a .5480 R², and even better .7120 R² for FHART vs FF% Rel and .7600 for CHART vs CF% Rel. There may be other minor differences, but other than the obvious shifts with players that were acquired mid-season the numbers appear to follow rather closely.

We see a similar trend when comparing the HockeyAnalysis Close ZS data vs Progressive Hockey's "Adjust for zone starts and score state" numbers. They don't add up completely, and the goals based is way off base, but if we are looking at possession we see a .6580 R² when comparing the FF% and a .8090 R² when comparing the CF%. Again, some of the difference could indeed be that the numbers I grabbed were for jsut the Penguins last season, whereas HockeyAnalysis is unable to break it down like that.

I'm sure there are likely other differences in the way the numbers are calculated, I know that in order to adjust for zone starts, HockeyAnalysis removes the first 10 seconds immediately following any faceoff, so if Progressive Hockey uses a different method that could account for some of the minor differences. They may also be like some of the sites like BehindTheNet that include empty nets into their 5-on-5 calculations, whereas HockeyAnalysis and formerly ExtraSkater only used situations in which both teams had 5 skaters and a goalie on the ice.

Whatever the explanation is for the minor differences, it still stands that between WOWYs, relative GF%/FF%/CF% that relies on both QoC and QoT, and "Adjust for zone starts and score state" data, this new site does have quite an awful lot in common with HockeyAnalysis' numbers. Even the QoC and QoT, which you can be expressed as GF%, FF%, or CF% based, providing us with the same information we can get from looking at HockeyAnalysis' TM and Opp percentages.

But despite the similarities, Progressive Hockey does indeed provide us with some unique information. One new bit is that he provides us with Shot Quality percent. While it is certainly possible for one to acquire this information, for example using the Super Shot Search, it is unique to be able to find the numbers readily available in a database like this. In addition, he uses that Shot Quality data, combined with the other information, to create his own metric, Expected GF%. And that Expected Goals data can be used for QoT and QoC, he provides the relative value based on said QoC/QoT, and the values can be modified with the "Adjust for zone starts and score state" option.

So while the site isn't exactly novel, it does provide some interesting information that isn't necessarily readily available elsewhere, so it is definitely worth checking out. I will make sure to try to incorporate the information into future examinations into the Penguins performance.