A Note on Compiling Assists

In the last post, I briefly mentioned the idea of compiling assists for league-seasons which predate the concept of the assist. This is something of a controversial idea, inasmuch as there can be controversy in something as mundane as compiling historical hockey stats.

When compiling historical stats (which I've done an awful lot of over the years, with countless hours at microfilm machines and online, combing through 100-year-old newspapers), I always compile assists. I get them from the written game reports. Often a report will say something like "X scored on a pass from Y", or "X scored from centre after Y had carried the puck down the ice", or "Y shot, which was stopped neatly by the goalkeeper, but X followed up and put it in". In all of these cases I would assign an assist to player Y.

Now, there are some perfectly valid criticisms of doing this. Not every game was covered in the same level of detail, and indeed not every goal was. So you'll never be sure you captured every proto-assist. Moreover, the detail in the coverage often varied greatly from team to team and paper to paper, which results in some teams having a relatively high ratio of assists to goals, and others with almost none. So if you try to do something like adjusted points, and apply the same goal-to-assist ratio across the league, you end up with some players having a big advantage simply based on which team they played for.

These criticisms are valid, but only if you intend the use the stats as a historical record. That's not what I use the stats for. I want to draw meaning from the numbers, not just present the numbers as-is and let them speak for themselves, which they can't really do anyway. Stats are all about context, and that's the real impetus behind Point Allocation: to dig into the context in which numbers were compiled, and present them accordingly.

Point Allocation has a very large advantage over adjusted scoring in this regard, because offensive points are allocated on team basis, not on a league basis. As such, by adding unofficial assists to the official goal totals, all you're doing is rearranging the offensive points on a particular team amongst its players; a team is not going to suddenly have more offensive value because you add in assists. So even if your assist totals are incomplete, it's not a big deal. Remember, Point Allocation is only ever intended to be a rough estimate, but it we exclude some of the information we have (partial assist records, essentially), we needlessly reduce the accuracy of the estimate. We need to make sure the playmakers of yesteryear get their due, as much as we can.

Official records be damned. I want to know what the numbers mean.