THN Analytics: comparing NHL greats with new historical data

(Photo by Bruce Bennett/Bruce Bennett)
(Photo by Bruce Bennett/Bruce Bennett)

For many decades, the comparison of great NHLers has primarily been in the hands of storytellers, a case of trophies, or a rudimentary set of statistics. The concerted efforts of a community of researchers, The Hockey Summary Project, has helped open us up to much finer details of these players’ careers, and granted us the opportunity to create new points of comparison.

One big area of comparison, that transcends things like era scoring effects and rink counting bias, is looking at the proportions of shots and assists taken in the games the player participated (% of Team Shots and % of Team Assists – or %TSh and %TA, respectively). Quantifying contribution this way gets us a lot closer to how important a player can be to a team’s possession performance. Why shots and assists rather than goals? For one, the idea is to better capture how a player contributed to possessing the puck; going beyond goals (which are still counted as shots) gives us more evidence the player was an important part of the team’s puck movement.

Assists also tend to get short shrift, primarily for some of the dubious circumstances that an assist is awarded. The irony is that a) fluke goals also occur, and b) assists actually speak better to what I’m trying to capture. Some assists result from luck, but all assists occur because a player was around the puck – and in a different capacity than shooting.

As I mentioned above, The Hockey Summary Project (HSP) allows us the opportunity to calculate these kinds of contribution data. A massive compilation of game-by-game box scores dating back through the 1952-53 NHL season (and an added bonus, 1926-27), HSP’s site will absolutely floor any NHL history buffs. It takes a bit of work, but by compiling the data from the HSP box scores a researcher like myself can scratch the surface of player contributions, and even find some fresh ideas for comparing players.

For example, I wanted to start by looking at a group of known great players, both forwards and defensemen: Mike Bossy, Wayne Gretzky, Mario Lemieux, and Bobby Orr, Ray Bourque, and Paul Coffey. I wanted to compare their shooting (%TSh) and passing (%TA) contributions year-by-year to their age, and I figured to keep the groups separate because forwards would overtake the defensemen. Or so I thought:

Comparing NHL Greats with New Historical Data Fig 1

Comparing NHL Greats with New Historical Data Fig 2 Comparing NHL Greats with New Historical Data Fig 3 Comparing NHL Greats with New Historical Data Fig 4

In fact, the bigger difference was in assists, not shots. Obviously, the forwards scored more goals; their shots were from in closer, for most part. But it’s likely that shots from the point helped generate opportunities down low as well. The charts are also interesting in how they depict player styles: Bossy’s sniper credentials are legit, but that talent is humbled when placed alongside Gretzky’s clever opportunistic offense and Lemieux’s all-around dominance. And then there’s Orr, who is simply transcendent.

It’s also interesting to see the growth curves of the players; the selected defenders do seem to hang on to their shooting peak a bit later and longer, while the forwards seem to hold steady in assists as their instincts age far more gradually than their athleticism. Having already determined the defense and forward groups seem closer than I thought they would be, I figured it would be useful to put them together in a couple charts:

Comparing NHL Greats with New Historical Data Fig 5 Comparing NHL Greats with New Historical Data Fig 6

I still find it interesting that these players are so close in shooting contributions, but I’m a bit intrigued by the assist share. One additional thing we can do with HSP data is breakdown first and second assists; is it possible that a defenseman receives fewer first assists because of their relative position, and therefore second assists might be a better “equalizer?” Think of it this way: in addition to assists from moving the puck at the top of the zone, defenders also get a fair number of assists setting up the rush. While these plays are important, they’re more likely to place defenders as the second assist, or even the uncredited third assist.

Comparing NHL Greats with New Historical Data Fig 7

Well, that seems fairly convincing…then, in theory, second assists should pull these players closer to even with one another if looking at something like first versus second assists per game.

Comparing NHL Greats with New Historical Data Fig 8

I think what we might be capturing here is closer to the essence of contribution. If first assists can be fluky, second assists can be even more so…yet the rule still holds that second assists will result because the player is around the puck. Might the building blocks for historical player contribution include a combination of shooting and second assist contributions? We can save that for a later post.

Often considered a bit of an statistical “Dark Age,” groups like HSP are giving us the tools necessary to shed some light on the NHL pre-1997 (1997-98 being the first year the NHL began sharing more in-depth box score data). They even give us some grounds for comparing the game across eras. We’re really just scratching the surface above; the opportunity is ripe for many other ways to dig into the past, and ultimately connect the past to the present.