Hockey Graphs

The Finals predictions, based on truculence, size, and experience. Up to this point, truculence and size have returned an accuracy of 5 for 14 series, and experience 6 for 14. In other words, none of these supposed virtues are matching a coin flip.

Two charts to convince you on the importance of things like possession and zone entries in the NHL...in 2000-01 and 2001-02, the NHL tracked time in the offensive zone, and then they stopped. You video gamers out there will rightly note that the EA games continue to track this stat, so don’t ask me why the NHL decided to stop. Anyway, I had the data from those two seasons of every game; by focusing on the team performances in 2 periods, rather than 3, we can remove the majority of score effects and see just how closely these two measures run together. What we actually find? Zone time has a tighter, more normal distribution. What does this mean? There are better and worse ways to use your time in the offensive zone, which is why there are so many people interested in entering the zone with possession, and controlling the puck while in the zone (and soon, we will have measures to demonstrate “useful” or “productive” possession).

Charting the distribution of Overall ratings in the EA NHL series from NHL Hockey ‘93 through NHL 05 (basically, the years I had data easily available). Though it doesn’t go all the way to the present, it charts the progression to the point where some people started to gripe that any slug could get a 70 overall rating. And they’re right.

As with the 1st and 2nd round, predicting the Conference Finals using measures of truculence, size, and experience. Unfortunately, those team “virtues” haven’t done so well this year: truculence is 4 of 12 series, while size and experience are 5 of 12. For those keeping track, SAP has matched a coin flip thus far (6 of 12).

From 1992-93, I created a selected group of top-line forwards that had played at least 750 games in that time span and recorded 0.859 points per game or greater. That bar was set by looking at top-line forwards from 1992-93 to the present, assessed by the quartile that recorded the highest shots and assists per game. Using this group of 48, I observed their proportion of full season games played (within this era, we had changes in the number of games in a season, thus the proportion), then also observed what proportion recorded a season at each age. As a result, we get a pretty decent couple of curves looking specifically at how elite forwards age.

These are charts for a piece called “The Best Defense is a Good Offense,” demonstrating how teams are increasingly choosing defensemen who can contribute offensively, across all pairings. The abbreviation “OSP” refers to the new metric, on-ice shooting proportion, or the the percentage of on-ice shots-for taken by the individual player.

Comparison of percentage of all league shots to overall league shooting percentage, each year from 2008-09 to present. Data taken from Greg Sinclair’s Super Shot Search.

One last request, a comparison of Oates and Trotz’s Washington Capitals at even-strenght, looking at shots-against. As with the previous GIF, the shots-against aren’t exhibiting much difference in the distribution of location, just a difference of volume.

By request, a comparison of the Adam Oates and Barry Trotz at even strength. The location distribution is really no different, just the sheer number of shot attempts.

By request, a comparison of the Adam Oates and Barry Trotz powerplays. Trotz’s seems a bit more dispersed, though some of it is due to sheer number. That said, you can clearly see that Trotz has Ovechkin shooting from his favorite spot.

Comparing shot location plots on the powerplay between the coaching careers of Boudreau, Ruff, and Vigneault (2008-09 to present). The plots are screen grabs from Greg Sinclair’s excellent Super Shot Search.

Well, of the brackets, truculence went 3 for 8, experience went 3 for 8, but size went 4 for 8...for those who are counting, that matches SAP. Here are the second round predictions. Because of the drastic overlap, I flipped the predictions for age (it was sucking anyway).

Aaron Ekblad’s evolving Total Player Chart in 2014-15, using 20-game moving averages of the measures (league average and high held static).

I’m reviving an old type of chart I put together a couple years ago, the Total Player Charts (TPC). Basically, it’s expressing ice time as a percentage of available ice time, only at the three main strengths. I’m tinkering with variations of it, and want to make it into something filter-able, so don’t be surprised to see some more of these in some shape or form.

Comparing top and bottom line NHLers (top and bottom 25% in TOI performances, 2005-06 to present) over time. It appears that NHL teams are making better choices and leaning on more defensive deployment among their bottom 6/bottom pairing players.

Since the 2004 lockout, the salary cap has systematically pressured teams to stop employing players they cannot trust with regular even-strength minutes. The result is an overall lift in the average game 5v5 TOI among NHLers.

By request, I’m adding the peak age estimation by shift length chart in a version that uses seconds.