January 19 2013 09:24AM
by Kareemadel, via Wikimedia Commons
Keeping in mind Derek's solid post from yesterday, we at NHL Numbers nevertheless felt a need to put together a NHL Power Rankings list for the coming season (in fact, the first "full" NHL season during which this site has functioned). Including me, Jonathan Willis, Derek Zona, Cam Charron, Kent Wilson, and Chase W., each contributor was asked to make their rankings and include a brief sentence or two summary on their ranking. Jonathan used those rankings to create a composite ranking, and now I weave in the prose. To save some space, the commenters I'm using below include one whose ranking was a bit of an outlier from the others and one who provides a pretty good summation. Let's get to them, starting with a real hack pick if I ever saw one...
January 07 2013 08:56AM
In previous posts, I have stressed the value of understanding NHLers' time on-ice as a percentage of available time (Time On-Ice%, or TOI%) rather than as a lump sum. I then found that TOI% at 5v5, 5v4, and 4v5 provide data points for a pretty nifty radar plot, what I've called a "Total Player Chart," and used those TPCs to depict player value.
Since then, I've also expanded Total Player Charts to include "Team Total Player Charts," or TTPCs, which demonstrate some other nifty stuff for looking at individual season depth charts for teams (such as the Vancouver Canucks in 2011-12), and looking at the evolution of player depth over time for teams (such as Vancouver from 2007-08 through 2011-12).*
I promised to come back and include some analysis of 5v4% and 4v5%, and for now there was one stone I felt was still left unturned: TOI% including all of 5v5, 5v4, and 4v5. By looking at individual TOI% performances over the last five seasons, and in particular those that are leading by this metric, you are undoubtedly looking at real MVPs, the players with the largest amount of their team's minutes.
An overwhelming majority of the time, they are also logging the most crucial minutes. Now, I've heard some whispers of apprehension about trusting that coaches will always put the best player out there. They won't always, but it won't be for lack of trying. Traditionalists who refute statheads by saying we should defer to someone who "watches" or "plays" the game (I've done both, probably more than them, but I digress) should have no problem with deferring to a coaching assessment. And if they don't, well who the hell are they, anyway? Stats folks have a right to be more skeptical, but are also beholden to the fact that coaches making the wrong decision will be a small part of the sample.
Suffice to say, it's no sure measure, but it's one of the best.
*Incidentally, I draw attention to a critical distinction between the TTPCs in individual seasons versus multiple seasons as I present them. Explanation here.
December 10 2012 08:48AM
I've always felt pretty strongly about Tom Awad's now-two-year-old assertion that, absent something better, even-strength ice time is a pretty good proxy for overall player value.
His numbers bore that out. However, I suggested that we could better refine and increase predictability of even-strength ice time by expressing it as a percentage of the team's even-strength time in games the player played (a wordy way of describing 5v5%). Though it will probably be a post in-and-of-itself, 5v4% and 4v5% are both testing to be similarly positive refinements on those TOI metrics.
Rather than handle that today, I wanted to unveil a little idea that had been bouncing around my head for a while. I'll be damned if I can recall the post from Beyond the Box Score, the excellent baseball statistics web site, but they had a season preview a while back where they used radar plots of a number of metrics like OBP, Fangraphs' Baserunner rating, UZR/150 (something to that effect), and for each player on a given team their talent in those metrics filled a certain portion of the radar plot. The more you filled the radar plot, the better you were. Author Note: Having heard back from BtB's Justin Bopp, the creator of the Diamondview Composite Player Evaluation that inspired the Total Player Charts, I can put my agitation and accreditation worries to rest).
There's certainly room for doing this when the statistical hockey gods agree on a metric already (spoiler alert: they never will) - till that time I'm content to build a radar plot of three metrics, 5v5%, 5v4%, and 4v5%. In part, this is to represent player value, as well as where that value gets allocated (just even-strength, or powerplay and penalty kill as well). I'll call them Total Player Charts, or TPCs (which, it turns out, is also an acronym for all sorts of important ish)...you can do a lot of neat stuff with 'em.
Let's have a look.
December 03 2012 12:57PM
Dealing with sum data should inevitably make a fancy stats person a bit uneasy; sums perpetually have a wealth of additional factors a person needs to know before they try to conclude anything. For instance, let's say your team's prospect is Lukas Sutter, and you want to know what the hell is going on with his point totals (aka, sum data). Well, Sutter's ice time is suffering right now, and the penalties he's taking aren't helping him get out of the doghouse. Hence, low points. Let's also say I'm predicting that, had the season happened, Evander Kane was going to score 50 goals...would he be taking enough shots? Will he play enough games? Will he receive enough ice time? Et cetera, et cetera...
Well, one of my previous posts pointed out two things we know about team 5v5 time on-ice: a) it can be volatile and independent of team talent, and b) it has gradually increased over the years. This throws a little bit of a wrench into using raw 5v5 TOI/60 to look at player quality, although that wrench can pretty easily removed.
My thinking is that you could control for those two elements of volatility by taking a player's 5v5 TOI and divide it by the team total 5v5 TOI in the games the player played. Whenever you do something like that, you want to make sure that you're actually improving predictability and either easing access to or learning new information, otherwise there's no point in creating the new metric. So, how did creating 5v5% work for me?
November 26 2012 07:51AM
Farting around with % of Attempted Shots (%AttSh) and coach data is fun, but the real bedrocks of fantasy prediction have to be line data and the focus of today's post, playing time. The line data is useful for looking at playing time if you anticipate a drastic change for a forward, while the playing time data can be important in telling us something about how predictive it can be, and whether there are any trends within the data that we need to pay attention to. It's one of the dirtier secrets of boxcar statistics (goals, assists, points), that they are frequently driven by playing time as much as they are driven by skill. Pierre Parenteau could've lived in AHL obscurity forever, but he received the ice-time opportunity that made him a great fantasy add the last couple of seasons.
Let's have a look at what our ice-time data can tell us about the future...