One of the things this blog can count among its real accomplishments: Some very smart people have passed this way over the years. Some have gone on to work for NHL teams, others embarked on media careers and still others decided to become successful.
I tend to think of the folks who come and visit in terms of their time of arrival. Some of you go back to my HF days and others commented on oilfans before the blogs explosion. I have enjoyed all comments over the years, from what now count 38, 187 different user names (5,200 are Woodguy trying to sign in on his phone though).
I have enjoyed visiting with all of you, and always wonder where you’ll end up. Not long ago, a few giant minds formed a hive and their work began to take shape as WoodMoney. Many of the people involved have commented here over the years (Ganesh “GMoney” Murdeshwar, Zsolt Munoz, Derek Blasutti, Darcy McLeod) have been working on a new website that may revolutionize publicly available analytics.
Put another way: This is a big damned deal. The site is here, and is called Puck IQ. I asked Woodguy over for a drink (I drank rye, he got water from the tap and cheese for the hand) and perhaps the conversation can serve as a tutorial. From here until the end, questions are me, answers are Darcy McLeod.
What can Puck IQ tell me that other sources (like http://Stats.HockeyAnalysis.com ) cannot?
The whole goal of the WoodMoney QoC rating system was to figure out how each NHL player did against the best and the worst opponents.
When GMoney and I used to discuss players together we knew we had to deflate the results for Dman or forward if he was getting 3rd pair or 4th line minutes minutes or inflate a 1st pairing Dman’s results because they played a lot against the best, but we didn’t know how much to inflate or deflate those results.
Both of us would pore over the game sheets and match ups to see who the coach was playing against the best and the worst. You could see game by game that players would get better results vs the lesser opponents and worse results vs the better opponents. Who you play against must matter, but we didn’t know how much it mattered.
Using puckiq you can see each player’s results against the 3 categories of forwards that we created, “Elite, Middle, and Gritensity”. You can also see how much time they spent against each category of forward, so you can understand more fully about how they were deployed.
One of the first things I learned was that not all teams deploy their forward lines or Dpairs the same. In 15/16 NJD absolutely buried Greene/Larsson by playing them against Elite Forwards 42% of the time and giving them a massive share of the defensive face off starts. This allowed them to play the other two pair at about 27% vs Elite Forwards, so it was like they have a 1st pair and 2 3rd pairs.
Every team/coach is a little different on how they deploy their roster so you can’t just look at corsi or time on ice and figure you know which level of competition a player is playing.
The whole goal of WoodMoney was to figure out “generally accepted” categories for QoC and then check everyone’s results against those categories. The players in the categories past the smell test and we think we now have information that is very useful.
We think the general NHL fan community would find this stuff interesting so Puckiq is making this information available to anyone who wants it.
How reliable are these metrics? Should I consider 250 minutes against elite competition a reliable sample size?
As always, the bigger the sample size the better. I find 300 minutes is a good line in the sand as most results don’t vary a whole bunch when they past that, but there are always exceptions.
How do you decide who gets the ‘elite competition’ moniker?
Every forward who is considered “Elite” must meet all 4 of these criteria:
- Points/60 must exceed 2.21 for all game states (not just 5v5)
- Time on ice must exceed the 75 percentile for all forwards. The coach must not shelter them.
- Relative Corsi must exceed the 40th The player can’t be a drag on possession. We dropped it to 40 because some players play on great teams and good players like Malkin and Kane can actually have negative relative corsis.
- Relative Dangerous Fenwick must exceed 40th The player can’t be a saloon door in their own end (Dman) and should drive pucks to the paint in the ozone (forward) Similar to the 3rd requirement, this one uses DFF or Dangerous Fenwick which is a shot metric that weights the “dangerousness” of each shot using shot location and shot type (slap, wrist, backhand, tip, etc)
Are these the metrics NHL teams would be using?
I’m not sure what NHL teams are using. From what I’ve heard what they use can vary greatly from team to team. Some team might have done something similar, but I wouldn’t know.
Do they pass the logic test? I mean, does Patrice Bergeron perform well against elites?
“Passing the smell test” was very important to us. The whole reason we did this was because the fancystats community had decided that QoC didn’t matter, and that didn’t past the smell test. Patrice Bergeron has the best relative corsi vs Elite forwards in the NHL last year and the 2nd best relative DFF. McDavid had the best relative DFF. Smells good.
Do you look at the middle and lower bins? What are you looking for when measuring play against mid-level competition? Should I take my team’s second and third line and measure them against the mid-level (to make things even)?
I think the lower categories are still important. Our original idea was to see “who was doing well vs The Best?” and “who was zooming their numbers by playing the dregs?” All results are important so it’s best to examine all of them to get a full picture of the player.
Lastly, we are not anyone “why” a player got the results they did. There are still tons of grey areas in that respect and just because a player got a result it doesn’t mean the reason they got that particular result is easy to figure out.
To help that part of the process along one of the first things we are going to add WOWY, which are “with and without you” data. It’s the same data, but you can take a player and disassemble their results even further by giving the results they get with each team mate.
I’m very pleased to help with the launch and want to congratulate all (Ganesh “GMoney” Murdeshwar, Zsolt Munoz, Derek Blasutti, Darcy McLeod) gentlemen for their accomplishment. Ganesh and Darcy are manning their twitter machines to answer your analytics and theory questions, Zsolt and Derek make this thing shine and can help you to move the giant levers. Ready, set, go! PUCK IQ!