So, the other day Alec Brownscombe of Maple Leafs Hot Stove felt the need to draw attention on twitter to a comment by fellow MLHS blogger Declan Kerin on a posting by (yet another blogger) Michael Stephens.
Here is the comment Kerin posted on a player review write up about Carl Gunnarsson:
"Using possession metrics for a team like the Leafs is where big, ugly holes are poked in advanced stats."
"The Leafs are a good example of what happens when you look at stats without context and draw convenient conclusions for people that are advanced stats evangelists. Not saying you are, Michael, I think you use them as they were intended: to support your eyes, and not to make up your mind based on them 1st."
"The Leafs sag off the points and take away the critical areas of the ice. 2nd chance opportunities and rebounds are cleaned up by layers of defence and forwards and brought down the ice, or poorly cleared, as was an achilles heel of the team all season. Reimer had some easy nights of lots of routine, positional saves where he was giving up big-time rebounds. He had lots of support and benefited from plenty of shots from non-critical areas of the ice."
"Where it throws a wrench into the plans: Centers drive position. [sic] We are THIN at center. We have no cycle. Like, nothing. The cycle chews up time and skews these results so it shows poorly on the Leafs."
-Declan Kerin, May 27th, 2013
Ah, the realm of the "big ugly holes in advanced stats." We all know where they are, these holes. They sit somewhere between explaining why a team like the Leafs have no reason to assume they'll be remotely capable of repeating their 2012-13 seasonal performance next year and being used as justification to jettison half the team because they are bad at hockey.
Unfortunately most of the people that feel there are "big ugly holes" in the advanced stats and argue for "context" then wish to INVENT contexts that "explain" what the numbers tell them - even if those have nothing to do with anything that has been measured, observed, or tracked in any way. Numbers are thrown out the window in preference for assertions that are anecdotal, unverifiable, and generally don't make a lot of sense.
In regards to Kerin's points regarding context - he's right - context is important. But saying that the Leafs seasonal shot differential was a result of the forwards collapsing away from the point and clearing rebounds from in front of the crease - is sort of unrealistic as explanations go. It might be true that the team did that, but the problem is the assumption that they (a) were the only team that plays this style of hockey and (b) that the Leafs centres are the reason the Leafs shot and possession proxies look so bad.
Here are a couple of points I would make regarding the Leafs and their defensive short comings - context if you will:
1) The Leafs defense did do a decent job of clearing away shots from in close this season.
2) Despite clearing pucks away from the crease - the team could NOT clear the zone (because they suck at possession). The reality is, the Leafs have very few players capable of "carrying" the puck out of the zone, and they tend to get hemmed in. This is largely because their D are not particularly effective at gaining control of the puck and keeping it away from the opposition.
Think about who the Leafs best "possession" players are on the blue line? The guys that are capable of making solid outlet passes, or can skate the puck out of trouble. The ones that make decisions with the puck promptly and generally speaking do something with it that is reasonably safe. Now if you plan on excluding a guy because you think he makes a lot of giveaways - don't bother... just forget that aspect for the moment and worry about the player's skill set and OVERALL decision making. Who would you list in that group? Gardiner? Phaneuf? Gunnarsson? Franson? Liles maybe?
Guess who ranks at the top of the Leafs D corps in terms of DEFENSIVE possession metrics? Here is the ranking from the past two seasons:
Gunnarsson, Kostka, Liles, Phaneuf, Franson, Gardiner, O'Byrne, Fraser, Holzer.
Gardiner's ranking is largely based on his play during the 2011-12 season as he didn't play enough games this year to qualify for the rankings. Either way, he's well ahead of the bottom 3.
Now think about how much time O'Byrne, Fraser, and Holzer were given on the ice this season - then think about which of those guys are considered the best at "clearing the front of the net". I'm pretty sure the majority of you would rank O'Byrne, Fraser, and Holzer as having skills that tend towards "clearing the net" rather than "playing the puck". Which makes sense, as they block a lot of shots, hit a lot, and basically chase the play constantly in their own end. Sure they deal with rebounds, by flinging them to the corner, where the opposition picks the puck right back up again, and so it goes.
Now think about how those last 3 D men play the point offensively. If the puck begins to exit the zone, do you think they pinch effectively? Do you think they reliably maintain possession by passing well in the offensive end? Probably not.
This is where those "Big, Ugly Holes" - or at least the people that see them - confuse the hell out of me. What is it about advanced stats that disagrees with the eye test here? Where do your eyes disagree with what we're seeing? Yes they get a lot of blocked shots and clear the crease of loose pucks... great... that doesn't help the Leafs gain possession back.
This also means that Reimer is required to make far MORE saves, and I'm sorry, but the number of difficult ones is probably higher than most other goalies just as a result of how many shots he faces. Let me walk you through a simple logical argument for a second regarding the ideas around quality shots faced by Reimer.
Goalie A faces 27 shots a night for 33 games, around 29% of the shots he sees are dangerous... that's around 8 shots per game. Goalie A has a .916 SV%
Goalie B faces 30 shots a night for 33 games, around 25% of the shots he sees are dangerous... that's around 8 shots per game. Goalie B has a .924 SV%
Goalie A ends up facing 252 dangerous shots, Goalie B ends up facing 253 dangerous shots. Goalie A lets in 74 goals, Goalie B lets in 76 goals.
Now - Since BOTH goalies let in around 75 goals... which one is doing a better job? Well on the dangerous shots goalie A let in 43 goals and had a .829 SV%, and goalie B let in 35 goals and had a .862 SV%... so probably goalie B?
Goalie A is Marc-Andre Fleury (74 goals against), Goalie B is James Reimer (76 goals against). They both faced almost an equivalent number of "dangerous shots". So arguing that Reimer had an easier time of it because his shots were easier is sort of absurd (and yes I know I'm only comparing him to ONE goalie - but seriously can we put this argument down for a little while?)
Not only is he facing the same number of DANGEROUS shots, he's facing more shots in total, and thus has an increased chance of one of the pucks shot from further out going in. So how again does having D men focus on clearing the crease make such a huge difference again?
The reality is - we use NUMBERS to determine performance. Nobody is out there saying Sidney Crosby is the best player on the planet because he LOOKS good when he plays even if no goals or assists are being recorded. Most viewers like to crap all over Grabovski's lack of production - they care about numbers, they just care about DIFFERENT numbers.
The long and short of all of this is - people like small, compartmentalized, easily explained statistics. Things they can count on their own at home, that take little effort to keep up with. Oddly though, while most fans can see a shot attempt, they probably don't know exactly which shot attempts count as shots on goal, blocked shots, or missed shots. They know what these things are, they just don't typically count them on their own. Luckily the NHL does.
People also already do some simple math in their heads when computing and understanding things like +/- as a stat. They KNOW how to figure out what it means. The fact that it doesn't actually correlate particularly well with ANYTHING doesn't mean much to the people that use this statistic. It doesn't HELP you understand or describe what is happening on the ice. It measures a result, not a process. It measures a result that the given player may have had NOTHING to do with on the ice, and may happen with alarmingly rare frequency. But people still use it.
Nobody goes around asking about the CONTEXT of plus-minus numbers. No average fan gripes that you need to understand the CONTEXT of a guy having a great plus-minus or a horrible plus-minus. Nobody spends a lot of their day complaining that secondary assists require CONTEXT before they get counted on a score sheet.
Yet people sit down and argue that they have to add lots of "context" to explain away advanced stats as meaningless. There is copious context provided within advanced stats, and most people making use of them will gladly discuss the contextual implications with you. Understanding the explanations sometimes takes some effort, but to sit and assert that they're meaningless if you don't want to take the time to understand them is mentally lazy at best, and willfully ignorant at worst.
Shot Metrics are making a regular appearance in the playoffs on TSN and HNIC, these stats aren't going anywhere and if anything they're likely to get more prevalent, not less. Let's work on spreading knowledge, not twisting or ignoring it.