News and Notes – January 12 2015

It’s a practice day for your Calgary Flames as they prepare themselves for the remaining four games of their current “road trip” – a big one including stops in Glendale, San Jose, Los Angeles and Anaheim – so let’s take some time to poke around the various happenings regarding the local hockey club.

Brian McGrattan cleared waivers and was assigned to the AHL’s Adirondack Flames on Saturday. McGrattan is perhaps one of the nicest guys in hockey, and routinely is used as a resource by the NHLPA for any players who are having substance abuse problems and need someone to talk to. In short: he’s a tremendous human being. That said, he’s a marginal (at best) NHL hockey player, and he’s probably only slightly better than most of the players in the American League. That said, he could be a pretty solid resource for a young Adirondack locker room – similar to how captain Nolan Yonkman is being used there. For now, the Flames will be carrying 22 bodies.

Mark Giordano was invited to the NHL All-Star Game in Columbus, while Johnny Gaudreau will represent the Flames in the Skills Competition. Giordano’s one of 12 defenders invited, while Gaudreau is one of six rookies. It’s the first trip to the festivities for both. Giordano’s the first Flames All-Star not named Iginla, Phaneuf or Kiprusoff since Valeri Bure and Phil Housley in 2000. The All-Star Game is January 25 in Columbus, and the Flames will be off entirely for four days (January 22-25) for the All-Star break.

Morgan Klimchuk, Eric Roy and the Brandon Wheat Kings beat the Calgary Hitmen 5-4 yesterday. It was a pretty entertaining game, all things considered. Klimchuk was minus-1, but that goal was scored shorthanded off a defensive gaffe by draft-eligible Russian Ivan Provorov – who made up for it with a three-point effort. Klimchuk was solid defensively overall, but never really was able to pull the trigger when he did get the puck in the high-percentage areas. Lots of double-clutching offensively, but he was effective on the forecheck and using his speed to generate chances and turnovers. He also started off a nice passing sequence in the Hitmen zone on a power-play that led to a goal.

Eric Roy was a tale of two halves of the game: he’s not a particularly great defensive player, and he coughed the puck up a few times and generated some Hitmen scoring chances as a result. However, Roy did a solid job jumping into the rush in the second and third periods, including having the first assist on Brandon’s second goal.

On the other side of the ice, Keegan Kanzig was adequate for the Hitmen. His game isn’t flashy, in that he doesn’t make a lot of flashy plays but is usually in position. He got into a first period fight, but beyond that was pretty quiet. He was matched up frequently against Klimchuk’s line, and while they didn’t score on his pairing, he did seem to have trouble with the forward unit’s speed, as he was frequently beaten for loose pucks (and on the forecheck) by Klimchuk and company.

-And now, a graph (5-game rolling Corsi For percentage):

Screen Shot 2015-01-11 at 5.28.55 PM

The Flames are 29th in the NHL for Corsi For this season (only ahead of Buffalo) and 28th in the league for Corsi Close this season (ahead of Buffalo and Colorado). They’re a 44% Corsi team right now, which basically means they’re spending roughly 56% of the time in their own end.

They are what they are, at this point. If they get the bounces and execute a bit better, as they were earlier in this season, they may be able to nudge their percentages to about 46% or 47%, at best. That may equate to a few more wins over the long-term, but if things stay about where they are, expect the team to slide a ways down the standings from here on out.

This road trip is definitely going to be key to the direction the team will take for the remainder of the season.

  • piscera.infada

    Ryan,
    I loved the review of the prospects at the Hitman-Pats game, but you went off the rails with the Corsi schtick.

    The big Rolling Corsi peak corresponds to the Flames 8-game losing streak, when the team definitely wasn’t playing as well, in spite of what the Corsi may be saying. Perhaps you’d be better off using your talents on figuring out why the Corsi story is wrong and coming up with a better measure to indicate when the Flames are actually playing well.

    • mk

      An interesting take on an issue with Corsi ratings:
      http://blogs.edmontonjournal.com/2013/08/01/why-corsi-numbers-get-it-wrong-with-some-players/

      Corsi IMO:

      1. Is a great tool which improves upon the old +/- stats which were used to gauge a player’s effectiveness on the ice, not just his direct contribution to scoring (goals, assists).

      2. Corsi/fenwick only came around because shot rate data was becoming more readily available & accurate. Before this, Corsi was impossible to calculate, so +/- was used.

      3. The subjective nature of a shot vs. not-a-shot is small. I.e. it is easy to say whether an event was a shot or not-a-shot than to rate its ‘quality’.

      4. The natural progression away from Corsi would be to track the quality of shots (i.e. explaining why Corsi is wrong sometimes). This is where we will get eventually. But:

      5. Shot quality data is not tracked by the NHL & is not readily available right now. Some people are tracking it for a specific team, but there is no single authority which gives shot quality data for the league now or past seasons.

      6. Shot location (i.e. distance) is not currently a reliable proxy for quality as the shot location is tracked VERY badly by the league. League scorers have a history or unreliable location tracking (look up a graph of where they mark down a face-off vs. the actual face off dots).

      7. Video tracking is being developed – but it is mostly behind closed doors (by individual teams). It is (currently) not accessible enough to create stats around.

      Tl;dr:
      Corsi/fenwick is not perfect, but we don’t have good enough shot quality data to improve it. People are working on it.

      In the meantime:
      5v5 Fenwick Close is the best predictor (at a team level) of future success (wins) of any stat we track currently, even better than current winning %.

  • piscera.infada

    The NHL should establish draft position by using Corsi Close! This is a better reflection of team play than Wins and Losses! πŸ™‚

    The first period of the Hitmen game was kind of boring so I switched to the NFL game..could have been the early start that effected play since the teams both played Saturday night. Only the AHL has worse schedulinhg – Friday, Saturday nights and Sunday afternoon games…..

  • everton fc

    I like Eric Roy. He’s a good kid. Sometimes I think if he is signed and taught to play the position, he’ll be okay. In fact, with an adequate professional pairing, he might be quite a nice offencive defenceman. Other times, he sounds more like Anton Babchuk. In any case, II hope he gets a long look.

    I still don’t think Kanzig has the skating tools to make it, but I hope I’m wrong.

  • mattyc

    Looking at the rolling corsi chart, it’s pretty clear (to me at least), which part of the season has been an outlier so far.

    Related: the Flames defensive game has been especially bad the last few games. This is something they’ll need to figure unless they expect to squeak through games like they did against the canucks on saturday.

    also: does anyone have any insight into Drew Shore?

    • mattyc

      “also: does anyone have any insight into Drew Shore?”

      Nothing more than what I read…that BT has been monitoring this kid for something like 5 years and likes what he has seen.

      It would be strange for the Flames to put him on the roster without having a good evaluation in the AHL of his NHL potential…one more game and he is subject to waivers if he isn’t NHL ready according to the Flames systems.

  • Derzie

    The Corsi chart shows a ‘peak'(good) during the Flames losing streak (bad). That alone is a point that can’t be ignored in the quest to understand the value of the stat.

      • crapshoot

        I see the value of corsi etc. to a certain extent but it does not correlate very much to standings at this point and I don’t understand the need to defend it like that. Edmonton and Carolina are pretty good in the corsi department, but they flat out suck. Montreal is very poor when it comes to corsi, but they are rolling along just fine. The Flames are a lot worse than the Kings and the Jets according to corsi but they still fight for the same position. The Avs are getting back in the hunt despite very poor corsi numbers. It’s not like corsi is missing the mark just that one time.

      • JMK

        I think the frustration with the stats community at times is it’s over reliance on certain stats. Corsi is mentioned a lot on this site, and while I’m sure it’s a useful tool there rarely seems to be a question of how good of a tool it is by the writers on FN (bar Kent Wilson his take is usually fairly grounded and explains his points well). I don’t mean to insult the other writers, I love this site and read it every day, but there is constant theme of “well Calgary’s Corsi is still in the bottom 3 so that’s where they should be” kind of comments. Corsi is based on large samples and I get that so saying Calgary had a good Corsi for their 8 game losing streak shows Corsi’s failure is inaccurate. But my question would be when is the sample size big enough?? Calgary at this stage of the season (42 games) have shown they are either going to scrape the playoffs or just miss out. So instead of being bottom 3 like their Corsi suggests they have maintained a 14-18th kind of range. It was fair enough at the start of the season to say Calgary were going to come back down to earth, but are Calgary not showing that Corsi isn’t the godsend of statistics?

        I’m a relatively new fan to the world of hockey but I can’t understand how it’s always corsi “possession” hockey teams that do well. Is there no room for tactics in this sport? For example a team that may invite teams on to them and then counter attack would have a low corsi but may have an effective system. Correct me if I’m wrong, maybe it’s my influence of other sports, but it just baffles me that the only “good” teams are “possession” based teams. Also for instance what if a coach wants his team to only take shots from high percentage areas, they play around more and take fewer shots in the offensive zone – this team might actually be a good “possession” team but it wouldn’t show up in the Corsi side of things.

        • JMK

          Also for instance what if a coach wants his team to only take shots from high percentage areas, they play around more and take fewer shots in the offensive zone – this team might actually be a good “possession” team but it wouldn’t show up in the Corsi side of things.

          Actually it would, because if they have the puck that much they’d still out-shoot (attempt) their opposition. They’d just be an extremely low-event team.

          That is, if they were good. The above quote is exactly–verbatim–the argument Leafs fans made a couple years ago. But the Leafs were so bad that as soon as their goalies weren’t stopping 93% of the shots they faced, and the team wasn’t scoring on 15% of its shots (which would only be possible if every single shot on goal was a scoring chance) they started to lose. Colorado has gone through this twice now.

          The mythical “shot quality” can make up for a shortcoming in shot attempts–but only to an extent. Either in a very small sample, or making a team with even possession numbers a contending team because of shooting talent (and ideally goaltending).

          The proverbial shot quality team still has to out-chance their opposition.

          • JMK

            I get your point and agree that they would be a low-event team but would they necessarily out-shoot their opponent? I don’t think so but it was just a hypothetical argument.

            You see this is what I’m talking about, fair points to make about Colorado and the Leafs, but that’s the same two comparisons that is constantly made. And I’ll agree this applied to early season Calgary, but shooting percentage and save percentage (or PDO if you like) has returned to normal now but Corsi has stayed low. Yet Calgary remain a lot higher than their corsi suggests. The point I’m making is the advanced stats community will make statements about corsi using it as the main factor for Calgary’s future regression. At least early in the season it was accompanied with the high shooting and save percentages which I thought were fair points. But now that Calgary is still in the playoff hunt despite a low corsi after what I think is a large sample of 42 games (correct me if I’m wrong but one of my questions is what is deemed a large enough smaple) and there is still no question to the usefulness of Corsi unaccompanied by other stats. Which I still question is there no way that you can see that tactics could render corsi useless for certain systematic teams?

            I’ll give you another example, let’s say a team is extremely disciplined and doesn’t give away a lot of penalties and also draws a lot of penalties. Basically a team of Johnny Gaudreaus. So this team spends a lot of time on the power play and thus generates less 5 v 5 corsi events than usual. So they also score a lot on power plays. Now I know you’ll say 5 v 5 they should still do well, but when teams go ahead they can tend to sit back in a lot of circumstances. So they could have a lot of shot attempts but not necessarily 5 v 5 shot attempts and thus may have a low corsi yet still be a good team. Interested in your thoughts as I’m sure I’m not the first to question corsi and advanced stats. And I’m not against advanced stats, I’m against them being used as the bible and ignoring times when they seem to be contradicted.

            I’ll give you another reason I’m asking about the sample size for Calgary. Let’s say similar to Calgary at the start of the season their goalie was playing out of his skin with a save percentage upward of 0.93. Yet his previous 3 seasons of NHL experience his save percentage of 0.91. At what point does this guy banish the stats crowd saying he’ll regress to his career average? If he kept up the save percentage of 0.93 for let’s pick a random number of 42 games would the stats crowd still say he’s going to regress or perhaps would people start to say maybe this guy has gotten better??

            It just frustrates me a bit when people will point to the stats will regress without looking further. I mean Glencross is a high percentage shooting guy, let’s say BT signed a bunch of these guys. Calgary’s PDO would improve, then would people say the PDO is bound to regress or will they look at the individual players and say hmm maybe the PDO will stay high. And I’m sure some people will do the in depth analysis but I think people can be too quick to say naw the PDO will regress.

          • Byron Bader

            Interestingly, Calgary was outchanced in 6 of the 8 games they lost in a row. Earlier in the year the Flames outchanced or were even with opposing team 17 out of 26 times.

            Kent’s point that their corsi would probably be a lot higher if they generated more shots as they don’t tend to give up that many attempts against also suggests they wait for their opportunities.

            I definitely think there needs to be more focus on shot quality, as you suggest. A muffin top point shot from a horrible angle at the back of the blue line should be weighted as 1/50 the corsi event of a close in open one-timer. The analytics haven’t got here yet. Definitely making strides though.

          • Burnward

            And there are lots of cool new stats out there.

            I suggest checking out Hextally at war-on-ice.com. Really neat look at overall team play strength and weaknesses.

          • mattyc

            And unless something has changed dramatically in the last couple weeks, I believe this data suggests that the Flames don’t actually prevent in-close shots more than other teams (one of the suggestions people have made as to why the Flames seemingly ‘contradict’ possession theory).

          • Burnward

            Yes, they allow more shots, but the shooting percentage from the scoring area in front is lower.

            Also shows that Calgary scores at a higher percentage from the higher slot area. Probably due to Gio/Wideman.

            Edited for me being a dummy.

          • JMK

            So this team spends a lot of time on the power play and thus generates less 5 v 5 corsi events than usual. So they also score a lot on power plays. Now I know you’ll say 5 v 5 they should still do well, but when teams go ahead they can tend to sit back in a lot of circumstances.

            That’s why we look at score close.

            Yet Calgary remain a lot higher than their corsi suggests.

            You forget that Calgary was something like 6th in the league before the regression hit. Now they’re out of the playoffs. Regression doesn’t erase the head start, it just lets everyone else catch up.

          • JMK

            Okay but generally their average corsi has stayed roughly the same throughout has it not (bar the spike on the losing streak). So it’s PDO (save percentage and shooting percentage) that regressed towards the normal figures. I’m not saying Calgary are good enough to be a top 6 team but 42 games in (with stretches of high and low PDO) they are 18th, not 28th which Corsi 5v5 would imply.

            But the line “Regression doesn’t erase the head start, it just lets everyone else catch up.”; is this not the same thing?? Calgary rode the wave of high PDO something like 102-103, then collapsed to something like 96-97 during their losing streak. So teams caught up due to their regression, no?? If they had regressed to 100 for a stretch of games, then fair enough their head start would still count but then their overall PDO would still remain fairly high.

            But even put the Calgary example aside for a minute do you not get my points on corsi??

          • JMK

            No, and you’re arguing against yourself. Scoring chances and shot attempts(Corsi) aren’t the same thing. If the one team ONLY took shots within the Scoring Chance zone and the other fired the puck from anywhere inside the Blueline the rate of success(goals scored) differential for the two zones would more than make up for the poor-Corsi team’s success, even with far fewer shot attempts.

            Corsi is a poor man’s stat that is easy to determine and is transferable. That’s about it. Much better is needed.

          • JMK

            No, and you’re arguing against yourself.

            No, I’m not. And I never even implied that scoring chances and shot attempts were the same thing.

            In order to “work the puck around” enough to get it into the scoring chance area you have to have the puck. While you have it, the other team doesn’t. You take a shot. 1-0 Corsi. 1-0 scoring chances. A successful “shot quality” team would BY DEFINITION out-chance their opposition, and would, necessarily, out-shoot them with the score close as well. Simply by nature of the purely hypothetical system.

            By the way, what’s the best (or at least most consistent) way to generate a scoring chance? Take a damn shot.

          • JMK

            You seem to be assuming that the time it takes to shot the puck is the same no matter what type of shot that is. Let’s look at two teams, SJS and DRW.

            San Jose brings the puck into the zone and quickly shoots it at the net. They are big and strong and own the boards so get 2 additional chances, also at the net. Within 30 seconds they have 4 long-distance Corsi events (1 blocked) in 30 seconds and finally the puck gets out. Detroit carries the puck in, they dipsy doodle around, probing for a weakness, cycling. After a while they make a backdoor play and shoot it at the net, getting 1 Corsi event in 30 seconds. Detroit had by far the better chance to score even though they are getting killed at Corsi, while actually possession time-wise is exactly the same.

            If you assume a Corsi event equals a certain amount of time of possession you can be mislead into thinking a team dominated when perhaps it did not. When you watch a game you can tell whether or not that is the case. Also, I haven’t checked but I’m pretty sure if you checked the goals/Corsi event stat (not tallied yet) you’d find a strong negative (inverse) correlation. This would have a lot to do with overall team play and tactics rather than the largely misleading Corsi stat.

        • Burnward

          Great comment.

          Hartley’s strategy with this team seems to be “Collapse and counter.”

          Of course they’re going to lose the Corgi battles with this system.

          When they play sound defensively to get in lanes, eliminate second chances and frustrate the hell out of teams is when they are most effective.

  • Bean-counting cowboy

    Button’s rankings has several good D men that will be in the Flames wheelhouse. If we could land one of (ranked 5 – 8):
    – Hannifin
    – Kylington
    – Povarov
    – Werenski – I would be stoked!

    Other options that look good:
    – Mikko Rantanen: 6’4″ 211 lbs RIGHT WING! ranked 11th. Loved his play at World Juniors
    – Lawson Crouse – same size as Rantanen but is a left wing – adds some much need size

    Another interesting note. Tyler Wotherspoon’s younger brother Parker (also a defensemen) is ranked 35th. Don’t know much about him. Maybe we can snag him with our 2nd rounder? They could form a brother defensive pairing for us!

    • KiLLKiND

      Rantanen seems amazing and I would be fine if we used our first round pick on him or any of the D listed above. but what do you think would be required to get another pick to get both picks? assuming Flames finish and draft 11th overall what would we need to trade to say get 7th overall pick as well?

      • Bean-counting cowboy

        That would be nice in a deep draft. Not sure if we could pull that off. Teams don’t give up those high of picks in deep drafts very often.

        My hope was to land one of them wherever we end up picking. One of them could fall like Maatta did a few years ago… maybe the Russian, as teams tend to shy away a bit (although the KHL draw is less than what it used to be).

        My thought is any team pegged to draft in that 7 range would require a return too high for us to consider… but who knows. And what we would be offering (vets) likely wouldn’t match well with what those teams would be trying to look for.

  • mattyc

    At a certain point, people need to acknowledge no one thinks or argues seriously that a) possession is the only thing that matters, b) that shot quality doesn’t exist.

    But the thing is, there is overwhelming data, over a half-dozen seasons, that show that possession metrics are the best predictor of future success, and that while shot quality exists, it’s not nearly as important as people think, and is swamped by ‘bounces’, bad goaltending, and other non-repeatable factors.

    • JMK

      And I completely understand where the confidence in corsi is coming from. I also know that not everyone argues seriously that possession is the only thing that matters but it’s quotes from the article above like “if things stay about where they are, expect the team to slide a ways down the standings from here on out” in reference to corsi and only corsi that can be frustrating.

      Look I’m looking at this from a non-hockey background so from a soccer/gaelic football/rugby background. And yes generally possession teams do quite well in the long run. But there a lot of non-possession team that come around and can win a few stanley cup equivalents. And it just baffles me that the opinion I constantly get when I question advanced stats is that the overwhelming data shows possession teams do better. It’s a fair point but are corsi possession teams always the best teams? I guess maybe I should ask before the 6 years of corsi data collection could there have been systems where teams that didn’t outshoot their opponents work well and win a lot of games?

      Just found an interesting stat on the NHL website. And now I’m guessing shots mean shots on goal rather than corsi for events but still interesting. It measures the winning percentage of teams when they outshoot their opponents and also the winning percentage when they are outshot. So the league average winning percentage for outshooting an opponent is 53.3% and the winning percentage when outshot by an opponent is 46.7%. Which I’m sure you’ll agree links closely to the corsi philosophy. Now Calgary’s winning percentage when they outshoot an opponent is 38.9% but when they are outshot it is 59.1%. That’s a large differential. Now my question is how large of a sample does this need to continue for to perhaps contradict corsi?? For reference Calgary last year was 32.4% and 53.5%.

      • mattyc

        Depends quite a bit when the team was leading from. A team that is up all game is going to sit back more, and go in shell mode, and probably get outshot to a non-trivial degree, even thought the game was never in doubt. You can call this ‘score effects’, which is why sometimes it’s better to look at Corsi (or fenwick) only in situations where the score is close, to try to eliminate that effect on the data.

        • JMK

          And that’s the kind of in depth analysis I’m looking for. But my point at the beginning is it seems to me some people are too quick to say oh look Calgarys 5 v 5 corsi is 3rd lowest in the league therefore they should be 3rd lowest in the league rather than taking a range of topics into consideration. I’m not saying corsi isn’t useful I just think it’s brandished far too often without any backup. I think corsi relative is quite useful when comparing players for example but at the same time you can’t look at corsi relative without having some idea what kind of role that player plays either.

      • everton fc

        In the EPL Everton is a high possession team that can’t win of late, nor can they score of late. There possession percentage mirrors Chelsea. Yet they sit 3 points from the relegation zone. Thought I’d throw that in!

        • JMK

          Exactly that’s the kind of thing I’m talking about. But every time I ask does that transfer over to hockey I’m told there’s so much data that shows this and this. Same way that Chelsea were ultra defensive not so long ago with their park the bus mentality yet beat a lot of teams out the gate. I’m sure if soccer staticians counted corsi Chelsea would have been very low yet won the Champions League beating pure possession teams like barca. Like a common tactic in gaelic football lately is teams start defensive and as they progress further in their development slowly become more attacking. It’s working well for a lot of teams. I just want to know does any of this transfer over at all.

          • everton fc

            I think it does. I don’t know the Devils Corsi when they played the way they did in their best years. It’d be interesting to see.

            Look at Stoke over the years. “Parked the bus” all those years w/Pulis. Every team from United to Arsenal to Chelsea had trouble with them at Brittania. Palace did the same when Pulis was there, and stayed up. I’m sure West Brom will do the same now. It’s sports. It translates. (Good point on Chelsea, by the way!)

            On this side of the ocean, people won’t pay for the type of game teams like Stoke play. I’ll add this: watch any EPL game out of UK/Ireland and you’ll never see a stat posted while the game’s live. I love this. We get to watch the game. Watch any North American sport, and every 30 seconds, we get stats. Hate it.

            Kybb79 said it best. “Whats with all the corsi talk? I come on here to read about some real hockey talk like trade suggestions, rumours and about the team in general.” Spot on. Hockey’s not math. Or science. It’s sport. I remember in college we were outshot 43-14. We won 4-2. It happens. How are the Leafs doing w/their statistician guru Dubas?

            Glad to see another GAA fan on here. I pull for Co.s Donegal and Mayo myself. Where my “people” are from! πŸ™‚ I loved when Setanta had those Tuesday evening GAA nights! We pulled the plug on t.v. in ’09. Should find a place on the net to catch the games live. Welcome aboard. Hockey’s a great game. Masculine, like hurling and Gaelic football. In fact, English and Scottish soccer is heavy contact, as well. Just look at Steve Bruce’s face! πŸ˜‰

  • mattyc

    Often on TV brodcsats, reference is made to “quality scoring chances”. Who is tracking this data and what factors determine a “quality scoring chance”? Is this only tracked for TV games?

  • piscera.infada

    Did anyone hear Treliving’s interview on the Fan this afternoon? I was in a meeting and had to miss it–I’ll definitely be listening to it tomorrow morning. Did he say anything good/bad/interesting?

    Heard he was talking prospects and the general direction of the team in the first half. I’ve found he’s usually a pretty good interview: insightful, honest, and well spoken.