[Rumour] Flames hire David Johnson - Analytics Team

[transplant99]

Originally Posted by wretched34

Thank goodness for analysts..... I'm sure no hockey player ever, was smart enough to realize his best chance of scoring was when the goalie was out of position, and from in front of the net......
I get that advanced stats play a role, and I suppose in their infancy they need to quantify common sense to prove value, but wow, it's amazing guys get paid to be "Professionals" by giving advice like that to Hockey players.

I could save them a fortune.

Just put on an old VHS tape of Tim Kerr when he played for the Flyers, sit them down and say "watch what he did and do that".

[Robbob]

You will get some that I think will publicly poo poo advanced stats and analytics, but the reality is that I would bet every team has its methodology and process they use as a closely guarded source of info.

The landscape for the NHL is pretty even so any little thing you can do to gain an advantage you do it. If you come across something from reviewing all the data and proves to be repeatable and predictable you will definitely keep it close to the vest. Treliving mentioned Chicago has a "system" they use and they keep it very close to the vest.

[VilleN]

Originally Posted by wretched34

Thank goodness for analysts..... I'm sure no hockey player ever, was smart enough to realize his best chance of scoring was when the goalie was out of position, and from in front of the net......

I get that advanced stats play a role, and I suppose in their infancy they need to quantify common sense to prove value, but wow, it's amazing guys get paid to be "Professionals" by giving advice like that to Hockey players.

Yeah, there's a little more to it than that.


Sent from my iPhone using Tapatalk

[Heavy Jack]

A calculatooor !

[bubbsy]

i'm always curious what these guys are given as objectives for the team..

Applying analytics to amateur scouting/drafting?
Applying analytics to pro scouting (ie. which type of players to target from an acquisition standpoint)?
Applying analytics to the coaching staff/style of play decisions, and/or line combo decisions?
all of the above??

I would imagine it can't be all of those, as that would be a massive workload, then again, i also don't know how big an "analytic staff" size is, other than these known names that get mentioned when hired.

[iloveicedhockey]

Seems like an interesting hire, so long as we don't promote him extremely quickly to GM.

[Matt Reeeeead]

Presumably just more of a guy to help with coding and database management. Nothing to see here. A behind the scenes move.

[Coach]

Originally Posted by bubbsy

i'm always curious what these guys are given as objectives for the team..

Applying analytics to amateur scouting/drafting?
Applying analytics to pro scouting (ie. which type of players to target from an acquisition standpoint)?
Applying analytics to the coaching staff/style of play decisions, and/or line combo decisions?
all of the above??

I would imagine it can't be all of those, as that would be a massive workload, then again, i also don't know how big an "analytic staff" size is, other than these known names that get mentioned when hired.

I would definitely think it's "all of the above".

Thing is, these guys aren't working from scratch. They're basically being hired for their intellectual property, not necessarily their personal hockey input. They have all the algorithms and statistical regressions made already, they just need data. Now, instead of collecting less data on more teams and publishing his results publicly, now he will get very specific data on one team and give them the results exclusively. It will be up to Treliving et al to decide what to do with the data.

So it's probably more likely working the scouting staffs in all departments to give them new/different parameters to track.

[Philly06Cup]

Originally Posted by Robbob

You will get some that I think will publicly poo poo advanced stats and analytics, but the reality is that I would bet every team has its methodology and process they use as a closely guarded source of info.

The landscape for the NHL is pretty even so any little thing you can do to gain an advantage you do it. If you come across something from reviewing all the data and proves to be repeatable and predictable you will definitely keep it close to the vest. Treliving mentioned Chicago has a "system" they use and they keep it very close to the vest.

I'm not so sure. Most notably, the NFL is rife with terrible, terrible coaching decisions. Any analytic papers/reports will say to use a much higher pass-run ratio and to gamble more on 4th down. Professional sports are still an old boys club following archaic models.

[dustygoon]

Originally Posted by Freeway

Johnson's data work is top-notch (WOWY analysis especially), but some of the views he expressed, especially regarding stuff like shot quality, are rather regressive compared to where analytics have gone in the last decade.

You don't think adjusting for shot quality is relevant? High possession team with all shots from the wall is better than slightly lower possession team with all shots from the slot?

[Calgary4LIfe]

IIRC, Treliving himself stated that the Flames don't use CORSI as a stat, but rather have their own in-house metrics. I always thought CORSI was a stupid stat to follow anyways, as it is an indirect measure of possession.

I am assuming that David Johnson has metrics that he has developed that will add to the Flames' in-house analytics, and/or will be working on creating/modifying existing ones.

I must say, the one that that baffles me is how often some 'advanced metric gurus' use analytics to prove something about a player. In my opinion, the one clear strength of analytics is to remove bias. Why some people go and use some specific analytics while intentionally ignoring others is what I think turns people off at times.

For example, things like HERO charts are basically a waste of time for me, especially when you are comparing two different players under two different situations, against differing quality of opposition. You get asinine examples of how a player who is no better than a #4 defencemen will out-shine a quality top pairing guy. Stupid things like that.

I find analytics interesting, and there are some really bright people drawing some really good conclusions, but there are a bunch of bozos out there just adding more bias to their favorite/hated players instead of removing bias. That's what makes me find dislike analytics at times. That, and stats like team PDO (this is stupid) and using CORSI as a measure of possession.

I like that the Flames are adding to this area. I like that the Flames are growing their hockey ops into one of the biggest in the league, actually, and how they are very progressive.

[Robbob]

Originally Posted by Philly06Cup

I'm not so sure. Most notably, the NFL is rife with terrible, terrible coaching decisions. Any analytic papers/reports will say to use a much higher pass-run ratio and to gamble more on 4th down. Professional sports are still an old boys club following archaic models.

It is changing though, at least in hockey. The Flames have their own set of metrics, as mentioned above (Trevling and CORSIesque metric).

I wonder how much of this has to do with adding analytics or data points vs blocking other teams getting the info.

[Oil Stain]

Originally Posted by Philly06Cup

I'm not so sure. Most notably, the NFL is rife with terrible, terrible coaching decisions. Any analytic papers/reports will say to use a much higher pass-run ratio and to gamble more on 4th down. Professional sports are still an old boys club following archaic models.

I don't watch football, but I see similar sounding arguments in hockey. I'm not saying you aren't right about those instances in football. For all I know you are 100% correct, It just reminded me of something I have seen said in hockey.



Stats guys have found that the likelihood of an attacking team scoring a goal when crossing the blueline with the puck is much higher than after dumping the puck in.

That's just common sense, but I think people were pretty shocked by how much more successful carrying the puck in was. Like just blew dump ins out of the water by a margin that was pretty unexpected.

As a result I've seen a strong push from some stats guys towards teams avoiding dumping the puck as much as possible because the alternative is so much more successful.

But, that fails to take into account that carry ins are way more successful at creating goals because NHL players and their coaching staff make decisions that create this specific situation.

For instance, if third and fourth liners starting trying to dangle at the blue line with higher frequency or guys tried it at the end of a shift, the ratio of goals scored after gaining the blueline with possession would drop. Also, the number of odd man rushes in the other direction would increase (Which is never mentioned oddly).

Other teams would also quickly adapt and just line up their players at their own blue line, and the ratio of successful carry ins would drop like a rock.

So while it seems when only looking at one specific stat that NHL teams aren't following a smart strategy, it makes much more sense when looking at the big picture.
.
.
.

[Oil Stain]

To throw a bone to analytics after my last post which was kind of anti-stats, what about the New York Rangers?

Historically, every team has had some tough guys and defensive/checking guys at the bottom of the roster that scored very little. There were and are some very famous no offence/shutdown guys.

While I think everyone agreed that your Boogaards didn't add much of anything in the hockey playing department, your Bouma/Hathaway types were and still are seen as very valuable players to have.

The majority of stats guys have never liked any of these types of players and have advocated against them for years.


The Rangers last season built a forward core of basically all skill based players. Last year they were 4th in GF and 12th in GA. 4th in the league in scoring despite having zero 60+ points players. They just employed 12 forwards that could all score.

Their top three forwards in PIMS had 58, 32 and 26 minutes which has to be the lowest total in the league by a fair margin.


It seems like the make up of this forward core was inspired by analytics, and it performed very well last season.

[14]

Originally Posted by Philly06Cup

Any move towards more analytics is a good move.

Combined with seasoned experience like Burke and Maloney and Newer minds like Treliving and Conroy. Covering all areas very nicely.

I like the organizational model very much.

[Coach]

To kind of pile onto Oilstain's point, one the things statistics has a difficult time tracking, particularly in hockey, is the opponent.

What is their gameplay? Who is their coach? Who is playing against who. That person was in the box at this time so that player got out against this player and scored. All the starting goalie in his division were hurt for periods during the season so his goal total is maybe higher than normal.

The one on one instances are largely chaotic. Whereas is baseball and football positions are separate, opponents are static, plays are readable, styles are predictably better/worse against others. They're just not as variable.

[jammies]

Originally Posted by Oil Stain

Stats guys have found that the likelihood of an attacking team scoring a goal when crossing the blueline with the puck is much higher than after dumping the puck in.

I'd say much of this correlation comes with players being much more likely to carry the puck in when defenders are out of position and cannot force them to dump the puck. Not to mention every breakaway and odd man rush is going to start with carrying the puck in.

[Boreal]

Originally Posted by wretched34

Thank goodness for analysts..... I'm sure no hockey player ever, was smart enough to realize his best chance of scoring was when the goalie was out of position, and from in front of the net......
I get that advanced stats play a role, and I suppose in their infancy they need to quantify common sense to prove value, but wow, it's amazing guys get paid to be "Professionals" by giving advice like that to Hockey players.

That is not what he is saying. The points he was making are considerations of correlation vs. causation when assessing a player holistically. Were they good, lucky, or both. Better questions & better considerations mean better answers.

If you think analytics are used to give players tips on where to shoot keep your head in the ground with the other ostriches and carry on.

[Enoch Root]

Originally Posted by jammies

I'd say much of this correlation comes with players being much more likely to carry the puck in when defenders are out of position and cannot force them to dump the puck. Not to mention every breakaway and odd man rush is going to start with carrying the puck in.

Exactly.

The other thing missed by that singular stat is turnovers and scoring the other way.

Dump-ins result in less scoring, but they are safer defensively. Carrying it in results in more blue-line turnovers that create odd man rushes the other way.

Until there is a stat that can factor those results in with the positive results, this stat doesn't impress me much.

In fact, it illustrates the primary problem I have with analytics in hockey: people are too quick to draw conclusions from single stats or 'perceived' situations.

[Robbob]

I am pretty sure there is a stat or analytic about the turnovers that lead to points. I think we named it the DERP (Decision/Errors Resulting in Points). Pretty sure it came out of a Taylor Hall article.