NHL Thread: ENTER NOW FOR ANNUAL LEAFS MELTDOWN

Congrats BWC

 

Low rent wearing the jersey of a former player at your wedding though. Go out and get a Crosby jersey or something.

 

https://theathletic.com/471276/2018/08/16/how-well-does-each-team-forecheck/?redirected=1

deep into the weeds of fore-checking and analytics. great stuff here imo.

Spoiler: for those who don't subscribe

How well does each team forecheck?


By Ryan Stimson 23m ago
The hockey analytics community has spent a good amount of time exploring the value in isolating blueline events, specifically entries and exits. For those unfamiliar with these events, an entry is the act of entering the offensive zone, which is commonly done with possession as a carry in or without possession as a dump in. An exit is much the same process, but exiting the defensive zone. There has been a wealth of information written about the former and some written about the latter, but this piece is about what the inverse of these events can tell us about teams, specifically forechecking.

In examining the percentage of zone exits a team’s opponents complete while maintaining possession of the puck, we can evaluate how successful teams forecheck and disrupt the opposition during this phase of the game. If we can measure it, we can then assign values to it and learn how important that aspect of the game is in generating shots and goals compared to others. Let’s get to it.

The Data

Special thanks to Corey Sznajder as always for collecting the data the actual league decides not to. Also, shout out to Alex Novet for helping in the data cleaning process. Everything is 5v5 and data not collected by Sznajder is from Corsica.

The Process

Using Corey’s data, we can look at every exit attempt a team made in the 2017 and 2018 seasons, about 768 games in total. Additionally, we can look at each outcome: a controlled exit, a failed exit and/or turnover, an icing and an exit without possession.

Last year, at the 2017 RIT Hockey Analytics Conference, Alex Novetpresented some intriguing findings on the value of exiting and entering the zone. One of his key findings was that if the team exiting their zone can do so with possession, they are significantly more likely to attempt a zone entry, which confirms earlier work on a smaller sample of data by Jen Lute Costellafrom several years ago. What was new was how Alex used pressure as a way to quantify just how much impact that can have on the zone exit result. Observe the below slide from Alex’s presentation.



If there is no pressure on the player attempting to exit the zone, they will exit with possession 57% of the time, but only 17% of the time if they are pressured. That’s a huge change. Additionally, the likelihood of a turnover or failed exit increases from 7% to 38%. That sounds like it’s something that should be important, so how can we measure and evaluate it further at the team level?

After investigating Corey’s data, which you can access by donating to his pateron, I wondered how we could use this idea of the controlled exit percentage of the opposing team to measure the forechecking effectiveness of each team in the league. If not letting the opposition exit their zone with possession is advantageous, there should be some signal in a regression. Naturally, we’d want to investigate what impact, if any, the opposition controlled exit percentage (OppCE%) would have on the forechecking team’s shot generation and goal scoring.

The Teams



The best team in Corey’s data set is the Tampa Bay Lightning, whose opponents exited their zone with possession 35% of the time. The worst team is the Calgary Flames at 46%. The Buffalo Sabres, Edmonton Oilers, and New Jersey Devils were next worst at 45%.

While it’s a narrow range, the positive impact of having a more effective forecheck adds up quickly due to teams regularly facing close to 100 zone exits in a game during 5v5 play. So, between the Lightning and Sabres, that 10% difference means that the Lightning opponents are either punting or turning the puck over ten more times per game than the Sabres’ opponents are. That can have a big advantage.

Just how much of an advantage? Let’s look.



What I’ve done is use one each half of Corey’s data to predict the goals scored in the other (two-fold cross-validation due to sample size). Each team had at least thirty games in the regression, so the only ones removed were New Jersey, Vancouver, and Vegas.

We observe a slight negative relationship between the opposition controlled entry percentage and goals for per sixty minutes. The rate at which your opponent exits with possession explains about 10% of your team’s goal-scoring. This is a statistically significant number as the p-value is 0.01. This is what we would expect to see if there is a relationship between forechecking and goal-scoring. We know from the fundamentals of the game that successful forechecking wins possession, forces turnovers, and leads to more offense. Lacking detailed forechecking data, we can use OppCE% as a measure of overall success – did your opponent break out cleanly or not?

Due to the wealth of Corey’s data, there are several metrics we can create and analyze: the percentage of exits that fail/result in turnovers, the total exits per sixty minutes, dump out rates, controlled exit rates – all were analyzed, but the opposition controlled exit percentage had the strongest signal and statistical significance related to predicting goals.

I will say that in a small project last year I collected very detailed forechecking data and there was some signal in how having an aggressive forecheck can be advantageous to a more conservative approach. With even close to 800 games analyzed here in Corey’s data, it’s still less than a full season, so we have to temper our conclusions, but this is the level of detail that should go into team and system evaluation. The more we can quantify each area of the game, the better informed we can be about optimal strategy as well as what players excel at.



As well as predicting goal-scoring, it’s worthwhile to note that the percentage of opposition exits that are controlled is repeatable, meaning that teams largely are good, average, or bad at this consistently.

Additionally, using regression analysis is important as it takes a lot of the guess work out of things. Simply charting data that appears related may or may not be related at all. The regression equation also can be used to provide measurable goals for the coaching staff and players. The equation basically works out to say that for every percentage point we can reduce the opposition’s controlled exit percentage by, we’ll see our goal-scoring rate increase by 0.05 per sixty minutes, simply plug in the OppCE% for x and compute the equation. Therefore, in order to gain an additional tenth of a goal per hour (or one every ten games or so), we’d need to increase our team’s forechecking ability by 2%. Doing so would give a team about eight more goals a season, possibly more depending on their scoring talent. This is an actionable approach using data.

System or Talent?

Some teams do this better than others and with a high rate of repeatability (how well the metric predicts itself), it’s reasonable to assume that teams play a specific forechecking system. Many teams play similar systems on the surface, so is it simply individual talent that’s driving the results here? That’s what I initially thought, but then you do see some teams like Detroit and Montreal around 38%, which is a solid mark and those teams are anywhere near solid in terms of talent. Are there nuances to how some teams forecheck that allow them to excel in this area of the game?

Like many things, it’s likely a bit of both, but certain teams will exhibit better habits in certain areas than others.

When people hear the word forecheck or pressure, they often thing of forwards laying hits on the opposing team’s backs, forcing turnovers and bang-bang plays to score goals. While pressuring the puck carrier is certainly important in order to cut off angles and passing options, teams must pressure at various levels and be disciplined in doing so. Successful forechecking pressure is maintained at various depths in order to maximize its effectiveness. Observe the below situation where the Sabres fail in this aspect.

Most teams have a set play that looks like this when they win a draw in their own end. The back will wheel around the net for protection and look to move the puck up the ice along the far boards to Josh Anderson. The Sabres have both wingers pressure the puck, but fail to cut off the boards. Ryan O’Reilly isn’t in a bad position, but isn’t able to jump onto this pass. Anderson’s first touch isn’t great otherwise he’d be able to pass across the zone and the Jackets would have a 3v2 going into the Sabres end. A little more discipline from the wingers or O’Reilly and this play likely ends earlier with a turnover or David Savard dumping the puck out from the pressure both in his face and up the ice.

In this clip, the Sabres do an excellent job of maintaining pressure both close to and away from the puck. They gradually squeeze the play and force the Blue Jackets to work within a small amount of space. Johan Larsson (22) forces the puck up the wall, where Seth Griffith (25) is there to deny advancing the puck. Griffith follows the play to continue applying pressure while Larsson reloads to maintain a high defensive position, but ready to pressure anything up the near side. When the puck does come that way, he’s quick to force a pass over where Jordan Nolan (17) is there to apply pressure.

Because of the Blue Jackets static play and forwards working back to help out the backs, Jake McCabe (19) is able to step up and nullify this play on the zone exit attempt, forcing a change of possession. The backs play is key here to maintaining pressure – if they leave too large a gap between forwards and backs, then teams will continually get through the zones in transition quite easily.

Let’s look at one more where the Sabres don’t do as good a job throughout the varying levels.

Here the Sabres are pressuring in a 2-1-2, and are in good position early. A quirky bounce results in Savard getting a chance to play the puck up the wall to Cam Atkinson (13). Zemgus Girgensons (28) is in the high position here as F3 and his job is to cut off passes to the center and read the play to apply pressure. While the pass through center is exactly how you often break a 2-1-2, once Savard has his back to the center of the ice and has pressure coming, Girgensons should read that the likely play here is up the wall to Atkinson. He’s late in making this adjustment and Atkinson is able to initiate the exit.

While you can find clips of every team doing things poorly and doing things quite well, the data we have to analyze show the Sabres as one of the worst teams last season at allowing the opposition to exit the zone with possession. Let’s look at the best.

Tampa Bay’s preferred forechecking scheme is 1 -2-2 in that they’ll have F1 angle towards the puck, Ryan Callahan (24), while the two other forwards cover the boards and middle, limiting passing options. Notice the backs slide over to mirror the forwards as well, which squeezes the field of play and allows the Lightning to pick off the errant clear and move back on offense.

Here you’ll see an exchange between the right back and forward. The puck is cleared up the ice and Dan Girardi (5) activates and beats Jonathan Huberdeau (11) to the puck. Vlad Namestikov (90) sees this unfold and circles back to the point to cover for Girardi. This movement creates more possession and nearly works out to great chance for Girardi. Lots of teams preach activating your defense, but it’s likely Tampa Bay is doing it more often than others as they were the best team at denying clean exits in Corey’s data.

So Tampa was the best, but it’s easy to point at the talent on the team as a reason why. Let’s look at some less-talented teams and see if there’s anything in their structure that allows them to turn this phase of the game to their advantage.

After this entry and shot, Ottawa looks to have an easy breakout. Bobby Ryan (9) has plenty of room in front of him and the Red Wings forwards were circling behind the net. However, good forechecking teams will have disciplined pursuit angles. Detroit has this here as their forwards take an angle that will eventually meet with Ryan if he continued carrying the puck. The purpose is to shrink the available field of play for Ottawa. The Senators don’t get much depth or width here so they help Detroit out quite a bit, but Detroit is able to compress Ryan’s options to a small area.

There’s no stretch option as the Red Wings’ backs have taken an aggressive position up ice. Have a look.



This is a second after Ryan makes the ill-advised pass that Detroit intercepts on their pursuit angles. Ottawa didn’t have much in the way of passing options for Ryan and that comes down to the forward pursuit angles, the backs playing aggressive on the forward options for Ottawa, and Martin Frk (42) closing down Ryan.

This is similar to the above play in that the principles of structured aggression are apparent. Ottawa simply has nowhere to go with this puck and opt to punt.

Conclusions

In order to improve team efficiency, data can be used to help answer questions like “How well are we doing at pressuring the opposition?” It also can lead to new questions like “What can we do to improve that aspect of our game?” and “How much impact can this have on us scoring goals?” Analyzing data this way is important because if we were to simply look at that chart of teams above, we wouldn’t how important those small differences can be. Also, teams generate offense in different ways, so some teams stand to gain from this (Buffalo, Edmonton, Calgary) more than others (Tampa Bay, Detroit).

I don’t believe we’ll ever have a definite answer on which system is best, but within each system there are important habits that matter to success. If a team does well at these, their system would likely be successful regardless of which approach it was.

Because of its fluid nature and player substitutions every minute or so, teams will naturally be allowed to exit with possession several times a game. This is why I believe where teams can gain a bigger edge is in the dead-puck situations (faceoffs) and how they pressure following a lost draw. Apart from that, with many teams using similar systems, it’s really down to habits and how disciplined players are related to angles, gap, and pressure. If we ever can get our hands on skating data, I’d wager that’s a big factor in how well players can forecheck.

It is always good to be aware of the limits we have with incomplete data sets. We can’t definitively assign causality to what I’ve done here, but it is an important step to begin to measure each phase of the game as best we can. Some factors that could impact this analysis are team effects. With a larger data set controlling for each team would be more feasible. There is also the the score state to consider and while teams didn’t have much difference across periods, looking at whether they are leading or trailing would be a logical variable to add.

The next step would be to combine all aspects of zone entry defense and offense and zone exit defense and offense to fully understand the impacts of each phase, in addition to the above variables I mentioned. Keep an eye out for that soon!

(Main photo: Ryan Callahan and Chris Kunitz battle for a puck against the Flyers. Credit:

 

meanwhile it's the Panthers 25th anniversary and there are no plans to have a 3rd jersey this season.

 

#Vancouver Canucks what are you doing?

 

guy had a chance to go to what would arguably the best team in the NHL and said no because there might be an ownership change which would make him want to stick around long term on one of the worst rosters in the NHL.

Your homer is showing enjj

 

so he's going to stick around in OTT hoping his buddy can be part of a group that is able to buy the team from a guy who has yet to show any willingness to sell the team and even if the sale does happen, he's gonna spend the last few years of his prime on a garbage roster. Again, your homer is showing.