clock menu more-arrow no yes mobile

Filed under:

New York Islanders Shift-start Project: An introduction

NHL shift data provides info on shifts begun at the dot...but what about changes on the fly?

Pittsburgh Penguins v New York Islanders
On the hunt...but in a positionally sound way
Photo by Mike Stobe/NHLI via Getty Images

The stats from this article do not include shifts from the Islanders’ memorable comeback win in Philadelphia Saturday night.

This season I have decided to track scoring chances at 5v5 “close*” in relation to the shift-starts of New York Islanders centers.

For years I have been wondering how much on-the-fly starts affect the fortunes of NHL lines and defensive pairs. It is great that public data keeps track of faceoff starts, but (a) self-inflicted defensive-zone faceoffs (and earned attack-zone starts) are typically lumped in with inherited faceoffs and (b) many shifts in the NHL start on-the-fly, with the play in various situations.

I feel that faceoffs only tell part of the story.

For this study I am dividing on-the-fly shift-starts into four different categories, along with the three categories of faceoff starts (defensive zone, neutral zone, and attack zone).

* ”Close” means the score is within one goal in the 1st or 2nd period, or tied in the 3rd.

Types of On-the-fly Starts

Dump-In: When the opponent has possession of the puck in their own zone and they are not pressured, as a new NYI center enters play from the bench, it is categorized as a “dump-in”. Typically there is no more than one player forechecking. The Islanders have four or five players in defensive position, preventing an easy transition-with-possession by the opponent.

Retrieval: When the Islanders have possession of the puck in their own zone and they are not pressured, as a new NYI center enters play from the bench, it is a “retrieval”. This is basically the inverse of the dump-in. Generally both teams have an opportunity to change players.

Disadvantage: When the puck is in the defensive zone (or entering the zone) while the NYI center steps into the play from the bench, with no opportunity to exit the zone without pressure, this goes down as a “disadvantage”. These starts are somewhat rare, because centers tend not to go to the bench when the team is under pressure.

Advantage: When the puck is in the attacking zone and the opposing team has no opportunity to exit the zone without pressure, this goes down as an “advantage”. This is not quite the inverse of a “disadvantage” start, because the opposing team’s center (and other players) are often tired, whereas for “disadvantage” the NYI center is fresh off the bench.

Here is an example of an on-the-fly advantage from the recent game against the Maple Leafs, courtesy of our own Spizzwolf. As you can see, Casey Cizikas, with the arrow over his head, enters play from the bench. Thanks to a timely pinch by Mayfield, Johnston is receiving the puck along the left wing boards as Cizikas joins the play.

Even if it was merely a 50/50 battle for the puck as Cizikas begins to impact the play, it would have been categorized as an on-the-fly advantage. The key is that the Leafs do not have an easy-out.

Scoring Chances

I will explain what constitutes a “scoring chance” in the FAQs at the bottom, as a lot of criteria go into that. If a shot is taken from the faceoff dot (or more advantageous location) and it gets through to/past the goalie unblocked, it most often goes down as a scoring chance.

For this study I am only concerned with the initial scoring chance from each shift’s start. I’ll explain more in the FAQs, but with the way the Islanders play 5v5, there is rarely an immediate counter-attack for or against, after the initial scoring chance.

Team Performance

We can see below that while the Islanders have been out-chanced from defensive-zone faceoffs, neutral-zone faceoffs, and dump-ins, they have out-chanced opponents via shifts stemming from attack-zone draws, retrievals, and most definitely on-the-fly advantage starts (36 to 12).

Are these numbers good?

As far as individual categories, I have little idea how good/bad these stats are. I don’t believe there is a publicly available baseline for any of these on-the-fly categories. We can safely say that (a) losing scoring chances 28 to 18 from neutral-zone faceoffs is not a good start to the season and (b) leading in on-the-fly chances 86 to 66 looks like Isles are doing something right.

We can hopefully have an idea of where centers may be struggling by comparing to the other NYI centers, at this point, but we are definitely working with limited sample-sizes.

Note that: (1) for faceoff starts, the opposing center is never on the ice longer than the NYI center and (2) for on-the-fly starts the NYI center more often has the fresher legs— but not always.

For today we’ll focus largely on the “on-the-fly advantage” category and highlight Casey Cizikas (and his line), after going over each individual center’s on-ice stats.

Mathew Barzal’s line has been out-chanced 25 to 18 from faceoff starts, but the on-the-fly numbers are strong, as we see below.

Brock Nelson’s line has been the opposite: solid off of faceoffs, but out-chanced 15-to-8 for dump-in and retrieval combo when Nelson started his shifts on-the-fly.

Derick Brassard’s line (when he has played center) has not had an initial scoring chance at 5v5 (with score close) so far this season stemming from defensive zone or neutral zone draws, while they have allowed 10, combined. The on-the-fly stats look fine, in the small sample, at 10-to-8.

Cole Bardreau’s line fared well by these stats, out-chancing opponents 11-to-8, in limited time.

Casey Cizikas’s line has been dominant from faceoff starts (10-to-5, with a lot of defensive-zone draws), but they have also been a remarkable 11-to-0 when starting with an on-the-fly advantage, utilizing their forechecking abilities.

During the team’s 10-game win streak (six for Casey, since he was injured the first four), the Islanders out-chanced opponents 23-to-5 on Cizikas’s shifts. The majority of those six games his wingers were Michael Dal Colle and Cal Clutterbuck, though he also had Martin on his left wing part of the six games.

Has Cizikas faced inferior opponents, compared to Barzal and Nelson?

Some may assume that Cizikas’s line faces primarily third and fourth-line opponents, but this has not been the case. Here is a table of the most common 5v5 opponents for Isles centers for each game this season. While Nelson’s line has drawn many of the power-on-power matchups at home, Cizikas’s most common competition appears comparable in the past seven games. (Note: this is all 5v5 TOI, not just “close” shifts.)

Of the 23 shifts at 5v5 “close” to begin with an on-the-fly advantage, Cizikas’s line had the initial scoring chance 11 times (48%), while they haven’t allowed the initial chance on any of the other 12 shifts.

Barzal’s line is the second most-impressive by this stat, of the four most used centers, with 14 chances (38%) for and five (14%) against, in 37 opportunities.

Of Cizikas’s 101 starts following a shift by another NYI center, his line began with an on-the-fly advantage 23 times. It has been a mix of centers who he has followed, so I find it interesting that his rate is substantially higher than that of Nelson or Brassard. (Chart below.)

This seems to indicate that it may be Cizikas’s line influencing the type of on-the-fly situation, as they step onto the ice. (Often a winger or two will step on before Cizikas.)

As far as a center’s line “creating” on-the-fly advantage for the oncoming center, the four pivots span from 15 percent (Cizikas and Barzal) to 18 percent (Nelson), which is much narrower range than the chart above. Often the defensemen play a big part in creating an on-the-fly advantage for the line entering play, as we saw in the GIF, with Mayfield protecting the puck and then swinging it around the boards to Johnston.

It will be interesting to see with more data whether Cizikas’s line continues to begin their shifts with the OTF advantage more often than other NYI lines— and whether they can continue to dominate shifts that begin with the forechecking opportunity.

Next time perhaps we’ll take a look at comparing dump-ins with retrievals: the battle of possession vs territory.



1. What counts as a scoring chance?

Shot location and whether the shot is blocked are the two main factors. The faceoff dots mark the sharpest angle for a typical scoring chance. A blocked shot can only be counted as a scoring chance if (a) the shot would otherwise be counted and (b) the blocker of the shot is closer to the goal than he is to the original location of the shot.... Shot velocity factors in as well as if there is quick lateral movement of the puck, such as from a cross-ice pass, causing the goalie to scramble.... A shot from the point can be counted if (a) the goalie is visibly screened— often leaning on one leg with head diagonal, as long as other criteria apply. But generally hard, unscreened shots from distance are not counted.... Shot velocity (visual) does factor in, especially for shots that are otherwise borderline.... Deflections from good angles are counted as scoring chances, as long as they meet the other criteria.... A goal doesn’t necessarily count as a scoring chance. Toews’s triple-deflection goal is an example.

2. Why only count the initial scoring chance on shifts?

I am mostly interested in the advantage / disadvantage stemming from different types of shift starts. For NYI shifts it is somewhat rare for there to be more than one scoring chance for or against (due to style of play, under Barry Trotz). It is most often for the same team as the original chance.... When there is a quick counter-attack chance it typically is due to a defenseman overcommitting. (I am taking notes for any counter-attack chances.)

3. Don’t the defensemen / wingers matter?

They absolutely do matter. I am focussing on the NYI centers because I have a hard time focussing on more than one thing at a time.... The study is as much about the line as it is the center, but wingers change from game-to-game and often shift-to-shift. It would be a headache to try to categorize on-the-fly shift-starts for entire lines, because players leave + enter at different times, while the puck and players continue to move.

4. Are the Islanders a good team for this study?

I think they are a great team for this study because (a) their more deliberate style of 5v5 hockey makes it easier to categorize on-the-fly shift-starts than I imagine it would be with other teams and (b) all three defensive pairs are quality this season. Pelech-Pulock is the best pair, but the other defensemen are very capable, so I don’t have to worry about adjusting for how much each center is playing with each defensive pair, at least over the course of a full season. (It is mixed quite a bit, anyway.)... Pelech-Pulock generally face tougher competition, so that also helps to balance out performance among defensive pairs, as it does for most teams.

5. Why limit the study to “Close” shifts?

Even if coaches do not dictate a different style-of-play, the skaters themselves play the game differently when the score becomes lopsided. I am interested in looking for patterns for “regular” 5v5 play.

6. How long does the shift extend?

The shift continues until the center goes to the bench. This may be through several faceoffs, but it still stems back to the initial start category. (So a center lives with the “self-inflicted” defensive-zone faceoffs and “earned” attack-zone faceoffs if he remains on the ice.)

7. Why do you say “attack zone” instead of “offensive zone”.

The abreviation “OZ” would look too much like “DZ” in the spreadsheet, and I want to be consistent.

8. Wait, didn’t you start this project last year?

Yes, but I realized a few games through that I didn’t have the time to devote to it, due to other endeavors. There are only so many hockey-hours in the week. This season I believe I can stay on-pace throughout the season.

9. What happens when a center steps onto the ice, but does not have a chance to enter the play before a scoring chance?

In that case it does not affect the stats of the entering center, if he does not have an opportunity to affect the play. The scoring chance “for” or “against” would count for the exiting center.... The categorization of the shift-start occurs when the center enters play, not when he steps onto the ice from the bench.