Last time, I went over how different forechecking styles and aggression levels and the risk/reward nature of each type. In this piece, I'm going to do some similar things with breakout data. Again, all data is from 5v5 situations and from only eleven games that were tracked before Christmas. This should be seen as exploratory analysis and the things we can do with this type of data and the value in continuing this project (over the summer, perhaps?). Kevin and I tracked these games. Muneeb pulled the shot data from NHL play-by-play files.
The Right Amount of Support
Since the first piece went over the value in forechecking with two or three forecheckers below the faceoff dots - how aggressive a team is - the flip side of that is how many players the team with the puck has below the faceoff dots to support the breakout. Situations with one or five players in support had the fewest number of events (only about forty total between them), so they were removed. Non-formation breakouts refer to situations following rebounds, opposition turnovers in your defensive zone, and defensive zone faceoffs.
An interesting pattern emerges that with more support below the dots, the team gives up more shots per breakout. It may seem counterintuitive that while the forechecking team should increase their aggression to generate offense, the team breaking out of the zone should decrease theirs. The reason for this is that in order to bypass a team's forecheck the backs* (or what I feel defensemen should be called) need passing options. On a typical breakout, the center will swing through the house and the forwards will offer support on the wall or as a stretch option, relative to which side of the ice the puck is on.
If the forwards are all low in the zone, there's no threat to the forechecking team and they can squeeze the zone to try and get the puck back. By having one, two, or three forwards at varying depths in the zone, there's several angles and targets for the backs to hit with a pass. Think of it like a quarterback facing the rush and deciding where to go with the ball: he usually has a short, conservative outlet in his fullback to the flat (forward on the wall), a middle tight end crosser a little deeper in behind (center swinging through and picking up speed through the center lane), or a receiver breaking on deep cross or post even further (stretch forward). If all three options are in the same area of the field, the defending team can cover them much easier and the pressure will likely get to the quarterback.
In formation especially, we see that as the number of support players below the faceoff dots increases, it increases the forechecking team's chances of generating a shot. With non-formation plays, it's not as clear-cut, but there is actually a small decrease between two and three players below the dots. The difference is negligible, however.
Similar to how I adjusted for entry attempts in the forechecking piece, I did the same thing here. Basically, this accounts for how often each support level generates a zone entry and weights those by the type of entry (controlled or uncontrolled) based on the expected number of shots derived from Corey Sznajder's work. The formula looks like this:
Shots Against per Breakout Support Level - (shots per entry for* % of breakouts leading to an entry)
The net differential column is negative as we are only taking into account the breakout and subsequent entry, so this is just a slice of the phases within the game. The closer to zero the differential is, the better. We can quickly see that the range of generating offense from the breakout is not all that much. There are greater differences in how effective they are at shot suppression. More is not always better in terms of support below the faceoff dots.
In formation, the more players below the faceoff dots, the worse off you are, given this sample of data. Following rebounds, turnovers, and faceoffs, the data suggests that using two players (i.e. the backs) below the faceoff dots in support is ideal. Using three has the worst differential. It's possible that if not enough players get back to defend or get forward to present as breakout targets, that in-between support neutralizes the team's options.
The Best Breakout Plays
Now that we've quantified what we can expect in terms of shot suppression and entry rates for each level of support, we need to look at which plays are the most optimal. Which ones suppress shots at a better rate? Which ones lead to more entries? I should point out there are a few events that don't happen all that frequently - namely controlled, goalie, and reverse plays in non-formation situations.
Here's the same formula as above, just with play type rather than support level.
And this is how we determine the risk/reward nature of what breakout play to use. In a previous piece, I analyzed breakout plays and how certain ones correlated with breaking out with possession and suppressing shots against. We see similar trends here in the net differential column (remember that closer to zero is better). It's unsurprising that controlled breakouts (i.e. set plays where a back sets up behind the net) are quite effective. They are also quite rare given the volume of other plays.
Similar to my findings in my other piece on breakouts, we see that over, wheel, and reverse plays all do better than goalie, up, and rim plays. You'll see that in the shots against per breakout play column, wheel is tied with goalie plays as the two highest. So, Ryan, how could a wheel play, or even a reverse play which isn't too far behind, be a good play if it allows shots at the same rate as goalie breakout plays? It's because of how often those plays result in entries and shots. That's why we'd want to take into account the net shot differential of each phase, from one end of the ice to another.
Also, unless the goalie is stopping a puck behind the net for their back to make a play, just stay in the net and stop the puck, eh? No need to try anything else.
Here's how the Devils backs fared over these eleven games. Breakout Percentage is simply the percentage of breakouts they were on the ice for that were with possession. Yohann Auvitu led the group in non-formation controlled breakout percentage as well as overall. Andy Greene was a curious case of excelling when the opposition forechecked in formation, but was the worst when facing rebounds, turnovers, or following faceoffs. Small sample quirk, perhaps? Or, are there skills among backs that make them much better for open play situations as opposed to non-formation breakout opportunities?
John Moore was the opposite of Greene: doing quite poorly in formation, but very well in non-formation situations. It would be very interesting to see larger samples of this data with regards to player evaluation. Are there deployment situations that NHL coaches could send in specific personnel groupings that work best together? What is it about certain players that lead to better breakout rates? I think once you isolate and analyze these plays and find hard data you can use to your advantage, you can quickly zero in on how to get the most out of each player in each situation.
If certain breakout and forecheck tactics can be tied to better defensive and offensive outcomes, then those fractions of shot differentials in your favor add up quite a bit over the course of a game and, especially, season. This is how you can "coach" to improve a team's offense and offense without the puck. Tactics certainly matter, but only to the extent you quantify why and how much. It's pretty easy to show a few clips of a play where a goal is scored or allowed and assign a narrative that this is a good or bad play, but unless there's something quantified in that series of play that can be analyzed, there's very little credence that should be given to amorphous analysis full of buzz words.
Certainly, you cannot always attempt the same breakout play as the opposition and circumstance determine part of what you do as a team. However, as we quantify and learn more about these nuances within the game, teams can be smarter about how they want to break out of the zone, and have better alternatives if their first read isn't there. Each player with the puck is a quarterback, and each player without the puck is a receiver, but based on their distance to the puck carrier, they'll have different routes. If the play is designed to be an over play with two players in support but it's not available, players then need to recognize that and go to a "hot route."
At the same time, teams can study how an opponent breaks out of their zone, which backs are more likely to make certain plays, which forwards tend to go lower in the zone to support the breakout, and create forechecking plays off those reads. Hockey is game in which players should have a series of reads/triggers that inform what they should do next. The game is very fluid and often difficult to discern what is trying to be done at various points, but having data like this to back up certain tactical assignments will have an effect over time.
Lastly, understanding and quantifying what leads to breakout success allows you to find value in players and unearth how they drive their shot generation and suppression, so it improves the decision-making process when looking to acquire a player for a specific role. The Devils are in great position ahead of the expansion draft and off-season, but really need some help on the back end to arrive in a hurry if they are to become even an average team again.
Any questions? Comments? Suggestions? Sound off below!
*I go into this in much greater detail in the piece linked alongside it, but the reason I feel this change is warranted is due to the evolving nature of the position. Depending on how a team wishes to set up or play in certain situations, backs should not necessarily always be the conservative defenders we traditionally see. Hockey is a fluid game, but the roles of each position can be quite rigid. Positions should be fluid as well, something I discussed here at #RITHAC last fall. Each position should have multiple roles in each zone and positional flexibility is something that could be vital to improving breakout plays especially. Too often we see defensemen "defending" an empty patch of ice when they could be much more useful in a forward position. If we want more intelligent discussion of the game, we may need to push this ourselves. I'll have much more on this in the near future.*