Last time, I discussed how we can use breakout data to explain how a team might tactically alter their breakout plays in order to generate more controlled exits. Today, I'm going to go over how we can quantify the expected shot rate based on how many aggressive forecheckers there were (number of players below the faceoff dots) as well as the type of forecheck (2-1-2 or 1-2-2). This would not have been possible without the data assistance of Muneeb Alam, who managed to extract certain data I wanted from the NHL play-by-play files. Thanks to Kevin for helping track games with me last fall. All this data is from 5v5 situations from games earlier this season.
Before I get to the data, there's some good news and bad news. The good news is that it's very cool information that will be able to support tactical decisions regarding how aggressive teams should forecheck. The bad news is that this project died around Christmas, so I'm still using those same eleven New Jersey Devils games. However, considering how many of these events there are in a game, we still have a little over 1000 events to analyze and look for patterns. It's unfortunate there weren't many volunteers, but hopefully that will change going forward. Getting involved in projects like this is what furthers how we can analyze and discuss hockey.
Finally, all data is from 5v5 situations, and this is focused more on the descriptive analysis of the game. There isn't enough data to conduct predictive analysis, but in looking at tactics, I do feel there's value in isolating phases of the game and quantifying shot values in these different phases.
Quantifying Aggression Levels of a Forecheck
Back to the lecture at hand. The first thing we'd want to do is measure the shots per forecheck at each aggression level. This is identical to how Eric Tulsky, Geoffrey Detweiler, Robert Spencer, and Corey Sznajder quantified the value of controlled and uncontrolled entries several years ago. I'll present the data as the total rates, formation rates (forecheck was in a defined 2-1-2, 1-2-2, 2-3, etc), and non-formation rates (rebounds/blocked/missed shots, lost offensive zone faceoffs, and turnover situations in which a team pressures to get the puck back). Let's have a look.
Unsurprisingly, we see that as the number of forecheckers rise, the greater number of shots are generated per forecheck. There's a much higher rate following rebounds, turnovers, and faceoffs, and I believe this owes to the fact that it's more difficult to interrupt and generate offense during a traditional forechecking event (in formation) due to a team's defensemen not up the ice in the offensive zone that often. In fact, pressuring aggressively off of puck battles in non-formation situations yields the same expected shot rates as controlled entries.
If you listen closely, you can hear conservative coaches shouting that while you may generate more shots with an aggressive forecheck, you may give up chances the other way and so it's not worth it in the long run. Well, I'm here to tell you that that is possibly wrong.
How can we be sure of this? Well, through our tracking, we recorded each exit and entry, if there was one, attempt by the team breaking out of their zone. Since we know the shot values of controlled and uncontrolled entries (0.66 and 0.29 respectively from Corey Sznajder's 2013 - 2014 season of data), we can create an expected shots against for each forecheck where an entry was yielded.
It's worth pointing out that while a dump in generates 0.29 shots per entry, you can see the differences based on how aggressive the team is in going after that dump in. Being aggressive in going after pucks following rebounds, lost faceoffs, and offensive zone turnovers generates almost as many shots as entering the zone with possession. I think that's important to note for the simple fact that if coaches wish to encourage players to enter with possession or create a tactical approach in transition to increase the likelihood of a controlled entry, they can now encourage their players to be aggressive in those situations and design where the players should go, based on where those events occur in the zone.
Now, since there isn't an entry for each forechecking sequence, we must properly measure the impact. Overall, there are entries yielded on forechecking events only 22% of the time. This may seem low, but there are two points of emphasis here: 1) We have a strict definition of a forecheck; and 2) Think of all the other stoppages that can occur during a game - goal, penalty, save, offsides, icing - in addition to dump outs that lead to another entry by the team originally on the forecheck.
When we break it down by aggression levels, here are the expected shot rates for entries against as well as how often each aggression level yields an entry.
A pattern does emerge of higher shots per entry as aggression level increases, as well as a lower percentage of entries attempted by your opposition. So, it is fair to say, on average, if you are more aggressive overall, you may give up a few more shots than if you had been conservative or balanced; however, it is also fair to say that you give up entries less often if you are aggressive.
With these different rates of shots against and proportion of forechecks that yield entries against, how are we to properly weight each component here? We can create a formula to arrive at the net shot differential that incorporates the aggression level of each forecheck, as well as the risk associated with how often entries are given up. It looks like this:
Shots per forecheck aggression level - (shots per entry allowed * % of forechecks yielding an entry)
As you can see, the differences in the risk associated with giving up tons of chances the other way is overblown, given this small sample of games. There is no attempt to adjust for the quality of chances given up the other way (though there were only eight recorded odd-man situations against the forechecking team out of 1018 events, which equates to slightly less than 1% and is below our passing data rate of 2% of all shot sequences being an odd man situation, but makes sense given we are not tracking all passing and shooting situations here), but I do think this at least forces us to consider championing being more aggressive with a team's forecheck.
2-1-2 or 1-2-2?
So there are choices a coach can make with how aggressive to be on the forecheck, but there is also the choice of what formation to use?
Both types of forechecks yield the same about of shots. There's a slight increase in shot generation from a 2-1-2 than a 1-2-2, based on this data. Not a huge difference between the two, but with how many of these events there are per game, that difference does add up. Say a team forechecks in formation 100 times over a stretch of games (likely three based on the data). In a 2-1-2, on average, that team would generate thirty-seven shots. If they used a 1-2-2, they would generate twenty-eight. That might only be a difference of three more shots each game from their forecheck, but the zone time and possession required to generate those shots might lead to a few power play chances or offensive zone faceoffs, which would then yield more shots. It's a small piece of the puzzle. In fact, I think the forecheck and zone entry work are the building blocks of any good offense, so both deserve to be quantified.
When to Pressure Loose Pucks?
In addition to formation forechecks, there are many situations during a game where puck battles occur and teams apply pressure to varying degrees. Are there certain situations teams should be more aggressive than others? Is it more valuable to pressure after a faceoff? A rebound? After you turn the puck over?
Based on this data, if a team loses a faceoff or turns the puck over, they are still incentivized to go after that puck aggressively. However, while a team may err on the side of caution in these situations, teams absolutely should go after rebounds, missed, and blocked shots as these are highly important moments to sustained offensive pressure. The shots per forecheck on rebounds actually exceeds that of a controlled entry. So, I'd love to get more data to further confirm this, but tactically, teams should be going all-out on rebounds to pressure the opposition. It makes much more sense to change on faceoffs and offensive zone turnovers. That's another point here - quantifying each phase of the game's impact on shot differentials can lead to optimal times to change lines. Playing efficiently means taking advantage of all phases.
How Aggressive Were the Devils
I'm going to split the formation and non-formation forechecks to analyze how aggressive and successful the Devils were against their opponents. I believe it's right to look at these phases separately. First, we'll look at how the Devils and their opposition each performed while forechecking in formation.
The Devils pressured with three aggressive forecheckers in formation about 6% more often than their opponents. Additionally, the Devils were far more effective than their opponents at generating shots with three aggressive forecheckers (0.51 shots per forecheck compared to 0.29). When they are this successfull, one wonders why you would only spend 22.9% of your time using three aggressive forecheckers? Certainly circumstance (line changes, if someone is late getting into the zone, etc) dictates some of that, but the coach plays a big role in something like this. Granted, a team doesn't want to become predictable in how they forecheck, but certainly the Devils could become more aggressive.
The opposition pressured about that same 6% difference with two aggressive forecheckers, 59.9% to the Devils 53.1%. This was the preferred (through some combination of intent and circumstance undoubtedly) method of forechecking pressure for both teams, but more heavily used by the opposition. As a result, the team's total net shot differential per forecheck closely mirrors their more balanced rates.
Let's now look at how both teams fared in non-formation pressure situations.
Here we see a decided advantage towards the opposition. Whether they pressured the Devils with one, two, or three aggressive forecheckers didn't matter - the opposition were more effective generating shots from these situations, most notably when only pressuring with two aggressive forecheckers. This raises a few reasons as to why this could be happening: 1) Lack of talented defensemen or 2) Poor coverage/breakout tactics. The talent on the back end is likely the main culprit, but if "off the glass and out" is something that is preached or a move the defensemen go to far too often (looking at Corey's zone exit data so far this season on the Devils, this means you Ben Lovejoy), that should be discouraged whenever possible.
Which Devils were the Best?
Here were the forwards (since forechecking is traditionally about forward involvement) with the most time-on-ice in this sample. The columns indicate the shots per forecheck each forward generated at each aggression level, as well as their total rate, and how they related to the rest of the team. Taylor Hall is good.
Tactical Takeaways / Conclusions
You can cut off a team's ability to transition out of their own end by applying an aggressive forecheck. This has a positive effect on your team's shot generation and does not necessarily mean you will be unable to defend your blue line. Reducing the number of entry attempts is paramount to how you defend at your blue line. Teams can generate shots from entering your zone - they can't generate as many shots if they enter your zone less often.
Also, this should be a first step towards better understanding and analyzing team tactics regarding zone entries. It has long been held, and with good evidence in favor of it for sure, that entering the zone with possession was better than dumping the puck in. I believe it is still preferable, but depending on the aggressiveness of the forecheck going after that puck, can still produce offense higher than the standard dump-in rate of 0.29 expected shots (0.41 shots per forecheck with three aggressive forechecks compared to 0.66 per controlled entry).
Furthermore, pressuring after rebounds, turnovers, and lost faceoffs, the reward for being aggressive is at its highest. Specifically, aggressively going after rebounds leads to the highest number of shots generated, even exceeding controlled entries. Depending on where a shot goes in the zone, a team should have a plan in place so each player knows where to go and what their responsibilities are and how they change.
While this represents a small number of games, my hope is that others will continue working on this project. There is now a template (courtesy of Ingrid and Alex) to capture this type of data each game as it's tracked. Similar to zone entries, there is a lot of information and knowledge to obtain from forechecking and breakout phases of the game.
Despite the small sample, I think what we see here makes intuitive sense: being aggressive is the best way to prevent shots and entries against, and the conservative approach is likely erroneously influenced by the few times a pinch goes bad or and odd-man rush goes against you - those are the events that stick in our minds.
Traditionally, teams will have sent one, two, or three forecheckers into the zone and kept their defensemen at the blue line. Occasionally some will pinch, but this is often seen, again, as a risky play. Of the 127 forechecking events with a defenseman jumping into the play to pinch or keep the attack going, only 12.7% of these resulted in an entry against the forechecking team. So, again, it's likely those few odd man rushes that you remember rather than anything else. We all have our own biases that cloud our analysis, no matter how much video we watch. Data helps cut through that.
The role of the defensemen will continue to change as it has since the 2005 lockout. I've written about it a bit recently. Teams could be far more aggressive on the forecheck and use their defensemen to maintain pressure while a forward reloads (climbs higher in the zone before applying pressure again), so there is never a lull in the pressure exerted upon the team trying to exit the zone. Confused? Let me show you a quick example of what I'm talking about.
In the Devils opener at Florida, we see them pressuring the puck aggressively at the beginning of the game. The Panthers' Jason Demers (55), ideally, would take a few strides and then reverse this play. If the defenseman can lose Mike Cammalleri, he can skate it up, or have a 2v1 passing option with Kyle Palmieri expected to mark both Michael Matheson (19) and Vincent Trocheck (21) of the Panthers.
Demers opts to send it up the boards towards Reilly Smith (18). Andy Greene steps up to keep possession. Palmieri here will cover for Greene while Travis Zajac reloads to attack again.
Greene will carry the puck to the corner, but a team doesn't want to play in a crowded area of the ice - they should want to switch to the open ice. This is especially true against aggressive overload defenses like the Panthers are using here. Zajac knows Greene is about to switch play and heads to the release point with a step on the defense. The Devils would release the point behind the net after this and get a good option in front thwarted by the Panthers at the last moment.
This is a quick example of why playing with four attackers in the offensive zone is critical.
Through data like this we can analyze the small pieces of the game that lead up to more offense over the long haul. This is why we use shots rather than goals to predict the future. Shots are good. If we can use forechecking and entry data to quantify tactical approaches to creating more shots, that's a good use of a video analyst's time and a team's money. Using passing data can then take that even further by quantifying and testing what types of shots and sequences are most likely to turn those shots into goals. Everything flows into the other.
I'll have a companion piece on Monday that details the reverse side of this: how many players should a team have in support (below the faceoff dots) on their breakout? Sound off below with questions, statements, concerns, jokes, or any other words of your choosing.