The Nashville Predators are one of the hottest teams in the league. They earned 21 of a possible 28 points in the month of November. Since the Kyle Turris trade the team is 9-2-1, and has scored 4+ goals in five of those games. The JOFA line is clicking again, and Turris has solidified the second line, flanked by Kevin Fiala and Craig Smith. Combined this line had 22 points in 10 games (before Thursday's game against Vancouver).
As armchair GMs and coaches, how each team plays, whether that’s to the strength of their roster, or to the preferences of the coach is always worthy of discussion. To give you a few examples, the Dallas Stars have slowed down their pace of play this year with new boss Ken Hitchcock, the notoriously defensive minded coach, calling the shots. Mike Babcock has gotten heat for his line combinations in Toronto, the Jets are winning despite terrible possession numbers, and on the flip side, Travis Green has done some good things tactically to give Vancouver a chance to win. Coaching tactics make media chatter at the extremes, especially with dramatic shifts like the post-Sullivan Penguins or why Claude Julien’s teams usually have great possession numbers but can’t score.
Although not perfect, puck possession, which simply measures the number of shots for and against, has been used as a proxy for winning. Intuitively this makes sense. If you consistently out-shoot your opponent, you’re giving your team more opportunities to score, so you should score more goals and win more games. Yet there are teams with good shot metrics like the Los Angeles Kings, the Boston Bruins and Carolina Hurricanes of the last several years who can’t consistently churn out wins.
Conversely there have been some teams with awful metrics that have put together good seasons like the Calgary Flames of the Bob Hartley days, or more notoriously the Lockout Leafs or the Colorado Avalanche of the next season who were “Saved by Roy”. Each of those teams came back to reality the following season once the luck of seasons’ past disappeared. Positive puck possession is a good thing. The next wave of hockey analysis will revolve around how to create positive puck possession. I’ve mentioned some of those ways before with zone entries and exits, as well as the kind of shots/goals a team generates.
Positive team possession numbers are a good thing, but should all teams play uptempo? Is the slow it down style of Ken Hitchcock a bad thing? Well that all depends on the roster.
Vegas establishes betting lines for each NHL game. Outside of the occasional pick-em game, there is a team that is favored to win for a number of factors, usually based simply on the strength of the team. Other factors like how much rest a team has had, who the starting goalie is, or if a team is home or away will shift the line in one direction or another. Hockey is a random game, but after enough games the cream rises to the top, the good teams are good and the bad teams are bad.
Throughout the course of the year some teams will be favored more than others. The Predators would be favored against most of the league right now while the Arizona Coyotes would not. Since hockey is a random sport, we don’t often see drastic differences in the betting line. It’s rare to see a line higher than -200, which implies a 66.6% chance of the favored team winning
In a vacuum, if those two teams faced off against each other 100 times we’d expect to observe the favored team win 66 or 67 of those games. Vegas didn’t build all those fancy hotels because the sports books were losing.
Let’s say the Preds are on the road next month in Tampa. On the road against the team in the league with the best record, Nashville will be underdogs. Let’s presume they are only slight underdogs with the line reading Tampa Bay -120. Vegas is giving Nashville about a 45% chance to win. Now presume we had a weighted coin that had a probability of landing heads 55% of the time and tails 45% of the time. If Tampa represented heads and Nashville represented tails that would give us a general comparison to the expected outcome of a hockey game.
The game of hockey can be considered a series of coin flips. Tactics and strategies are used to gain and advantage over the opponent, but like any other sport inches matter. A puck that rings off the crossbar one night is a goal the next night. Probabilities, even unlikely events like any particular shot resulting in a goal do happen. When those unlikely events occur the dynamics of the game change.
If we treat the coin flips as the likelihood of winning, how we respond to the current game state depends on the score. This is where we get the introduction of score effects. Especially late in games, teams will play differently if they are behind as compared to when they are ahead. In football, teams will execute a two minute drill if they are behind, trying to move the ball up the field quickly before time expires. In basketball a team will willingly trade the points from a made three-pointer for the two free throws you give the other team after a quick foul. The idea in both sports is to make up the score deficit quickly. In hockey we see it at the end of games when the trailing team pulls the goalie. It’s a risky move, but like a power-play team are more likely to score with a numerical advantage in the offensive zone.
Now I’m not advocating for playing an entire hockey game 6 on 5, but what if the more talented team pushed their advantage throughout the game, essentially increasing the amount of coin flips?
In this scenario, if I am the Nashville Predators on the road in Tampa, I know that I will be the losing team in this situation 55 out of 100 times. The longer the game is played the more likely my team is to lose, but how can teams use this to there advantage.
Lets say after the first flip of the game, it comes up tails. Nashville has a 1-0 advantage. I know the coin is weighted however and if flipped enough times I will come out on the losing end. How should Nashville treat the game after taking the early lead? If the rest of the game is a continuous series of equally weighted coin flips, Nashville would like to see the coin flipped as few times as possible for the remainder of the game. If I can limit the game to say 60 flips, I’ll still likely lose, but I’ve given myself the best possible advantage. If I can limit the game even further down to 50 flips, I’ll have an even greater probability of winning. On the other hand, if the favored team falls behind they will want to pick up the pace of the game, essentially increasing the number of coin flips to take advantage of my increased chance of success.
A Hockey Example
In simple teams, the favored team should be trying to push their advantage as soon as the puck is dropped. Don’t play passively, don’t wait for something bad to happen, step on the gas from the get-go. If my roster has more talent than yours, I want to do everything I can to exploit that advantage. I can do that by increasing the number of events in the game. These events of course being shots. Hypothetically speaking if Tampa played Arizona tomorrow, Tampa should aim to skate the Coyotes right out of the building. Press the advantage early, push the pace, and work to get to the lead. Once a sufficient lead is established, you can essentially slow the game back down and bleed out the clock. If I have a less talented roster, I want to limit the on-ice events to as few as possible to give luck more of a chance to work in my favor. I’d be ok with a game consisting of 40 shots.
Is Nashville Actually Good?
The short answer is, not as good as we think they are, at least at even strength. The below chart is score and venue adjusted Team Pace, which means that once we account for score effects and where the game is being played, how exactly are each of these teams playing if we eliminate as many variables as possible? Nashville is straddling the line between the “Good” and “Dull” quadrants. They are almost exactly at league average for Corsi For, and just slightly better than league average in Corsi Against.
If we filter out all blocked shots and just used unblocked shot attempts, The Predators drift just slightly to the “Dull” quadrant. Dull doesn’t mean bad, just that they are playing lower event hockey than the league average. Dull hockey can be traced in one way or another to the roster or tactics employed.
Nashville has been a perfectly average team in the territorial battle at 5 on 5 so far this year. The driver behind the Turris trade was the inability to generate any offense beyond the JOFA line.
Simple shot attempts don’t tell the whole story because some teams do generate higher quality looks than other teams with the same exact shot totals. If we look at expected goals, again a not perfect, but explanatory model, we see that Nashville is one of the worst teams in the league at generating offense at 5 on 5. The expected goal model built essentially on shot totals and shot distance, has Nashville pegged at 1.94 xG/60, just a tick above both Arizona and Vancouver at 1.93, not exactly company the Predators want to keep. On the defensive side, like their possession totals, the Predators are right at league average in expected goals against at 2.25/60
It’s been better since the Turris trade, right? Well it has, but it’s not been a seismic difference. Nashville has improved their xGF numbers from 1.94 to to 2.04. Last year on average teams spent 3831 minutes at 5 on 5. So a one-tenth of a goal difference per 60 minutes would equal about 6.4 goals more, or two standings points difference by the end of the year. Expected goals against has fallen from 2.25 to 2.20, an even smaller difference than the goals for change. Either way, the Preds have played better at 5 on 5 since the addition of Turris, but they aren’t some completely new team.
Why are Nashville’s 5 on 5 expected results worse comparatively than the possession totals? Well there are a few conjectures I can make. First I wanted to look at shot quality as measured by the team-level expected goals per unblocked shot attempt. Graphed below, Nashville is in a poor location with an overall shot quality of 5.3%, tied with St. Louis and Buffalo and ahead of only Colorado.
What the expected goal metric doesn’t measure however is the amount of pre-shot movement that occurs in the offensive zone. It’s been established that shots that occur after a pass in the offensive zone are more likely to result in goals, especially if that pass originated from below the goal line or across the royal road. Since NHL play-by-play data doesn’t include that information, we are left to use other methods to try to assess shot quality.
One simple way of doing so is measuring how often the Predators shoot from each location on the ice compared to the league average. HockeyViz does this with heat maps. Shown below, the Predators heat map shows exactly why team expected goal numbers are so low. There is an ice-cold blue mess right in the slot area. On the whole, the Predators are not getting to the high-danger areas of the ice. The only place on the ice where Nashville is significantly out-shooting the league average is from the right circle. If Nashville fails to develop any real slot presence in front of the opponent’s net, it’s going to be a problem.
Corsica doesn’t track scoring chances anymore, instead they’ve opted for expected goals. Natural Stat Trick still tallies them though. Their numerical results are in line with the Hockey Viz graphic. On the season Nashville has been downright abysmal on SCF%. Adjusted for score and venue, they have the 9th worst total in the league.
Filtering out the games since Turris, the results are even a touch worse, slotting the team in at 8th worst, and about one percentage point worse than they were before Turris.
But they are fine on defense right? Well they’ve been perfectly average there too, allowing more than the league average shots right at the goal crease, higher than average in the right circle (when looking out from the goalie), a below league average total in the low slot, and slightly higher in the high slot.
The Predators haven’t been tremendous when compared to the league average in how they’ve played in front of the goalie. Rinne is on the chart below just slightly in the “Friday Afternoon” quadrant. He has an expected save percentage of 92.4% compared to the league average of 92.2%, and he faces 30.36 shots against to the league’s 30.48. Again all these numbers are adjusted for score and venue, giving another indication of Nashville’s totally pedestrian yet sufficient defensive play.
My “Goalie Heart Rate” chart measures the percentage of shots against that are considered high-danger shots. Rinne is in the “Lance Armstrong in his Prime” quadrant with 17.8% of shots against being high-danger with a league average of 18.9%. This measured difference can likely be traced almost entirely to his ability to control rebounds. Saros is at 22.5% and in the “Intermittent Panic” quadrant on this same chart. I discussed Rinne’s prowess briefly before the season, and my friend Cole Anderson took a deep analytical whack at it with this tremendous piece just a few days ago.
Rinne is saving the team about one goal against for every 180 minutes of even strength ice time above the league average.
This will be fine until he hits about start 55, where his performance has historically declined, outside of last year’s stretch run where he was tremendous. Saros will need to win some games and ease the burden on Rinne sometime this season.
How Should They Play?
To go back to the Vegas analogy, the Preds are going to be favored likely for every home game the rest of the season and even some games on the road. Their perceived advantage has been limited to special teams where they have the 5th best GF/60 on the power-play and 7th best GA/60 on the penalty kill. Pekka Rinne has been playing above expectations so far. However their adjusted 5 on 5 goals +/- stands at -0.98. Even with the addition of Kyle Turris, they’ve continued to play like a league average team 5 on 5. On the whole they should work to play fast against teams with a more pronounced negative expected goals differential than they have, and look to slow the game down against teams that have a positive expected goal differential. Sadly the list of teams they should play fast against isn’t that long.
If we look at the team at the individual level, there are really only two lines who are out chancing their opponents. When we adjust for score and venue the JOFA line and the newly formed Fiala-Turris-Smith line are the ones driving positive possession and scoring chance numbers. Ironically, Miika Salomaki looks good here too in limited time. You could easily make the case he should see some time on the 3rd line to see if some combination of he, Bonino and Jarnkrok can give the Preds what we could consider a top nine.
Analytics folks get annoyed when teams dump and chase, but if we look at a dump as a play that will at least burn a few seconds off the block before the puck comes back out in the other direction, we are putting our less skilled players in a position to not give up a chance in their own end. We are limiting the number of events in the game. If the 4th liners are out there against a skilled line from the opposing team, a dump in is likely the correct choice. Slow the game down and let the more offensively skilled players catch their wind on the bench.
For the Predators they should do everything they can to grind the game down to a standstill while the 3rd and 4th line are on the ice. The faster those lines play the more likely they are to succumb to their scoring deficits. The JOFA line should be given free reign to create at will and the 2nd line, should look to create offensive with creative passing plays. The talent on that line is there although they aren’t significantly outplaying their opponents.
Will this improve when Ryan Ellis returns? The answer is probably. The pending Ellis return will let Coach Laviolette move Emelin/Irwin down to the 3rd pair. I’m sure I like Matt Irwin as a player more than you do, but the data shows him being an effective player, even if it’s in a 3rd pairing role. He tilts the ice in Nashville’s favor while playing low event hockey.
Will David Poile make a move at the deadline to add another scoring winger? He probably should. Unless someone can develop along Bonino’s wing, we’ve seen in recent playoff runs that the teams that can run an effective three scoring lines give themselves a chance to win when it matters; and that’s the ultimate goal of this team.