Better Together: An Analytical Look at the JOFA Line - Part II
The trio of Ryan Johansen, Filip Forsberg, and Viktor Arvidsson have faced some ups and downs this season as injuries have reared their ugly heads. Here’s an in-depth analytical look at the numbers behind their performance.
[In the first half of this, OTF staffer Shaun Smith had asked up-and-coming analytics contributor Bryan Bastin to chat about everyone’s favorite subject, the JOFA line—more specifically about what was working, what wasn’t, and what to do about it.]
Continuing from our last installment, we had just seen Ryan Johansen going it alone while both Filip Forsberg and Viktor Arvidsson were recovering from injuries. As each player returned, would they be able to find their success once again?
The Band Gets Back Together (January)
Arvidsson returned to the lineup December 27th, and he was joined by Filip Forsberg on the 7th of January. The JOFA line was reunited; however, other issues began to arise—Kyle Turris left the lineup for over a month and the overachieving line of Watson–Bonino–Sissons slowed to a snail’s pace production-wise even before Watson’s eventual re-suspension. While Freddy Gaudreau and Rocco Grimaldi were providing stability on the fourth line, the question still remained: could the JOFA line carry the team as they did in October?
The top line came out firing, figuratively and literally. The JOFA line had a Corsi differential/60 of +36.93, triple the previous month’s. However, the rest of the team did not follow suit, giving up nearly 8 more shots than they took. Perhaps more worrying was in the instances where the line was incomplete, whether before Forsberg’s return on the 7th or during a two-game stretch where the lines were broken up and posted a -13 differential. When it came to shooting, the JOFA line was doing all the heavy lifting, but only when all together.
When it came to quality of shots, the story was mostly the same: JOFA produced an incredible +2.99 xG/60 differential. However, all other combinations had a deficit, though of less than a single goal. The top line was continuing to bear the load, shooting high-quality shots and lots of them, yet the results were unexpected.
The month was filled with both lopsided victories and defeats, as well as three OT games (of which Nashville only won one). While the rest of the team managed a surprising +1.73 goals/60 differential, the JOFA line had a slight deficit, at -0.84 goals/60, and when JOFA was split up, they were dead even.
The top line knew they had to drive the Predators’ success, and did so by playing a lot of high-risk, high-reward hockey that—while impressive on the stat sheet—resulted in some blowouts on both sides. The difference was made up by some depth. The wins were coming, despite seriously depressed goaltending due to inconsistent efforts from both Rinne and Saros. Yet the team came out of the month with a 7-3-2 record, pulling them out of the roughly 0.500 winning percentage from the previous two months. If they could follow the same gameplan, would the results be the same?
Playoff Push — Inconsistency and Lines Split Apart (February to Mid-March)
(12-10-2 through March 23rd)
They followed the same gameplan, and some odd things happened: the non-JOFA lines stepped up to provide improved Corsi rates, from a nearly -9 in January, to a +4.8 in February and a +5.43 in the first half of March—about half of what JOFA was producing in October. With this came an slight improvement in quality, as the team ended February with a slightly positive xG differential, later offset by a -0.32 xGdiff in March after the trade deadline.
Speaking of the trade deadline, the lineups were shuffled quite a bit both in the lead-up to and directly after the deadline. Kevin Fiala and others all saw time on the first line, and in these two months, about thirty percent of the team’s time on ice was spent with one member of the JOFA line separated and playing somewhere else. This was a very, very bad decision:
The coaching staff decided to ice a broken-up JOFA line for eight games in February and March. Three of those never saw the JOFA line together at any point at all. In February alone, there was a difference of 25 Corsi shots, 2.1 expected goals, and 6.4 actual goals that could have been gained by keeping the line together, rather than a 70/30 split.
Yet in March, two entire games were played in which the JOFA line saw no time whatsoever—against Carolina and Anaheim, two bad losses both on paper and on the ice. Over the first ten games of March, TOI was split 70/30 again between JOFA and JOFA minus one, and the results were still bad: 31 Corsi, 0.6 expected goals, and 2.7 goals scored.
Would that pace be enough to take both of those games? Likely not against Carolina, but definitely in what should’ve been a gimmie against the Ducks.
Still, even though the JOFA line intact was clearly a better option, the numbers weren’t up to the standards of their explosive months of October and January. In fact, the team continued to play almost exactly 0.500 hockey with a record of 12-10-2. While both JOFA and the non-JOFA lines managed a positive Corsi differential over this time period, it wasn’t near the standards set earlier in the year.
In fact, while in January the JOFA line was carrying the rest of the team’s negative differential, only a couple months later, the JOFA line was producing less than the rest of the team (3.10 vs 5.43 Corsi shots per hour). In a similar vein, quality was exactly the same, with both groups managing a 0.07 xG/60, about the value of one or two decent shots.
The scoreboard reflected this trend: JOFA had a +0.86 xG/60 in February, but regressed to dead even in March; the rest of the team went from a paltry +0.12 xG/60 to -0.32 xG/60. Fpr the last five games of March, the line was kept mostly together (barring a stretch against San Jose in which Arvidsson got hurt), but they couldn’t keep up the pace they saw in October and January—JOFA was struggling, as was the team.
The Season So Far: A Closer Look at PDO
As noted before, the two components of PDO are shooting percentage (goals / shots on goal) and save percentage (saves / shots on goal). The rolling average graph above has the save % in blue; this is shots saved while the entire JOFA line was on the ice. The first 12 bars represent the 12 games in October that the JOFA line played together—the goaltending was solid at 93.9%, and the average PDO for the month was at a blistering 110, meaning that the JOFA line or the defenders they were sharing the ice with were scoring a goal on 15% of the shots they were taking on 5 on 5.
As mentioned previously, PDO league-wide averages between 102 and 98, so to have a 110 means that the shooting percentage was unsustainable—even the best teams regress to no more than around 11.5%. The months of November and December are not shown aside from the JOFA line’s only game together (November 10th), which is the 13th bar on the chart.
The next seven were the games in January after Arvidsson and Forsberg both came off injured reserve. January was another high-performance month for the line and the team; however, this time it came during a stretch of bad goaltending. The average over that month for save percentage was an awful 80.6%. Once again, the JOFA line was rolling; their PDO was a much-lower 94, but that was due to the line shooting at 15% once again. This 15% shooting average over two healthy months was great—imagine if they had avoided injuries the two months prior.
Riding this momentum into February (or, of course, Filbruary), Rinne and Saros returned to form, returning to a great 92.6% save percentage. However, the line’s shooting percentage came crashing back to average. For the month, the JOFA line had a 9% shooting percentage, which is not bad by any means, but was a departure from their last two months of work. March was more of the same; goaltending managed to increase again (95%), yet shooting percentage was a pedestrian 8%.
One possible answer is the effect of splitting the lines up unnecessarily: shooting percentage took a dip every time the lines were split up, and often took a game to recover back to normal levels. They were contributing to better defense for the goaltenders while they were on the ice, but offensively, they were in a rut, and have been for the better part of two months.
The Home Stretch — KEEP. JOFA. TOGETHER.
At this point it should be fairly obvious what the team should do with the first line heading into the final, crucial seven-game stretch into the playoffs. The quality of competition is incredibly high (Pittsburgh, Columbus, Toronto), and with the recent two-game benching of Kyle Turris, nothing is certain for the other nine forwards as the lineup continually shifts.
Järnkrok–Sissons–Smith had caught fire, which fits with Sissons’s career year, but with Fiala gone, Turris in the doghouse, Rinaldo on IR, and Salomäki and Watson on a conditioning loan, the coaching staff has to figure something out quick with their three trade acquisitions.
Whatever they do, one thing is certain—they have to keep the JOFA line together, all game, every game. Let the chemistry return. The numbers show that they will perform much better together rather than split apart, even with suspect goaltending (or not). A lot of criticism around the team is that it’s only one line. Whether that’s true or not, the truth remains: let that line do work.
Still, the last two month’s regression of the JOFA line is enough to be seriously concerned—if their best months are behind them, a hot start won’t make a difference for a lackluster playoff performance. However, the combination of goaltending/defense increasing while JOFA is on the ice, and the potential that we have seen the line perform with earlier this season, leave room for hope that the line can carry this team in the playoffs.
The criticism of the team early in the year was that it was “a one line team, JOFA and not much else.” There’s a lot of truth in this, but Laviolette and company would be putting this team in the best position by icing this line together, all game, every game. Let the one-line team be carried by that line—or at least let them try. The alternative is much, much worse.