By Leon Lifschutz | Follow me @BruinsBreakdown
The debate over what matters more, offense or defense, has raged in sports likely since their creation. Does Novak Djokovic’s attack succeed versus Rafael Nadal’s ability to defend and return? Does the stalwart defense of the 1985 Chicago Bears hold off the offensive onslaught of the 2013 Denver Broncos? How does the dynamic offense of the Michael Jordan-led Bulls stack up against their own tenacious and determined defending? The topic makes for a fun debate amongst sports fans and holds key insights for team decision-makers.
In today’s post, we will try and answer the question for the present-day Boston Bruins. The Bruins have been a successful squad over the past five seasons winning the President’s Trophy for best regular-season record in 2019-20 and never finishing lower than 3rd in the Atlantic Division. So what has been the biggest driver of their success? Has it been an offense paced by an elite top line and powerplay, or a defense anchored by strong goaltending and a team commitment to suppressing opponents?
In order to examine this topic, we will take a look at several pieces of data. Obviously, goals are the most important stat to look at as the entire point of the game is to score more than your opponent. So we will start by looking at goals for and goals against to see if one or the other has a greater association with, or is a better predictor of, a team’s winning percentage. We will also examine goals for percentages.
Hockey is a variable sport. You can play a great game with lots of chances but hit a couple of posts, give up a fluky goal, and lose. Some teams can ride lucky streaks through the whole season, for example, the 2016-17 Ottawa Senators. Despite finishing 20th in expected goals for percentage, they were an overtime goal away from the Stanley Cup finals. As we know, the team derailed miserably the following season and is just now on the verge of recovery. Therefore, it can be prudent to examine underlying numbers to get a clearer picture of how a team plays. Expected goals for, against, and expected goals for percentage can be useful metrics. While not as powerful as goals, over time, expected goals do have strong predictive powers for both teams and individuals. Expected goals can also help demonstrate how a team plays stylistically better than actual goals can. It will serve as a helpful aid in identifying how teams, like the Bruins, win.
Finally, we will look at PDO which is a composite stat combining shooting and save percentages. PDO usually ends up around 100. Finishing higher than average means one of two things – your team is talented and can sustain it or your team was lucky and expect regression. This will provide us a little additional context into the other numbers we are looking at.
The Bruins’ five seasons between 2015-20, while insightful, do not provide enough data points to be statistically significant. A single outlier season would influence twenty percent of the analysis. Therefore we will first take a look at every NHL team from the past five seasons, 153 in total, to make sure the patterns we see with the Bruins coincide with the larger, statistically significant, trends league-wide.
The scatterplot is adjusted into Z-score which just means that numbers are standardized in distance from the average, the gray line through the middle of the chart. This particular chart is pretty self-explanatory. If you score more or let up fewer goals, you have a better chance of winning. However, there are some key takeaways here to aid our analysis. League-wide, scoring goals has a slightly higher correlation to accumulating points than preventing them. This does make sense as you do need to outscore your opponent. The other key point is whether you are an offensive or defensive-minded team, being better at it than your opponent, evidenced by an extremely strong correlation of goals for percentage to winning, is most important.
Having higher expected goals for correlates slightly higher with winning than expected goals against. That leads us to the conclusion that in the past five NHL seasons, playing a little more offensive-minded, albeit minimally, has been a stronger driver of winning than defense. Examining the expected goals for percentage, we again note that regardless of your style doing it better than the other team is most significant.
A final and really important data point from this chart is PDO. PDO shows stronger correlations with winning than all of our metrics expect goals for percentage. This would suggest that regardless of your style of play talent and luck play a huge role in winning in the NHL. Some teams can put together strong streaks of exceeding expectations in a given season and have a high PDO. Those teams, like our Senators example earlier, will come crashing back to earth. Others, like the Washington Capitals over the past decade, have the talent to continuously outpace their analytical expectations. One other example is the New York Islanders under Barry Trotz who, due to their style of play, continue to exceed analytical expectations.
The below table shows the Bruins output (with league-wide rank in parentheses) for each of the metrics we are examining over the past five seasons.
|2019-20||71% (1st)||227 (7th)||167 (1st)||57.6% (1st)||190 (18th)||171 (4th)||52.6%(5th)||1.022 (1st)|
|2018-19||65% (2nd)||257 (11th)||212 (3rd)||54.8% (3rd)||223 (14th)||204 (4th)||52.3% (8th)||1.008 (8th)|
|2017-18||68% (4th)||267 (6th)||211 (4th)||55.9% (2nd)||236 (9th)||195 (1st)||54.7% (1st)||1.011 (10th)|
|2016-17||58% (13th)||232 (13th)||209 (9th)||52.6% (10th)||228 (5th)||186 (2nd)||55.1% (1st)||0.990 (21st)|
|2015-16||57% (15th)||236 (5th)||228 (20th)||50.9% (15th)||202 (9th)||207 (10th)||49.4 (18th)||0.999 (18th)|
The Bruins’ playing style over the past five years has had a heavy emphasis on defense first. The team has not been out of the top-ten in expected goals against the whole stretch and only slipped out of the top-five once. In goals against, the team only once fell out of the top-ten and have been a top-five team in the last three seasons. In regards to the offense, the team has exceeded expectations every year except for 2016-17 showing they are an opportunistic bunch. It is also important to note that the statistics above include all situations. This past season, at 5v5 the team performed fourth-worst in the league in expected goals for, further exhibiting their conservative style and defense-first mentality. While the team can score, neither their goals for nor expected goals for totals rival their defensive numbers. This has been an organization with a clear identity – limit chances against, be a strong possession team, and finish your opportunities.
Boston Bruins GF and xGF versus Winning Percentage 2015-20 pic.twitter.com/JDEhYbMBOy— BruinsBreakdown (@BruinsBreakdown) December 13, 2020
Scoring more goals or letting in less have a strong impact on winning for the Bruins much as we saw league-wide and frankly expected. According to the above chart, preventing goals has a stronger association with winning for the Bruins than scoring. Given the league-wide trend suggesting scoring is slightly more important, our Bruins’ data could be misleading. However, given what we have determined around the Bruins’ desire to play with a defensive conscious we should also consider that Boston may be an outlier for whom defense has greater importance.
Percentages again matter. Goals for percentage has a more significant impact than goals for or against. How important PDO is should not be understated with a strong and significant trendline. Using statistical software, PDO was the only Bruins’ variable to prove statistically significant in correlation or regression analysis despite the limited data points. The Bruins have regularly been above average in this category and their ability to sustain that will be a big factor in future success. A potent powerplay, several above-average shooters, strong goaltending, and good team defense suggest the Bruins may be able to stay above the mean.
Offense Or Defense?
League-wide, there appear to be many ways to win – though offense has the slightest edge at the moment. Whether you are a defensive-minded team or an offensive powerhouse, you just have to be able to do either better than your opponent. The Dallas Stars and the Tampa Bay Lightning, stylistically opposed teams, both made the Cup finals. Regardless of your style, performing above expectations with strong shooting and save percentages can be a huge boon for a team’s success. While luck can be a factor in PDO, having a team laden with talent makes a big difference in sustaining success.
Some might argue that these conclusions are fine for the regular season but that the old adage, “defense wins championships”, holds firm. Building off of our Stars and Lightning example, Dom Luszczyszyn of The Athletic recently took on the task of determining whether offensive or defensive-minded teams are more likely to win a playoff series. His analysis shows that there is more than one way to win. However one of his key takeaways is an offensive-minded team, as long as their defense is at least above average, can fair quite well.
This brings us to the big bad Bruins who have an elite defense and an above-average offense. Boston plays a defensive-minded style as evidenced by the metrics we examined and others like conceding the least amount of counter-attacks across the whole league in 2019-20. They have been defensive stalwarts over the past five seasons. A strong defensive structure supported by quality goaltending has been the more prominent factor for Boston versus offense. The Bruins certainly have also had success on offense but it’s been driven more by a strong powerplay and above-average shooters than an offensive-minded style. The Bruins will likely put an emphasis on defending again next year. With a revamped and less experienced defensive core, along with aging goaltenders, it will be interesting to see if the strategy continues to pay off.