Last week, we joined in the ongoing debate about whether Mitch Marner and Auston Matthews should play together on the same line or be split onto different lines. We’re not the first to weigh in on the topic, and we won’t be the last. This possibility has been a topic of discussion among Toronto Maple Leafs fans and hockey analysts for several seasons now.
In our article last week, we dug into the impact of this pairing using a simple approach by tracking the 5-on-5 goals for and goals against statistics from the past two seasons when each of the different combinations (Matthews with Marner, or each with another partner) was on the ice. We concluded, given simple addition and subtraction, that the Maple Leafs might benefit from Marner and Matthews being on separate lines. We also looked at the numbers when combined with their other two sometimes linemates, John Tavares (with Marner) and William Nylander (with Matthews).
Related: Guy Lafleur Gave Me a Lifetime of Memories
However, this time we’re going to do the process again but digging more deeply into the analysis using one advanced analysis — that’s the concept of expected goals (xG) as our measuring stick.
Expected Goals: What Are They?
The concept of Expected Goals, often denoted as xG, provides a more comprehensive insight into a player’s performance by eliminating the unpredictable factor of goaltending. xG is the statistical probability of an unblocked shot becoming a goal. While it’s not a perfect metric and relies on NHL’s play-by-play data, which can be faulty, it offers a more direct measure of shot quality.
(Jess Starr/The Hockey Writers)
Here are the key points to know about expected goals:
- xG quantifies the likelihood of a shot resulting in a goal.
- xG helps measure shot quality, accounting for the play-by-play data’s (what actually happened) limitations.
- The NHL’s play-by-play data, while publicly available, might have inaccuracies, inconsistencies, and omissions. While it tries to be as objective as possible, there remains an element of human decision-making that is factored in.
Analyzing Expected Goals for the Maple Leafs’ Pairings
For our analysis, we relied on on-ice, 5-on-5 statistics from Naturalstattrick.com over the last two seasons, with a focus on xG per 60 minutes played.
Related: Maple Leafs’ Matthews Learned His Hockey Skills in a Phoenix Thrift Store
- Matthews/Nylander: This pairing had the highest xG for per 60 at 3.86 but also the highest xG against per 60 at 2.71, resulting in a positive difference of 1.15 goals per 60 minutes played.
- Matthews/Marner: Matthews and Marner had the second-highest xG for per 60 at 3.56, paired with the lowest xG against per 60 at 2.09, resulting in a remarkable positive difference of 1.47 goals per 60 minutes.
- Nylander/Tavares: This duo boasted an xG for per 60 of 3.22 and an xG against per 60 of 2.58, leading to a difference of plus 0.64 goals per 60.
- Marner/Tavares: Marner with Tavares had the lowest xG for per 60 at 2.94 and an xG against per 60 of 2.51, resulting in a difference of plus 0.44 goals per 60.
When combining xG for per 60 for Marner/Matthews and Nylander/Tavares, we obtained a total of 6.78 xG per 60. Summing up the xG for for Marner/Tavares and Nylander/Matthews, we arrived at 6.80 xG per 60. The offensive difference is negligible at just 0.02 xG per 60, virtually the same.
Related: Revisiting the Capitals’ Disastrous Jagr Trade
The key difference lies in xG against per 60. With Nylander/Matthews and Marner/Tavares, the combined xG against per 60 was 5.22. Conversely, Marner/Matthews and Nylander/Tavares had a combined xG against per 60 of 4.67.
The Bottom Line
In our previous analysis using actual goals for and against, splitting Marner and Matthews showed a 1.04 goals per 60 improvement. However, when using expected goals as a guide, the improvement with Matthews and Marner together was 0.53 goals per 60.
When analyzing the impact of Marner and Matthews playing together or apart using actual goals scored and allowed, the data suggested that they performed better when apart. However, when we shift our focus to xG, a different picture emerges.
The xG statistic, which accounts for shot quality and eliminates goalie performance as a factor, indicates that Matthews and Marner together excel defensively, with an xG against per 60 of 2.09, which is notably lower than other combinations. This suggests that their partnership is effective in preventing opponents from generating high-quality scoring chances.
Related: Revisiting Maple Leafs & Marner’s 2019 Contract Negotiations
While xG for, which measures offensive potential, showed a marginal difference between the two pairings, the significant improvement in xG against when Matthews and Marner are together is an important finding. Therefore, when considering xG, the data leans towards the idea that having Matthews and Marner on the same line might be a more beneficial strategy for the Maple Leafs, particularly from a defensive perspective. It’s their combined defense that is the key factor.
The Value of Using Advanced Statistics
When answering the question “Should Matthews play with Marner or not?” applying advanced statistics doesn’t give THE answer. However, the statistics do give us more information we might apply to make a more reasoned and better informed decision.
(THE CANADIAN PRESS/Nathan Denette)
The beauty of using statistics to examine this issue lies in the debate between what was expected and what actually transpired. Unfortunately, as is often the case in hockey analysis, we cannot definitively conclude which combination is better.
The decision will ultimately depend on a variety of factors, including coaching strategies, player chemistry, and in-game situations. Nonetheless, this ongoing debate keeps the hockey world engaged, highlighting the intricacies of team dynamics and the beauty of the sport. Funny how two different uses statistics offer two different possibilities. That’s one thing that makes hockey fun.
[Note: I want to thank long-time Maple Leafs’ fan Stan Smith for collaborating with me on this post. Stan’s Facebook profile can be found here.]
