What’s going on with John Tavares, and why?


As the Captain and one of the highest paid players in the league, a lot is expected of John Tavares. When he decided to join Toronto in the summer of 2018, it was a joyous day. Adding a top flight center to what was already an elite offense brought mouthwatering visions of having two 30-plus-scorers (good name for a new dating site) on one team. His first season in Toronto seemed a blessed one, scoring 47 goals in a nearly unstoppable combination with Mitch Marner.

In this season, Tavares hasn’t had the same punch to his play. With 6 goals and 17 points in 21 games, it’s not like he’s been terrible in a producing role. However, something seems to be up with his ability to be an impactful player that can change the game the way an Auston Matthews, Mitch Marner, and even William Nylander can.

We know that Tavares is billed, primarily, as a points-scoring forward. There’s some talk of his defensive ability as a center — we’ll talk about that misnomer later — but, primarily he’s a scorer. All stats in this section are from naturalstattrick.com, unless otherwise noted.

Advertisement – Continue Reading Below

Production Rate

There’s a lot of permutations of scoring stats that can show you production from a different perspective. One such way is expressing it as a rate based on how much time they spend on the ice. It seems obvious that the more ice time you get, the more likely you are to score, assuming you have the ability to do so.

Expressing these stats as a rate per 60 minutes of ice time is most common, and typically we use either Even Strength or 5-on-5 numbers. Some players aren’t picked for powerplay time, or are shirked with shorthanded time, and that shouldn’t affect our analysis of their abilities, scoring or otherwise.

In terms of points at Even Strength, Auston Matthews is 6th in the league among players with more than 150 minutes of ice time (it’s important to filter out low-ice-time players as there’s a high chance of outliers, such as Rasmus Sandin’s 12 points per 60 having 1 point in just 5 minutes of ice time). Tavares, conversely, is 141st, scoring just 1.87 points for every 60 minutes of ice time. That’s lower even than a number of defenders, including Toronto’s own Jake Muzzin.

In Tavares’ 10 years in the league, this is, by a healthy margin, his worst season (so far) in points per 60 at even strength (from evolving-hockey.com).

Advertisement – Continue Reading Below

Season Team GP TOI Points/60
2011-12 NYI 82 1381.6 2.13
2012-13 NYI 48 807.27 2.16
2013-14 NYI 59 978.53 2.39
2014-15 NYI 82 1355.2 2.13
2015-16 NYI 78 1269.82 2.22
2016-17 NYI 77 1218.6 1.97
2017-18 NYI 82 1202.03 2.2
2018-19 TOR 82 1281.93 2.86
2019-20 TOR 63 997.17 1.99
2020-21 TOR 21 305.65 1.77

Primary Production Rate

One way we can improve how we look at points scoring is by prioritizing “primary” points, which are goals and 1st assists. It has been shown that secondary assists are very “noisy” from a statistical perspective, meaning that they vary widely from year-to-year for players, regardless of their skill level. Conversely, primary points tend to be “repeatable”, meaning that good players are able to be good at scoring primary points year-to-year, which is not true of 2nd assists. 

From naturalstattrick.com, Tavares sits 173rd this year in Primary Points/60, showing that he is having a higher portion of his points as secondary assists than some of the players behind him in Total Points/60, such as Sam Bennett or Phil Kessel.

In fact, we can show that Tavares’ ratio of primary points to secondary points is significantly worse than previous years (from evolving-hockey.com).

Advertisement – Continue Reading Below

Season Team GP TOI G/60 A1/60 Primary Points/60 A2/60 Points/60 Primary/Secondary Ratio
2011-12 NYI 82 1381.6 0.87 0.91 1.78 0.35 2.13 5.09
2012-13 NYI 48 807.27 1.26 0.67 1.93 0.22 2.16 8.77
2013-14 NYI 59 978.53 0.92 1.1 2.02 0.37 2.39 5.46
2014-15 NYI 82 1355.2 1.02 0.84 1.86 0.27 2.13 6.89
2015-16 NYI 78 1269.82 1.04 1.04 2.08 0.14 2.22 14.86
2016-17 NYI 77 1218.6 0.84 0.59 1.43 0.54 1.97 2.65
2017-18 NYI 82 1202.03 1 0.7 1.7 0.5 2.2 3.40
2018-19 TOR 82 1281.93 1.59 0.75 2.34 0.51 2.86 4.59
2019-20 TOR 63 997.17 0.96 0.72 1.68 0.3 1.99 5.60
2020-21 TOR 21 305.65 0.39 0.59 0.98 0.79 1.77 1.24

Score-Close Production Rate

Another way we can look at his impact on the game is what happens in those crucial times of the game when the score is close. This used to be pretty common, but with the advent of score adjustment, this is rarely used. However, we don’t have score-adjusted-scoring that I know of. Instead we will look at scoring when the score is within one goal. Again, we will use Even Strength numbers and express them as a rate per 60 minutes of ice time, and filter out those with less than 150 minutes of ice time this year.

In this area, again drawing comparison between Toronto’s top centers, Auston Matthews is 3rd in the league at 3.94 points/60 when the game is close, slightly worse than in all score states. Tavares is down at 158th in the league, with 1.46 points/60 when the game is on the line. This from naturalstattrick.com. 

We have shown, in detail, what Tavares’ results are this year as compared to the league and to previous versions of himself. The “Why” part of the analysis, while perhaps more interesting, is caked in many layers of nuance. There are a lot of factors that can influence why someone is performing poorly, only some of which we can show with data. And in those we can show with data, there’s a lot of overlapping, oversimplifying and underappreciating going on that the answer can’t be simple without combining stats into a multi-stat regression model, like Wins Above Replacement (WAR).

Advertisement – Continue Reading Below

Here we will look at some of the individual factors, as well as Evolving-Hockey’s WAR model, to try to answer the question of “Why?”

Shooting Percentage

Another way to look at scoring is shooting percentage. From evolving-hockey.com, here are John Tavares’ yearly individual shooting percentage for shots on goal, on-ice shooting percentages, on-ice save percentages, and the combined on-ice ‘PDO’ as a total expression of luck (for a detailed explanation of PDO, see the stats primer I wrote).

Season Team GP TOI iSh% Sh% Sv% PDO
2011-12 NYI 82 1381.6 9.3 8.4 90.04 98.44
2012-13 NYI 48 807.27 14.66 8.52 90.09 98.61
2013-14 NYI 59 978.53 11.72 10.31 91.81 102.12
2014-15 NYI 82 1355.2 11.73 8.14 91 99.14
2015-16 NYI 78 1269.82 12.57 8.56 93.11 101.67
2016-17 NYI 77 1218.6 9.6 8.73 91.31 100.04
2017-18 NYI 82 1202.03 12.5 9.54 91.4 100.94
2018-19 TOR 82 1281.93 16.5 11.32 92.13 103.45
2019-20 TOR 63 997.17 10.26 8.36 89.17 97.53
2020-21 TOR 21 305.65 5 5.39 95.2 100.59

It’s a common misconception that the higher your shooting percentage is the better scorer you are, however, what we often see is if shooting percentage is really high or really low, it regresses toward more average numbers. So, for Tavares’ 2020-21 season, his individual shooting percentage is bad at 5%, as is his on-ice shooting percentage at 5.39%. This is not an indication that Tavares is playing poorly; instead, it’s an indication that luck hasn’t been going his way. We can expect that he will have an uptick in scoring as the season goes on, regressing him towards a more average shooting percentage of 8-9%.

Shot Location

Within this one of many ways to look at how a player is performing, there are many ways to look at Shot Locations. One way is through shooting percentage versus expected. The table below shows Individual Fenwick shooting percentage (iFSh%) for shots on goal or shots that miss the net, and expected individual Fenwick shooting percentage (ixFSh%) for how good those unblocked shots should have been based on location.

Season Team GP TOI iFSh% ixFSh%
2011-12 NYI 82 1381.6 7.17 9.64
2012-13 NYI 48 807.27 11.04 8.22
2013-14 NYI 59 978.53 8.77 9.15
2014-15 NYI 82 1355.2 8.52 8.99
2015-16 NYI 78 1269.82 9.36 9.32
2016-17 NYI 77 1218.6 6.91 8.41
2017-18 NYI 82 1202.03 9.43 8.37
2018-19 TOR 82 1281.93 11.85 8.85
2019-20 TOR 63 997.17 7.21 6.84
2020-21 TOR 21 305.65 3.85 7.81

Another way is through heat maps, the prettier but harder to analyze method, available from hockeyviz.com. Below is a series of heat maps showing John Tavares’ shot locations for each season he’s been in the NHL. From the above table we can see there are some years with big discrepancies between shooting percentage and expected shooting percentage. These heat maps can help to show why that may be. It’s important to note that being “+” on offense is good, and being “+” on defense is bad.

Look at the years where there’s a big red pool in front of the top (offense) goal crease, that being basically every year except the last two years with Toronto. This corresponds with the two lowest years in ixFSh%, which makes sense. The fewer of his shots that come from that “home plate” shaped area in front of the net, the worse his shooting percentage should be. This is expressed in a different way on the heat maps through xGF%, which stands for Expected Goals For%, which I explain in my primer.


Corsi isn’t very good at explaining why, in previous games, players haven’t scored, so I’m hesitant to use it here in case it gets misconstrued that way. What Corsi is good at is identifying if players in future games should be expected to be a positive impact on the score sheet. This is certainly of interest overall, but it doesn’t help to answer that question of “what’s happening and why”, so I’ve not included it. You can find those numbers on either evolving-hockey.com or naturalstattrick.com, if you’re interested.

Unlike Corsi, a multi-stat regression model like Wins Above Replacement, or WAR, is better suited to explaining what’s happening now and why. The What is the WAR, and the Why is the xWAR, or Expected WAR.

The “What?” — WAR

10-11 NYI 82 1686.6 11.3 -2.4 3.2 -0.1 2.5 4.2 14.6 -2.5 6.7 18.8 3.3
11-12 NYI 48 997.2 5.4 0 1.6 0 1.5 1.1 7 0 2.6 9.6 1.8
13-14 NYI 59 1252.7 9 -4.8 3.8 -0.1 0.9 0.1 12.8 -4.8 0.9 8.9 1.7
14-15 NYI 82 1695.2 10.1 -4.1 6.6 0 0.5 0 16.6 -4.1 0.5 13.1 2.5
15-16 NYI 78 1559.6 9.2 -4.6 3.6 0 1.9 1.5 12.8 -4.6 3.4 11.6 2.2
16-17 NYI 77 1572.2 7.7 -1.4 -0.1 1.2 0 2.5 7.6 -0.2 2.5 9.9 1.9
17-18 NYI 82 1633.9 9.7 -6.1 3.4 0.1 0.7 0.4 13.2 -6 1.1 8.3 1.6
18-19 TOR 82 1565.3 15.8 -1.3 3.2 -0.4 1.1 1.1 19 -1.7 2.2 19.5 3.6
19-20 TOR 63 1231.6 5.6 -2.6 5.1 0 0.1 0.3 10.6 -2.6 0.4 8.5 1.5
20-21 TOR 21 376.4 -0.2 -0.2 2.1 0 0.3 -0.1 1.9 -0.2 0.1 1.8 0.3

As expected based on everything we’ve shown thus far, this is Tavares’ worst season yet in terms of WAR and GAR (Goals Above Replacement, the basis for WAR). It’s important to note that GAR is cumulative, so it will continue to grow or shrink with more ice time, assuming the same trends continue. So his being 2.1 in powerplay offense GAR isn’t actually worse than last year’s 5.1, as he’s played about a quarter of the ice time so far.

He’s been a negative contributor on offense and defense at even strength, his major contribution coming as an offensive threat on the powerplay. No other season has he been this bad at even strength. He has also been poorer than normal in terms of faceoffs, “Draw_GAR”.

In total, thanks to the powerplay contributions, Tavares is still  a positive player, but not by much. The next section will show us what direction we should expect that to trend in.

The “Why?” — xWAR

Expected GAR or WAR is a way to show what the underlying numbers say a player should be contributing, factoring out the luck factors that go into final results. WAR is what they have contributed, while xWAR is what they should contribute in the future. I’ll only show the last two seasons, as previous seasons’ expected WAR is no longer relevant. Last year’s is shown only for example’s sake on what a longer seasons’ xGAR might look like.

19-20 TOR 63 1231.6 7.1 -1.9 3.1 0 0.1 0.3 10.1 -1.9 0.4 8.7 1.5
20-21 TOR 21 376.4 -1 -0.7 1 0 0.3 -0.1 0 -0.7 0.1 -0.5 -0.1

It appears that from this perspective, Tavares is performing at a level that would have him as a negative WAR player for the first time in his career. However, with 21 of the 56 games played so far, and Tavares sitting at a slightly positive clip so far, it’s more likely that Tavares hovers around that 0.3 WAR mark, if his underlying play doesn’t significantly improve.

As a summary of the above, what is happening with Tavares is:

  1. He’s scoring at a lower points/60 and primary points/60
  2. He’s producing less when the score is close
  3. He is producing a slightly positive WAR, much worse than previous years

The why and what to expect in the future is:

  1. His shooting percentage on-ice and individually is lower than usual, and should increase
  2. His expected Fenwick shooting percentage is higher than his actual Fenwick shooting percentage, so the actual Fenwick shooting percentage should again increase
  3. Similar to last season, he is not taking as many shots from the dangerous locations in front of the net
  4. His xWAR shows that his already low WAR should decrease as the season goes

Overall, there’s some concerning and encouraging elements here, as one might expect. When taking a deep dive like this into the numbers, you’ll often find trends that conflict and leave you with an indefinite end perspective. I believe that this accurately represents reality, where there is no black-and-white answer on “is this person good or bad”, in terms of performance (but also in terms of morality, though that is unrelated).

As the season goes on, the Leafs will expect an improvement from John Tavares, and we can’t say for sure whether they’ll get what they expect. Most often, we Leafs fans don’t get what we expect, so I’m afraid I’m leaving this post with a pessimistic outlook on Tavares in this season thus far.