This is part three of my series on how roles impact statistics. In Part I I looked at a brief history of the debate and also described the methods used in this research. The history section is optional but I recommend reading the methods portion as it describes the statistics and methods used which will not be discussed in great detail here.
The second post in the series looked at the relationship between roles and offensive statistics (GF60 Rel, CF60 Rel, and Sh% Rel) for forwards. In this third post I will look at the defensive statistics (GA60 Rel, CA60 Rel and Sv% Rel) which ought to be interesting as defensive statistics have historically been difficult to get a handle on (compared to offensive statistics).
Leading and Trailing TOI%
The first role indicators that I am looking at is leading and trailing TOI%. Here is how they compare to 5v5close GA60 rates.
Goals against are bad so increasing them (blue) indicates player is worse off. Here we see that players that are given more ice time trailing also give up more goals against in close situations while those that get less ice time trailing give up fewer goals. Getting more defensive ice time protecting a lead does not necessarily mean giving up fewer goals though (moving left to right along a row does not necessarily mean lower GA60).
Another thing you may notices is that the scale in this chart is approximately half that of the GF60 Rel chart in Part II of this series of posts. While there is a clear relationship between Trailing TOI% and GA60 Rel, the relationship is not as strong as the relationship with GF60.
This is where things get a little interesting. There is not a significant pattern here, certainly not like the GA60 Rel chart which is quite interesting. Players that get more ice time playing catch-up hockey give up more goals, but don’t give up more shots. That can only mean one thing…
Yes, those players that get more ice time when trailing give up more goals in 5v5close situations because they have a negative impact on their own teams save percentage.
Again it is worth noting that the CA60 Rel and Sv%Rel charts have scales about half that of CF60 Rel and Sh%Rel charts from the last post so players do have a greater ability to impact offense than defense.
Defensive and Offensive Face Offs
Players that get the most offensvie zone face offs also give up the most goals against. Clearly starting in the offensive zone is not helping them in the goals against department. This is just further evidence that face off zone starts don’t really have significant impacts on results.
Not much of a pattern here either. Face off zone starts don’t seem to correlate with shots against.
There is somewhat of a pattern here. Those that get more offensive zone face offs do generally have a negative impact on save percentages. Getting more defensive zone face offs does not seem to have any correlation with save percentages.
Penalty Kill and Power Play
You probably guessed what you might see here but those that get the most powerplay ice time generally have a negative impact on goals against rates. Those players that get little to no powerplay ice time generally reduce goals against in 5v5close situations. In the middle there is quite a mix though generally moving down the chart brings in more red grid cells.
Again, not much of a connection between role and shots against. Some players getting significant defensive time end up giving up a lot of shots against.
And this chart confirms once again that offensive players are generally have worse save percentages relative to their teammates while those that get little or no power play ice time generally boost their teams save percentage in 5v5close situations.
In the last post we saw that players that get the offensive assignments also happen to produce the most offense. However in this post we see that these same players also give up the most goals against by having a negative impact on their goalies save percentage.
A few weeks ago I tweeted a poll about whether the best defense is a good offense and 52% of respondents agreed (22% thought it was most true and another 30% thought it was equally true) however from these charts this is not correct. The best defense is typically not a good offense. In fact a good offense often leads to a worse defense.
In hockey which of the following is most true?
— David Johnson (@hockeyanalysis) September 2, 2016
The positive for offensive players is that on average the positive impact to their teams offense is greater than their negative impact to their team defense so overall they are doing better. However it is clear that the best defense is not necessarily a good offense (52% of you got that wrong).
What we don’t see though is offensive players giving up more shots against so their negative influence on goals against is typically through negative impacts to their goalies save percentages. This doesn’t mean that forwards can’t impact shots against, it just can’t be generalized according to offensive and defensive roles as defined here. The clear takeaway from this though is that forwards can and do influence save percentage. In a future post in this series I’ll look at persistence and predictability of Sv%Rel so stay tuned but I’ll turn my attention to defensemen first.