Can Players Maintain or Increase Their Performance with an Increased Role?

Cover Image

Back in December, Scott Cullen wrote about relative possession numbers in his Statistically Speaking column. He took defensemen who averaged 15+ TOI/gm and forwards who averaged 12+ TOI/gm and sorted them based on their relative possession numbers (CF%Rel). He then analyzed defensemen with < 15 TOI/gm and forwards with < 12 TOI/gm. He wrote:

Then, some forwards that aren’t playing a whole lot but, based on relative possession stats, could be worth considering for more ice time.

This made me wonder, should a strong relative possession statistic be taken in consideration when determining if a depth player can handle a larger role? And if so, how much weight should it hold?

I came up with a list of 76 players that from one season to the next, saw their even strength ice time increase by over two minutes per game (‘07-‘08 to Present). I originally started with 90 players but eliminated anyone that didn’t reach EV 500 TOI in the season prior to seeing an increase in TOI/gm. I further cut down the list to exclude anyone that played defense along with forward (ie. Brett Burns).

Let’s look at how players with their 5v5 TOI/gm increased by two or more minutes do in respects to their CF% Rel from one season to the next:

CF%Rel

Overall, 48 of the 76 (63%) saw their CF% Rel increase with an increased ice time. There is a positive correlation between the two variables, shown by the trend line’s positive slope, but it appears to be heavily influenced by players negatively affecting their team’s shot attempt percentages in the first year having a less negative impact with more ice time. With increased ice time, player’s relative shot attempt numbers increased as well, but the trend line only carries a R-squared value of .2938.

Despite this, only 12 of the 33 (36%) players that saw a positive CF% Rel in the year before their ice time increased by 2+ minutes, saw further increase to their CF% Rel with increased ice time. This may be a cause for concern for my original question. Since the R-squared value of .2938 means that the best fit line only explains 29.38% of the data, we’ll have to further break it down to fully understand the big picture.

First, I ordered the player’s first year CF% Rel from largest to smallest and split the data into five even groups. I then averaged each group’s CF% Rel from both years to see how the smaller groups performed by their percentage quartile.

CF%Rel Quartile Analysis

The results show a much different picture than the prior graph. The bottom four percentile groups were able to post a higher average in the player’s second year, but the group at the top wasn’t able to sustain their higher relative shot attempt percentage. The players in the 20th percentile group had the largest change in impact (over three percentage points), turning a negative impact to a slight positive impact.

Every player in both the 60th and 80th percentile groups started with a positive impact in the first year. The players in the 60th percentile group were able to barely increase their average impact where the players in the 80th percentile group took a few steps backwards. It is important to note that the 80th percentile group still did positively impact possession in the second year, but not to the extent that they were able to originally.

From here on, I’ll only be focusing on the 33 players that were able to positively impact possession in the first year as they may be looked to before others, as Cullen suggested, to take on a larger role.

CF%Rel Positive CF%Rel 1st Yr

Recreating my earlier graph with only the positive first year CF% Rel points, shows that players can maintain a positive impact to relative shot attempt percentages, but not to the extent that they did with a smaller role. 25 of the 33 players (76%) were still able to positively impact possession statistics with increased ice time. Look specifically with players with at least a +2.00 CF% Rel in the first year (17 total), only Bobby Ryan, Ryan O’Reilly, and Teddy Purcell were able to further increase their relative shot attempt percentage.

It doesn’t seem very likely that players can maintain a high CF% Rel with an increase in ice time. But that doesn’t translate to a warning of “don’t give players who post a high relative shot attempt percentage in a limited role more ice time”. While players may have less of an impact to relative shot attempt percentages, their impact could increase elsewhere with an increased role. Let’s first look to production and goal-based metrics.

Production Chart

For the most part, with an increase in ice time, players could increase their production and goal-based metrics (the percentages are conditionally formatted where blue is a higher rating and red is a lower rating). This of course, is nothing new but important to note as scoring goals and winning games is the main objective. In total, 14 of the 33 players increased their G/60, A/60, and P/60 metrics collectively. While production increased, very few players were able to prevent goals against at the same rate. Players in this group struggled to prevent goals (in this case and for later in the blog, a higher mark in against metrics is referred to by a lower total in the second year). Despite this, a majority of the players were able to increase their GF% Rel.

Next we’ll look into shot attempts:

Shot Attempts Chart

As production increased, so did player’s shot attempts for. But, players saw a decrease in shot attempts against, shot attempts percentage, and relative shot attempts percentage, as we looked into earlier. This doesn’t necessarily mean they negatively impacted possession, but just not to the extent of the year prior. For example, Milan Lucic saw a his TOI/gm increase by 2.27 minutes in 2008-2009 with the Boston Bruins. While his CF% dropped from 52.07% to 51.73%, his G/60, P/60, and GF% Rel all increased, showing he in fact did have a positive impact to the Bruins with more ice time.

Taking this one step further, we’ll break down shot attempts into scoring chances and high-danger scoring chances:

Scoring Chances Chart

In a similar fashion to shot attempts, players were able to increase their scoring chances and high-danger scoring chances for but weren’t able to prevent scoring chances and high-danger scoring chances against at the same rate.

It is clear that players starting with a smaller role weren’t as strong in the defensive zone, both in suppression shot attempts and preventing goals against. With an increased role, opponents exposed their weakness in the defensive zone, shown by decreased GA/60 rates.

In total, only Andy McDonald (+2.07 TOI/gm in 2010-2011 for the St. Louis Blues) saw a stronger positive impact in the second season for all 20 categories. Not far behind, Ryan O’Reilly (+2.78 TOI/gm in 2011-2012 for the Colorado Avalanche) saw a stronger positive impact in 19 of the 20 categories (only his G/60 dropped from .59 to .49).

On the flip side, there were some players who performed worse in almost every category when given more ice time. Looking back, it is possible to question the coaches’ decision to give them a larger 5v5 role. Mikael Backlund (+2.39 TOI/gm in 2011-2012 for the Calgary Flames) saw his performance worsen in every category in a season that was plagued by injuries, only allowing him to play in 41 regular season games. He was able to rebound after that season. Brandon Prust, in his first full season with the New York Rangers, saw his ice time increase by almost five full even strength minutes, but only saw an increase in two categories (HSCF/60 & HSCF%). Prust was able to set career highs in goals (13), assists (16), and points (29) that season but did so at a much lower rate.

Players can still be successful taking on a larger role while posting a negative relative shot attempt percentage with less ice time. CF% Rel is just one of many factors that come into play. After a disappointing rookie year where Tyler Seguin mainly saw time on the fourth line and posted just 22 points in 74 games, his ice time increased by over three even strength minutes per game and he had a breakout year. This is despite recording a negative relative shot attempt percentage in his rookie season. In his breakout year, he saw increases in 19 of the 20 categories mentioned earlier. St. Louis’ Jaden Schwartz posted a –1.25% CF% Rel one season before seeing an increase in ice time of 2.08 TOI/gm. He was able to favorably impact all 20 categories when given a larger role.

In conclusion, a high CF% Rel should not be the only factor when deciding if a player can take on a larger role, but it may be an indicator for a player who is able to produce at a higher rate when given more ice time.

Follow Steve Ness on Twitter: @QuickkNess

All data from War-On-Ice.com.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s