Everything Else

Score Effects Pt. 1: Protecting A Lead

Score Effects. You may be familiar with this term, but for the uninitiated, a brief explanation. Score Effects is a term used to describe the way teams play in various score situations at full strength (5v5) during a game. When the game is tied or at “Score Close” we tend to see a truer measure of a team’s real strategy and playing ability. Score Close is defined as full strength play when the game is tied or within 1 goal in the first or second period. In the third period, Score Close is a tie game.

When the score is not close, i.e. a team has a 2 or more goal lead in the first or second period, we often see Score Effects come into play. Often when a team has a lead like this, it will play more conservatively. The team goes into a defensive shell to protect the lead. This is usually with the aim of reducing risk or playing it safe. The team down by 2 or more goals wants to bring the score even so it will often take more risk and play a more aggressive style offensively.

These two strategies combine to result in the trailing team making a big offensive push and the leading team struggling to get out of its defensive zone. Essentially, the leading team ends up being bombarded with shots, which we can all agree is far more risky than cycling the puck in the offensive zone, so many people feel that going into a defensive shell with a lead is a mistake. Regardless of the wisdom behind such a strategy, it happens very regularly and with profound effects on the shot totals in a game. This is why you will often see different shot based metrics include a note that the data is from a Score Close situation so it is not artificially inflated or deflated by Score Effects.

While doing some research, I decided to look at shot rates for different teams in a Score Effects situation. If some teams are better at 5v5 play than others, certainly the same must hold true of teams trying to dig themselves out of a hole or keep their opponent from doing so. Initially, I wanted to cover this in one post; however, I came to my senses and realized that would be foolish.

For the first installment in what should be a two or three part series, we will focus on the team playing with the lead. To do this, we will look at shooting and possession metrics for a team when Up 2 or more goals and compare them to how the teams perform at Score Close. When doing a team specific analysis, many more factors such as player usage and deployment, etc. would be considered; however, because the purpose here is to get an overview of the league that type of in-depth team specific analysis is not particularly practical.

*I have limited the analysis here to the 2013-14 regular season, but I intend to add other seasons in the future to see the correlation with specific systems as well.*

2013-14 Regular Season Leading 2+ Goals

When we look at the difference between how many total shots (Corsi) a team gives up when the score is close and compare that to the same mark when the team is leading by 2 goals or more, we start to get a picture of which teams are more susceptible to Score Effects. It is unsurprising that teams that tend to give up a lot of shots at Score Close also tend to give up a greater number of shots when they have a lead.

ca60 close up2

The teams known for being strong in shot suppression are at the lower left of the graph. These teams give up very few shots throughout their even strength systems. The New Jersey Devils actually surrendered shots against at a slightly lower rate this past season when they had a lead than at score close. This is the only team to end up that way although, the Winnipeg Jets and L.A. Kings both finished with Up 2+ Corsi Against rates less than 1 shot per 60 more than their Score Close rates, i.e. very little change in their CA rates.

The teams at the top right of the graph tend give up shots at a high frequency at even strength regardless of having a lead or a close score. The Nashville Predators showed the biggest change in their CA60 rate from Score Close to Up 2+ with an increase of 11.9. The Predators were followed closely in this jump in shots allowed based on the score by the Florida Panthers (11.4) and the Tampa Bay Lightning (10.7).

When we look at the shot rates from a Fenwick perspective, the situation gets a bit more complicated. Unlike Corsi, which includes shots on goal, missed shots and blocked shots, Fenwick excludes blocked shots.


The Corsi graph had an easy to follow look to it with a correlation (R2) of .8062. The Fenwick graph is very similar. We still see the good shot suppression teams like New Jersey, L.A., St. Louis and Chicago at the far left of the graph and the poor shot suppression teams such as the Edmonton Oilers, Buffalo Sabres and Toronto Maple Leafs at the right.

Upon closer inspection, we see that when Up 2 or more goals, a few teams actually allow Fenwick (unblocked) shots at a lower rate than when the score is close. These teams include Minnesota, St. Louis and Winnipeg. The theory of Score Effects tells us that teams allow more overall shots when they have a big lead, so why are there teams allowing unblocked shots against at lower rates?

To figure this out, we can compare the CA60 rates for each team at Score Close with the FA60 rates to see what their shot blocking rates were in a normal game situation. We can then compare those rates when the team is Up 2 or more goals.


We would expect that the changes in Corsi Against and Fenwick Against rates from Score Close to Up 2+ to be uniform if their tactics stayed static. Essentially, if the teams were approaching the game in the same way defensively when the score was close and when they had a lead, we would expect their shot blocking rate to show an increase consistent with the increase in shots being taken against them.

To find out if this was the case, we can compare the CA60 differential (Up 2 – Score Close), the FA60 differential (Up 2 – Score Close) and the Shot Blocking Rate differential to see if there is any consistency.

ca60 fa60 blocking

On the graph above, the teams showing the smallest change in their CA60 rate from Score Close to Up 2 or more goals are on the left. Those values increase as we move to the right. For example, New Jersey had a negative differential, meaning that the team actually yielded shots at a lower rate when playing with the lead. Winnipeg has almost no change in their CA60 from Score Close to Up 2 or more. Nashville showed the biggest increase in their CA60 when leading with Florida and Tampa Bay right on their heels in this regard.

When the differential in CA60 and FA60 are plotted together, the correlation is very strong (R2 0.7758). The higher the team appears on the graph, the higher its FA60 differential. Again, Nashville showed the biggest change in unblocked shots allowed when Up 2 or more goals as compared to the team’s rate at Score Close. New Jersey’s FA60 differential was a miniscule 0.1. Adding the Shot Blocking Rate differential in for each team, we see that there is little correlation between the increase in CA60 and a change in a team’s Shot Blocking Rate for the league as a whole. Each team seems to approach shot blocking with a lead a bit differently.

To get a better idea of which teams make a concerted effort to elevate their shot blocking when they have a lead, we can convert the teams’ shot blocking numbers to a percentage and compare them in different score situations.

shots blocked percentage

On the left are the teams with a lower shots blocked percentage at Score Close with those having a higher percentage to the right. The teams near the bottom of the graph have a lower percentage when playing with a lead of 2 or more goals. Those near the top have a higher percentage with a lead. The circles that are blue reflect those teams having a higher shots blocked percentage when playing Up 2 or more goals as opposed to their percentage at Score Close. The white circles represent teams with a lower shots blocked percentage when Up 2 or more goals than when the score is close. The numbers in the circles are the percentage change from Score Close.

Knowing this, we can see that the team showing the largest jump in shots blocked percentage when playing with a lead is the Minnesota Wild (+5.07% increase). The Vancouver Canucks, Philadelphia Flyers, St. Louis Blues, Winnipeg Jets, Chicago Blackhawks, Buffalo Sabres and Boston Bruins also show a substantial increase in shots blocked percentage when Up 2 or more goals. On the negative side of things, the New York Islanders displayed the largest decrease in shots blocked percentage, followed by the Nashville Predators, Calgary Flames and New Jersey Devils.


While some teams show dramatic drops in shots blocked percentage when leading, negative effect on their FA60 Differential is not overly obvious for some of them. It is fairly apparent for teams such as Nashville, New York (Islanders) and others. The change in shots blocked percentage did seem to benefit the teams showing a more marked increase as seen in the teams near the upper left of the graph.

Now that we know how teams performed in terms of shots against when protecting a lead, it is interesting to look at how their offensive approach changed in the same situation. To do this, we can again examine the difference in a team’s Corsi and Fenwick attempts at Score Close and Up 2 or more goals.

cf60 close up2

In the Corsi For graph above, we see the nice orderly grouping that was apparent in the Corsi Against graph. The only real outlier is Philadelphia, which oddly enough had a higher CF60 (Corsi For Per 60) rate when Up 2 or more goals than when the score was close. Ottawa and Dallas sneak in near the top right of the graph with the dominant possession teams, which may be a surprise until you consider that those are the Eastern Conference and Western Conference’s two highest event (shot) rate teams. Both teams give up a lot of shots but take a lot of shots too so it should not be a surprise to see them here. Likewise, Minnesota, New Jersey, Nashville and company are relatively low event rate teams so despite the fact that they suppress shots well, it is expected that they will show up near the lower end of the CF60 spectrum.

ff60 close up2

When we look at the Fenwick For graph, the teams are a bit more spread out, but still in a rather nice grouping with the teams in the same locations relatively speaking. Where we are able to see the real differences in how a team plays with a lead is when we quantify the change in their shot rates.


In order to show the change in a team’s approach to offense when trying to protect a lead, we must find the difference in a team’s CF60 (Corsi For Per 60) at Score Close and Up 2 or more goals. Again, only Philadelphia had a higher CF60 when protecting a lead than when the score was close. Teams closer to the right of the graph had a smaller difference in how frequently they took shots when Up 2 or more goals. The farther to the left of the axis on the graph, the bigger the change in the team’s CF60. It is important to note that several of the teams showing a rather low differential in CF60 are also not particularly aggressive offensively when the score is close either.

Some teams do show a more dramatic drop in offensive push when protecting a lead. We often hear the phrases “taking their foot off the gas” or “sitting on a lead” when this happens. Vancouver demonstrated the most dramatic drop in their CF60. When sh% (shooting percentage) for each team is taken into account at the different score situations, many teams saw an increase while Up 2 or more goals. Logically, this makes sense due to the lower number of shots. The New York Islanders showed a rather precipitous drop in sh% in this scenario falling 2.66% lower than their Score Close mark.

sfsa diff gfga diff up2

To see the concrete effects of this change in a team’s style of play, we look to shot rates and goal rates in the graph above. Shots as defined for the purpose of this data set, are shots on goal and exclude blocked and missed shots. Teams to the left took fewer shots on goal than their opponent. Teams to the right took nearly the same or more shots on goal than their opponent.

The other data in this plot shows how teams fared in terms of goal rates. Teams above the axis scored goals at a higher rate than they gave up when they were Up 2 or more goals. Teams below the axis gave up goals at a higher rate than they scored them in that situation. Calgary showed the biggest negative differential in goal scoring and also had the biggest change in Sv% (Save Percentage), dropping 4.44% from Score Close.

shot goal save diff overall

There are several teams that took more or nearly equal the rate of shots on goal as their opponents when Up 2 or more goals. The four teams with a higher SF60 than SA60 in this score situation were St. Louis, L.A., Boston and Chicago.


Including only the eight teams who were nearly equal or higher in terms of shot on goal rates than their opponents, it is interesting to note that some of the teams still fell behind their opponents in terms of goal scoring rates. The answer to why this happened may be clearer when the team’s Sv% is added in to the equation. L.A., St. Louis and San Jose all showed an increase in Sv% and ended up on the positive end of the goal scoring rates. Boston showed a small drop; however, given that they still largely out shot their opponents, were still in the positive goal rates area.

New Jersey showed a small increase in Sv% but it was not enough to overcome the shot on goal rate differential as they finished with a goal scoring rate slightly below their opponents. Winnipeg and Detroit were taking shots on goal at a slightly lower rate than their opponents and finished with lower goal scoring rates as well. This looks to be largely due to drops in Sv% of 1.44% and 3.88% respectively from Score Close.

Of the teams with a higher shot on goal rate than their opponents when playing Up 2 or more goals, only Chicago showed a lower goal scoring rate. It is not a coincidence since the team’s Sv% dropped by 2.67% in that situation. Given how close the differential was between the goal scoring rate of Chicago and their opponents, it is logical that had the Sv% held up anywhere near the Score Close Sv%, the team would have finished with a positive mark in this category.

This begs the question of whether the team’s increase in shot blocking played a role here. Goalies have often bemoaned aggressive shot blocking because their teammates end up screening their view of incoming pucks, so it is possible that this contributed to the drop in Sv% for Chicago. Additionally, Crawford struggled with perimeter shots throughout the 2013-14 season. He made the third highest number (836) of saves on shots from perimeter locations {trailing Lundqvist (929 saves) and tied with Rask} but also gave up more goals on perimeter shots (31) than other starting goalies in the league {followed by Mason (26), Lundqvist (24)}. All of this information lends credence to the theory that Chicago’s increased shot blocking rate when leading by 2 or more goals may actually have adversely affected the team’s Sv% and collaterally their goals for and against rate differential, making this an issue worthy of further exploration.

Final Observations

Now that we have an overview of how all of the teams in the league play when trying to protect a lead, a few final thoughts.

  • Reducing offensive pressure or lacking the defensive prowess to limit the offensive onslaught of the trailing team leads to giving up goals at a greater rate than your team is scoring them unless it is accompanied by a significant increase in Sv%.
  • Teams that were usually good at suppressing shots when the score was close, continued in that strength when protecting a lead.
  • Some teams showed a meaningful increase in the rate at which they blocked shots when Up 2 or more goals which appears to be a strategic approach to their defensive structure.
  • The vast majority of teams showed some decrease in their shooting rates when protecting a lead. Only four teams continued to take more shots on goal than their opponents when protecting a lead of 2 or more goals: Boston, Chicago, L.A. and St. Louis.
  • Some teams were far more likely to succumb to Score Effects than others. Those teams both backed off the offensive pressure and allowed a significant increase in their opponent’s offensive pressure.

Below is a graph of the “gross” change for the leading team due to Score Effects as shown through the overall change in Corsi (CF60, CA60), Fenwick (FF60, FA60) and Shots On Goal (SF60, SA60) for each. Teams with a higher “gross” change are those most significantly reducing their offense while allowing more pressure from their opponent’s offense and thus falling victim to Score Effects.

gross change score effects

In the next installment of this series on Score Effects, we will look at how teams play when Down 2 or more goals to get the other side of this story.

Leave a Reply

Your email address will not be published. Required fields are marked *