I have previously broken down the ways a leading team played in a Score Effects situation last season. We are now ready to look at the ways teams played when trailing 2 or more goals in a game. The theory of Score Effects tells us that teams get more aggressive offensively when trailing by 2 or more goals in the first and second period and even when down by one in the third period. The theory also posits that most teams go into a defensive shell in this situation or at the very least take less chances on offense thereby decreasing their shot output.
In order to determine how the trailing team plays, shot rates including Shots For, Corsi For and Fenwick For are helpful. When a team was down 2 or more goals, these rates can be compared to the same rates for each team at Score Close. A comparison of a trailing team’s sh% (Shooting Percentage) when down 2 and at Score Close is also useful.
2013-14 Regular Season
As a whole, teams down 2 or more goals gave up shots against at a lower rate than they did at Score Close.
CORSI AGAINST PER 60 AT SCORE CLOSE AND DOWN 2+ GOALS
Teams that gave up more shots at Score Close showed larger changes in their rates generally, but also remained near the higher end of the spectrum in terms of allowing offensive pressure by their opponent in those situations.
The Buffalo Sabres recorded the largest change in CA60 from Score Close to Down 2 or more goals last season. Their Score Close CA60 was the 2nd highest in the league; however, when they were Down 2 or more goals, their CA60 was only the 9th highest in the league.
Teams that were normally fairly stingy about giving up offensive pressure when the score was close remained so when Down 2 or more goals. These teams are at the lower end of the graph.
SHOT BLOCKING PERCENTAGE
A look at shot blocking percentage when Down 2 or more goals supports the theory that trailing teams became more offensively aggressive. With the exception of Carolina and Winnipeg, all teams showed a decrease in the percentage of shots they blocked when trailing. Some of the changes are rather small and thus suggest those teams did not make major adjustments to their style of defensive play, such as Dallas, Chicago, Colorado, New Jersey, Florida and Detroit. Calgary, which had the highest shot blocking % at Score Close, remained at the top of the list when trailing as well.
Teams such as Boston, Minnesota, San Jose and a few others showed rather drastic drops in their shot blocking % when trailing. There are few reasons that may be behind this. It could represent a defensive systems approach that does not stress shot blocking. It could also be caused by the type of offensive pressure the opponents were able to generate at that time as well. If the team’s offensive structure was such that the leading team was not able to get a cycle going, but instead was getting breakaway attempts or odd man rushes, it would make sense that there was not much opportunity to block those shots.
CA60 DEVIATION FROM AVERAGE AND DIFFERENTIAL FROM SCORE CLOSE
The graph above gives a more complete picture of the offensive pressure allowed when trailing. Teams to the left gave up shot attempts at a rate below the league average. Teams to the right gave up shot attempts at a rate higher than the league average. Teams near the bottom of the graph had a fairly large difference in their CA60 at Score Close as compared to that rate when Down 2 or more goals, i.e. they had the largest change in rate. Teams near the top did not have as large a change in shot attempts against rates from Score Close to Down 2 or more goals.
We see from the plot that teams such as Detroit, L.A., Chicago, Boston and New Jersey did not have a very large change in CA60, but are still on the right side of the average line in terms of shot suppression. The teams that showed bigger changes and managed to fall on the left side of the average line, were effectively adapting to the score situation in order to suppress the offensive pressure of the leading team.
Teams just over to the right of the average line were likely quite close to having a more effective response to the score situation because they were working to suppress offensive pressure. Teams farther to the right had a much bigger hill to climb. Buffalo drastically reduced their CA60 but still ended up a bit above the league average. Edmonton, Toronto and Montreal showed a large drop, but were still above the league average as well. This likely had a lot to do with the weaknesses of the defensive systems of these teams.
We can also look at how these teams performed by using SA60 (Shots Against Per 60), GA60 (Goals Against Per 60) and Save Percentage.
SHOTS AGAINST PER 60, GOALS AGAINST PER 60 AND SAVE PERCENTAGE DEVIATION FROM AVERAGE DOWN 2+ GOALS
This graph shows the deviation of each team from the average of each metric used. Because these are against rates, being lower than average in shots and goals is desirable. Being above average in Sv% is still better here of course. The circles represent the deviation from average of the team’s Sv% when Down 2 or more goals. Circles in white are negative deviation, or below average and those in blue are above average.
The first thing that jumps off of the page here is how poor Pittsburgh’s Sv% was when down last season. This contributed to the team’s poor GA60 as well obviously. Despite being two of the better teams in terms of SA60, both Chicago and Tampa Bay had poor showings in terms of SV% as well; however, their shot suppression was enough to keep them very close to the average line for GA60.
Phoenix had the best Sv% when Down 2 or more goals, was nicely below the average on GA60 as a result and was right on the line in terms of SA60. Another interesting result on this graph is Toronto’s position. Obviously we expect Toronto to be on the high end of shots allowed, but they had very good goaltending which helped them stay right around average in terms of GA60. Both Anaheim and Columbus had goaltending slightly below average, but due to being better at shot suppression when trailing, were able to maintain a position below the average in goals against rate, while teams with similar goaltending were above the average in GA60 because of a high frequency of shots against.
Now that we know which teams excelled and struggled with reducing the offensive pressure of the leading team, we can look at how they tried to get out of their goal deficit.
CF60 SCORE CLOSE, DOWN 2+ GOALS AND DIFFERENTIAL
Eastern Conference CF60 Score Close, Down 2+ Goals and Differential
Breaking this into 2 graphs makes it easier to see and appreciate the differences in the offensive pressure of the trailing teams. In the Eastern Conference, Carolina showed the largest change in CF60 when down in the score column. They are often a high event rate team at Score Close so the extra gear they found was fairly impressive. Toronto also showed a fairly significant increase in their shot attempt rates when Down 2 or more goals.
Only three teams showed a decrease in their overall shot attempt rate here, namely New Jersey, Pittsburgh and Boston. Boston had a high CF60 at Score Close this past season so this decrease when down in the score is a bit surprising. It may be explained by the personnel used when trailing in a game or the result of a certain style of offense employed at that time in the game.
New Jersey had the lowest event rate in the league at Score Close so it is not surprising to see a decrease in their CF60. Pittsburgh was also rather low in the event rate category so a decrease for them is not surprising even if it is a bit perplexing. The highest event rate team in the league, Ottawa, showed only the slightest increase in CF60 when Down 2 or more goals.
In the Western Conference and, when viewed with the first of these graphs, in the entire league, Chicago led the way in terms of not only CF60 when Down 2 or more goals, but also in terms of the team’s differential from Score Close. Chicago had a whopping 17.2 increase in CF60 from Score Close to Down 2 or more goals. Anaheim and Nashville also showed significant increases, although it should be noted that Nashville is usually a low event rate team. It would seem that they opened it up quite a bit when down in a game. So, while these teams tried to pile on the offense to get the game closer, were they making any headway?
BLOCKED SHOTS PERCENTAGE
Taking a quick look at which teams ended up with more of their shots blocked than others, we see that when the score was close, Philadelphia led the league last season in having the highest percentage of their shots blocked. When Down 2 or more goals, it was Montreal leading the way there. The tracking line is in there at 28.00% to give an idea of how much each team’s blocked shots percentage changed due to score situation. Because getting shots through or avoiding having shots blocked has been shown NOT to be a measurable skill, this probably says more about the opponents these teams faced when trailing than the teams’ shooters.
To get a better idea of how effective these more aggressive offenses were, we can use SF60 (Shots For Per 60), GF60 (Goals For Per 60) and sh% (shooting percentage).
SF60 GF60 SH% DEVIATION FROM AVERAGE
This first graph shows the deviation from average for each team. Teams above the horizontal axis were above average in GF60 and teams below were, you guessed it, below average. The teams with the above average GF60 were the same teams with the above average shooting percentages regardless of their position relative to the average in terms of SF60.
SF60 GF60 SH% DIFFERENTIAL FROM SCORE CLOSE
This becomes even more interesting when we look at these same metrics and measure not the deviation from average, but the difference in each from Score Close. The circles represent the differential of the team’s shooting percentage from Score Close, or how much the sh% changed when the team was Down 2 or more goals from what it was at Score Close. Blue circles indicate a higher sh% when down while white circles represent teams whose sh% was lower than their percentage at Score Close.
Teams higher on the graph scored at a higher rate (GF60) when Down 2 or more goals than they did when the score was close. Teams lower on the graph had a lower GF60 when trailing. Teams to the left of the vertical axis had a lower SF60 when trailing. As you can see, most teams had a higher SF60 in this score situation. The interesting thing about this way of looking at team play is that teams with even a slight drop in sh% tended to do better (i.e. have a higher GF60) as their SF60 increased. Essentially, teams such as Washington, Columbus, Vancouver and Anaheim compensated for a slight decrease in sh% from Score Close by increasing their SF60.
Chicago had the highest differential in SF60 from Score Close. When that was combined with a very small negative change in sh% (0.16%), Chicago remained well above their Score Close mark for GF60 when Down 2 or more goals. Carolina also showed a dramatic increase in SF60 in this situation; however, a drop in sh% of 1.97% was too much of a detriment to allow the team to score at a higher rate when trailing. This significant drop in sh% explains why teams who increased their SF60 still failed to increase their GF60 when trying to come back in a game.
CF60 AND CA60 DIFFERENTIAL WHEN DOWN 2+ GOALS FROM SCORE CLOSE
In the graph above, the teams regularly good at shot suppression did not show a very big decrease in CA60 when trailing, because frankly, they did not allow many shots against at Score Close. The teams that did allow more shots at Score Close naturally showed the biggest change when they were Down 2 or more goals and trying to get back in to the game. L.A., Boston, New Jersey and Pittsburgh decreased shots against but also decreased their CF60 at the same time, while Chicago and Nashville decreased shots against and increased shots for when trailing. Some teams that were poor at shot suppression continued to struggle with that while trailing, such as Colorado, New York (Islanders), Florida and Calgary.
When we look at how teams played from a Fenwick perspective, the results are similar.
FF60, FA60 DOWN 2+ GOALS
Once again, New Jersey led the way in suppressing shots, but it is interesting to note that not only was Chicago first in FF60 when trailing, but the team also suppressed shots more than any others (except New Jersey) in that situation. Winnipeg showed a common theme here again as well. They did a pretty nice job in terms of shot suppression when trailing last season.
To get a final lay of the land, I have split the Eastern and Western Conferences (simply to have a less crowded view) into two graphs with the teams’ Goals For, Goals Against and Time On Ice playing Down 2 or more goals.
Now that we can see how the teams fared in general, we can put this all together and get an idea of how effective a particular team was last season when trailing by 2 or more goals by looking at sh% and Sv%.
SH% SV% DEVIATION FROM AVERAGE DOWN 2+ GOALS
While changes in offensive style, i.e. more shots, may have helped overcome certain issues teams had, the results of those endeavors, GF%, played out in fairly close correlation with sh% and Sv%. This really isn’t a big revelation, but it is an easy way to gauge which teams ended up making a dent in those 2 goal deficits.
PDO (SH% + SV%) AND GF% DOWN 2+ GOALS
Pittsburgh’s struggles were evident in their GF%. While they suppressed shots against when trailing, they simply did not mount enough of an offensive attack to overcome some truly woeful goaltending numbers. Chicago was one of the best shot suppression teams in the league last season but saw a rather precipitous drop in Sv% when trailing. The team’s sh% remained almost the same as at Score Close and they piled on the shots, thus boosting their GF% when Down 2 or more goals and giving us the most notable outlier on this graph. Phoenix had an extremely high Sv% and sh% when trailing and this very obviously resulted in an excellent GF%. Calgary also had a high PDO and benefited from that.
In conclusion, while playing from behind in the score usually led to a more aggressive offensive attack in order to get back in the game, more shots alone did not guarantee a comeback. The trailing team’s goaltending was also extremely important in that situation. Further, if a team’s sh% did not increase while mounting the comeback, it at the very least could not decrease very much if the comeback was to be successful. Increasing shot volume did help to overcome some of those problems so long as the sh% decrease was not substantial.