Part 2: Boring, Boring Chelsea?

In Part 1 of this analysis, we took a critical look at Michael Cox's idea that the Premier League has traditionally been defensively-oriented, and only gained an offensive mindset in recent years. Now that the season has concluded, let's look more closely the author’s specific contention about Chelsea: that they "aren't a particularly defensive side."

Let’s test this assertion by measuring "defensive" in a few different ways, and compare the 2014-15 version of Chelsea to the Premier League as a whole, over its entire 23-year history. In some cases, since the "Chelsea is a defensive team" narrative often assigns credit (or blame) for the team's playing style to these two men, we will also look specifically at Chelsea in the years since Roman Abramovich, Russian billionaire, purchased the club (2003-04 forward), and/or the years during which Jose Mourinho was/is the manager (2004-05 season through 2006-07, and 2014-15).


For a base-level metric, we can simply say that teams who concede fewer goals are more defensive than other teams.

In the chart here we see all 466 team-seasons in the history of the Premier League, with poor defensive teams towards the top of the plot.  Chelsea teams from various seasons are highlighted in blue. 

In the 2014-15 season, Chelsea only conceded 0.842 goals per game, which was the best rate of all Premier League teams this year. (The median goals-allowed in all games in Premier League history is 1.316, and has been trending slightly higher over time. )

That level of stinginess is in the 90th-percentile for defense, being the 30th-best goals-against mark of all time. This is a laudable rate, but prior iterations of Chelsea are much stingier. There are six years in which their goals-against rate is superior to the 2014-15 season, including the best defensive season ever (2004-05, 0.395 GA/G) and four of the top six (2005-06, 0.579 GA/G; 2006-07, 0.632 GA/G; and 2008-09, also 0.632 GA/G).

Let's look at goals-allowed on a historical team-by-team basis, splitting it out so that we can easily see the difference between pre-Abramovich Chelsea and modern-day Chelsea.

You might notice a certain shape to these charts: teams that allow a lot of goals per game have decidedly shorter stays in the Premier League than teams that allow few.

It seems that this particular metric of defense is a key factor in a club's long-term survival and success in the Premier League...and Chelsea, the poster child for not allowing goals, has finished no worse than 6th since 1997. 

Therefore, if we were to use "goals allowed" as our measure of defensiveness, not only is Chelsea a "particularly defensive" side in 2015, but Chelsea, generally speaking, are THE defensive side in the Premier League--particularly in the Abramovich/Mourinho era.


We can also say that teams that tend to play in low-scoring games, taking into consideration goals scored by both teams, are "defensive" teams. This is because teams that are risk-averse--i.e., teams that emphasize goal prevention over goal scoring--will both score and concede fewer goals than teams that play a wide-open style.

Obviously, high scoring games can result from:

  • a team's utter dominance of its opposition;
  • two teams who habitually play an attack-oriented game;
  • two teams who have equally unsound defenses; or 
  • one team's sheer inability to stop its opponents.

Nevertheless, a team that is risk-averse is not going to have high-scoring games on a regular basis. 

  • Against inferior opponents, they will take a lead, and then close up shop to defend the three points in the standings that a win is worth.
  • Against superior or evenly-matched sides, they'll start out in a defensive position, and only attempt to score if they fall behind or have an obvious counter-attacking opportunity.

Let's see if Chelsea can still be considered a "defensive" side by total goals by both sides per game, which we'll call "action."

In the chart here we see all 466 team-seasons in the history of the Premier League; teams with higher action scores are towards the top.  Again, Chelsea teams are highlighted in blue.

By action, the idea of Chelsea as a defensive side completely disappears. Fans who watched this season's Chelsea matches were not subjected to a series of draws and low-scoring affairs. In fact, they saw an average of 2.763 goals per game--the fifth-most in the league, and more than 2/3 of all Premier League teams in history. (The median action score for the Premier League all-time is 2.605.)  

Of course, Chelsea doesn't play such high-scoring games on a regular basis, season after season. But neither do they play the “park the bus” style of defense purportedly favored by Jose Mourinho. Historically speaking, Chelsea is pretty much right in the middle of the league table when it comes to playing a crowd-pleasing style. They've never had the lowest action score in a season, but twice have had the highest, including in the 2009-10 season, when they won the League with the 4th-highest action score in history.

So as much as the "goals allowed" metric says that Chelsea is the quintessential defensive team in the Premier League, the "action" metric says they're actually quite balanced...and that's because they're also one of the best goal-scoring teams in the league, and have been throughout the Premier League's history.



An interesting sidebar about the "action" metric: the team with the highest action rank in each season has a dramatically increased chance of finishing at the very top or the very bottom of the league. 

In the 20-team Premier League (with the exception of the first three years, when 22 teams competed), 20% of the teams will finish in the coveted Champions' League qualifying spots, 1st through 4th. 15% of the teams, slots 18 through 20, will be relegated. That leaves 65% of the league--slots 5 through 17--to finish in mid-table, with no appreciable benefit or drawback other than surviving to compete in the League for another year

However, the top-ranked action team only finishes in a mid-table position about 30% of the time, or less than half as often as an average team.

Conversely, the bottom-ranked action teams over each season of the Premier League have been relegated no more frequently than an average team, but finish in mid-table 72% of the time, and have never won the league.

So, hypothetically speaking, say you have a team that is a middle-of-the-pack Premier League side, measured by an objective assessment of the talent of the players, and of your budget. Currently you play a balanced style of soccer, and you're trying to decide if you'd be better off if you switched your philosophy to a more extreme style.

  • Switching to a defense-heavy blueprint has no appreciable upside; your chances of being relegated are pretty much the same, but your chance of qualifying for Champions' League decline. 
  • Switching to an offensive-heavy style offers reward, but carries a lot of risk: the chance of finishing in the top four doubles, but your chance of relegation dramatically increases as well.

For a hypothetical team such as this, switching to a more defensively-oriented philosophy doesn't make much sense.

Incidentally, Chelsea was the top-action team twice; they finished 6th (mid-table, but pretty high up) in 2000-01, and they won the league in 2009-10. They were never the dullest team in any season.  Now, let's look at another potential metric of defensiveness: draws.


It's possible that teams who have a disproportionate number of draws are more defensive, because it could indicate a strategic willingness to accept a tie rather than risk a loss in pursuit of a win.

The problem with this strategy is twofold.

1. It's boring.

Even hardcore fans wouldn't stand for such a dull strategy. Nobody goes to the stadium week in and week out to see his team strive for guileless, tactical draws, and no neutral observer would be long entertained by such a team. (For our stateside readers: I'm sure that Pittsburgh Steelers and Baltimore Ravens fans are happy to watch their respective teams battle each other to miserable 9-6 field goal battles twice a season, but the rest of America wants to die an agonizing death rather than put themselves through that kind of torture on purpose. ) 

Here and there, there are tactical reasons to play for a draw (e.g., you are on the road, playing a superior team, in the middle of a rough stretch of games, and/or any number of other possible conditions). But it can't be an every-week plan, because of the second problem with this strategy.

2. Mathematics.

There is a huge premium for winning a match, versus the penalty for losing it. The payoff for a win is +2 points. The cost of a loss is only -1 point. If you are in a tied game, you're better off trying to win the game, and risking that single point a draw is worth, if your chance of winning is at least 33%.  (Actually, the threshold is even lower, because some of the time you'll still end up in a tie game.)

Consider two teams after 10 games played in a season. One goes for broke, and often gets there, but also scratches out some wins; one plays very conservatively, beating very bad teams and losing only to outstanding ones.

  • Gamblers: 10 games played, 4 wins, 0 draws, 6 losses.
  • Pragmatists: 10 games played, 1 win, 8 draws, 1 loss. 

The Gamblers have lost 60% of their games, while the Pragmatists have been defeated only once in 10 matches. But the Gamblers are ahead in the standings, with 12 points (4 wins x 3 points) to the Pragmatists' 11 (1 win x 3 points + 8 draws x 1 point). 

The only teams who would logically benefit from playing for draws on a regular basis are teams that think their chances of winning any given match are significantly lower than 33%--that is, bad teams.

However, if bad teams are truly employing this as a tactic, they're not doing it well. The fact is, there's almost no correlation between "number of games drawn" and "final league position" or "points per game" or anything else indicative of overall success.

Here we can see, for instance, the final point totals for all 466 team-seasons in the Premier League. The effect on total points that "percentage of games drawn" has is negligible, compared to "percentage of games won" or "percentage of games lost." 

Basically, irrespective of quality, teams in the Premier League draw between 6 and 13 games a season, with an average of about 10. Teams that draw less tend to finish with more points than teams that draw more, but that's mostly because games that end in draws award only 2 total points, and games with a winner and a loser award 50% more. 


Moreover, teams can finish relegated, in mid-table, or in Champions' League qualifying positions with almost any number of draws. To be fair, league champions draw fewer games (8.0) on average than teams finishing 2nd-4th (9.1), and those teams draw fewer than relegated teams (10.1)...but mid-table teams actually draw more games than any of these groups (10.7).

As far as this relates to Chelsea specifically....well, we know that the Blues have only drawn more than 11 games once in the last 16 years, and they draw fewer games (9.74/year) than the overall Premier League average (10.24).  In 2014-15, they drew 9 games while winning the league title. Meanwhile:

  • The league averaged 9.3 draws per team.
  • Manchester City qualified for Champions' League with fewer draws (7) than Chelsea; Manchester United qualified for Champions' League with more (10).
  • Queens Park Rangers were relegated with fewer draws (6) than Chelsea; Hull City (11) and Burnley (12) were relegated with more.

The irrelevance of "number of games drawn" as a metric is best highlighted by this oddity from the 2014-15 season: Sunderland drew nearly 50% of their games (17), and finished 16th, surviving to compete in the Premier League's 2015-16 season; QPR drew less than 20% of their games (6)--almost three times fewer than Sunderland--and were relegated. The difference is that Sunderland won 33% of the games they did not draw; QPR only won 25% of theirs.

Historically, Chelsea were a mid-table team in the pre-Abramovich era, and a powerhouse team thereafter. Their chances of beating most teams they have played likely exceeded, or at worst approached, 33%. So for the most part, the math would say that Chelsea rationally has had no incentive to play for draws, and it does not seem like they have done so.


We've seen that Michael Cox is not entirely correct when he says that Chelsea are not a particularly defensive team. On the contrary, by using the metric of goals conceded, they were the most defensive team in the league this year; historically, they are responsible for four of the six most defensive seasons ever; and since the purchase of the team by Roman Abramovich, they are the most defensive club in the Premier League.

However, what they are not is dull. They traditionally, and specifically in 2014-15, play entertaining football in the sense of providing goal-scoring excitement for their fans and for neutral observers. They also do not, and never have, displayed a propensity to draw matches at an unusually high rate (not that there's any benefit to doing so). 

I think we have to conclude that Chelsea is not playing negative football, or putting out a substandard entertainment product in service of grimly marching toward a long-term goal. As a Tottenham Hotspur supporter, it pains me to say this, but even I have to admit that 2015 Chelsea, when you get right down to it, was simply an excellent, balanced team, succeeding on the merits of both their offensive and defensive skills, and worthy of being the league champions.


Mike Cisneros can be found on Twitter at @mikevizneros.
Mike Brinn (@brinn_official) contributed to this blog post.