Join AI Pro

With the 2018-19 season underway, there is no shortage of betting sites and prediction models that tell us which teams are likely to finish first in the Premier League (PL). At the same time, there is no shortage of pundit and fan opinions on what it would take to win the title. One of the more common refrains I’ve heard recently, especially on social media, is that it is the offense that wins games, but it is the defense that wins titles. You frequently hear this adage for other sports, not just for PL football. But is it true?

What does it really take to win the PL title? If we look back at the 26 winners, going all the way back to 1992-93, do they have common characteristics besides mostly being based in Manchester or London? Unfortunately, I have no access to detailed statistics going all the way back to the first PL season, and certainly no access to advanced analytical models (such as Expected Goals, for example). But I was able to obtain the final PL season tables from the wonderful www.fbref.com site and decided to take a look.

I examined every season since 1992-93, and asked: Where did the team with the best defense (i.e., the fewest goals allowed – GA – for the season) finish? And how often did the team with the best GA win the title? Well, the answer is, fairly frequently. Of the 26 seasons, the team with the best GA won the title 11 times (42%), finished second 9 times (35%), finished third 3 times (12%), finished fourth 2 times (8%), and finished fifth 1 time (4%). So, while the team with the best defense won the title 42% of the time, it failed to win it 58% of the time. Is that good? I then looked at the teams with the best offense.

Of the 26 seasons, the team with the best offense (i.e., most goals scored – GS), won the title 16 times (62%), finished second 6 times (24%), finished third 2 times (8%) and finished fourth 2 times (8%). But when I looked at the teams with the best goal differential (GD), the numbers were most convincing. The team with the best GD won the title 18 times (69%), finished second 7 times (27%), finished third 1 time (4%) and never finished fourth or worse. Which team finished third with best GD? 15-16 Tottenham.

Now it is clear that these are not mutually exclusive – meaning a team with the best GD may have the best GS, or the best GA, or both. In fact, 8 times in PL history the team with the best GD had the best GS *and* the best GA. 6 of these 8 teams (75%) won the PL title (these teams were 00-01 Man United, 03-04 Arsenal, 05-06 Chelsea, 07-08 Man United, 11-12 Man City, and 17-18 Man City). The other 2 teams finished second (97-98 Man United and 16-17 Tottenham). So while having the best GD (ideally with the best GS and GA) will not absolutely guarantee that you will win the PL title, it will give you the best odds of finishing first. And having the best GD, even without having either the highest GS or the lowest GA, still gives you an excellent chance of winning the PL title, as it happened 12 times in 26 seasons (46%).

Also, if you go back to the teams with the lowest GA in each of PL season, and remove the 8 teams that also had the best GD, the correlation between having the lowest GA and winning the PL title is worse. In the 18 remaining seasons, the team with the lowest GA won the title only 5 times (28%). On the other hand, in these same 18 seasons, the team with the highest GS won the title 10 times (56%). So, having the best GD – ideally with highest GS and lowest GA – is optimal. After that, having the highest GS total seems to be better than having the lowest GA total. So, apparently, it’s the offense that wins PL titles.

Now that we know that having the best GD gives you the highest chance of winning the Premier League, how high of a GD do you really need to have a reasonable shot at the title? In the 26 PL seasons, the title winners averaged GD of 48.1 per season, the lowest being 32 (96-97 Man United, 14-15 Leicester City) and the highest being 79 (17-18 Man City). If you look at the last 10 seasons, which basically cover the period since Man City became a rich club and Tottenham began competing above their wage level, the PL title winners averaged GD of 53.2 per season. Winning the Premier League has become much harder.

Based on this observation, and consistent with Twitter exchanges I have had with Dan Kennett, I believe that GD of 50 is the minimum required achievement to be competing for a title. In the PL era, there have only been 16 teams with GD of 50+ (chart). 10 teams won the title, the other 6 finished second. 3 of the 6 teams that finished second, unfortunately, faced a team with a better GD and more points in that exact same season (04-05; 09-10; 13-14). In 11-12, the two Manchester teams had the same points but City won with the superior GD. In 16-17, Tottenham had superior GD but lost to Chelsea. And, alas, in 08-09, Liverpool, with GD of 50, lost to a Man United team that had inferior GD (44) but 4 more points. In fact, Liverpool have twice finished second in seasons where they had GD of 50 or more (in 08-09, 13-14). But Man United experienced the same disappointment twice as well (in 09-10, 11-12). Arsenal had the same heartbreak, albeit to a great Chelsea team (in 04-05), as did the above mentioned Tottenham in 16-17.

So, will GD of 50 be sufficient to win the PL title in 18-19? Looking at the GD totals achieved by the top teams over the last few years, and given how strong Man City were last season and still are this year and how strong the other top four teams look, especially Tottenham and Chelsea, we can safely assume that GD of 60 or better may be required to finish first this season. In fact, Man City achieved GD of 79 in 17-18 (the best in PL history) and Tottenham finished second in 16-17 with GD of 60. For Liverpool, the task may seem daunting. After all, only twice in their Premier League history has a Liverpool team had GD of 50+ (in 08-09, 50; and in 13-14, 51). And those are usually considered the best seasons the team has had in the Premier League (in their wonderful treble winning 00-01 season, Liverpool finished third). So this season, Liverpool will have to greatly outperform their best two Premier League seasons, if they are to achieve the 60+ GD that I believe will be required for a proper title run. Can they do it? I am hopeful.

Incidentally, Liverpool may be starting this season from a much stronger position than their fourth-place finish in 17-18 may imply, even ignoring the much-improved squad depth that the summer arrivals have brought. Last season, Liverpool had the second best GD (46) yet finished fourth, behind Man United (40) and Tottenham (38). In fact, that was the first time in the history of the Premier League that a team with GD of 45+ failed to finish first or second (chart). For comparison, in 14-15 Chelsea finished first with GD of 41, while Man City finished second with GD of 45. And in 15-16, Leicester won the title with GD of 32, while Arsenal finished second with 29 and Tottenham finished third with 34 (that was a fluke season).

The main reason Liverpool collected only 75 points and faced a “must win” game on the final day of the season to lock a top-four finish, was that they had too many draws (12), especially at home. It’s actually unusual for a team with such a high GS total, and such a high GD, to have so many draws. In fact, the last team to have GD of 46 or more and 10 or more draws in the same season was… Rafa’s 08-09 Liverpool. But while that team fell apart in the 09-10 season, Liverpool fans can be optimistic that the current team is stronger and deeper and more talented and more experienced than last year’s squad, and can have a proper fight for the PL title and Cup trophies. After all, going from GD of 46 to GD of 60, while difficult, is not impossible. Score 6 more goals (to go from 84 to 90) and allow 8 fewer goals (to go from 38 to 30), and you are in the thick of a title race in May. As Simon (Brundish) says, #BeatTheDrossWinTheLeague.

Join AI Pro