The Over-Expectation League

A look at how teams are performing versus their start-of-season expectation

The current Premier League table. Take a good long look at it coz it’s probably not going to change for a while.

As the Covid-19 lockdown begins (unless you’re living in the UK of course), and seemingly all top level football suspended, we may begin to wonder if Harvey Barnes’ goal versus Aston Villa will be the last ever Premier League goal.

With no restart date on the horizon, it does give us a chance to look at the Premier League season so far and how teams are performing versus their pre-season expectations. Liverpool may be the undisputed champions-elect and are undeniably out-performing their pre-season expectations, but how do they stack up versus the other season’s high fliers?

It’s all relative

Friday, 9^th August 2019. On the back of a 9 game winning streak and 17 game unbeaten run at the end of the 2018/19 season, Liverpool – second-favourites for the title behind Manchester City – go top of the table with a 4-1 win over relegation candidates Norwich City.

It’s no surprise that fans, pundits, journalists and bookmakers alike would have had Liverpool as big favourites to beat Norwich. Liverpool accumulated 97 points in the Premier League the previous season and unforgettably finished 2^nd – a single point behind Manchester City. Norwich on the other hand finished one place higher than Liverpool – albeit in the Championship – with 94 points in the Championship’s 46 game season. 

97 points in the Premier League vs 94 points in the Championship, easy to pick which side you fancy in a once off game. Prior to this game, if you were to predict what the Premier League table would look like at the end of the season, you’d likely have Liverpool somewhere among the top and Norwich somewhere among the bottom.

If we look at the current League Table, Liverpool are at the very top and Norwich are at the very bottom. But of all the Premier League teams, what team is most satisfied with their performance so far?

First, one must ask how do you measure a given club’s “pre-season expectations”. Once we know that, we can compare their season to date with these expectations.

To measure the pre-season expectations of a team, we will be using a handicap style method. This means that each club is given X number of points to add to their end of season total.

1)      Let’s say you expected Liverpool to finish 1^st and obtain 85 points in the 19/20 League Season and you expected Norwich to obtain 35 points. We would assign Norwich 50 Handicap Points (H Pts).

2)     Let’s say we expected Liverpool to finish 2^nd with 85 points, Man City to finish 1^st with 89 points and Norwich to finish with 35 points. We would now assign Norwich 54 Handicap Points and Liverpool 4 Handicap Points (and Man City 0 Handicap Points).

The idea is that at the end of the season, every team has a pretty reasonable chance of finishing in 1^st place in this league table. Norwich may never see themselves finishing above Liverpool in the actual league table, but they can target finishing above them in Handicap League Table.

So now that we have a reasonable model for judging pre-season expectations, how do we fill in the data? How do you get a reliable and independent forecast for a Premier League club’s pre-season expectations?  Ask any club official what their expectations for the coming season are, and you’re bound to get an ambitious answer. Fans too are often optimistic and can’t help but feel “this is our year”. Journalists, pundits and media are often incentivised to say whatever will sell the most papers or get the most clicks.

For independent forecasts of this nature, I always tend to follow the money and look at what the bookies are predicting. The only allegiance the bookies have is to their own bottom line so they are incentivised to get their predictions as accurate as possible.

At the beginning of the season, Bet365 offered the following market. Each team has the same odds* of winning, so in Bet365’s eyes, this is a fair handicap system and every team should be competing for first place.

*each team is given odds of 16.00 (if you bet €10, you receive €160 if your bet wins). If this was a ‘fair’ bet, the odds would be 20.00 - this margin is how bookmakers make money. The important point is that the odds are 16.00 for ALL teams, so Bet365 are incentivised to be as accurate as possible with their assigning of Handicap Points to teams. Bookmakers may change lines/odds/handicaps based on how the market reacts, so this is not a perfect model, but it is still the best option for our intents and purposes. These remarks aren't necessarily pertinent to the topic but I felt it should be mentioned. 
Also note, the only reason Leicester are highlighted in the image below is because they were the team I fancied most in this market!

The race for the “Over-Expectation Trophy”

This is the League Table following the Leicster City vs Aston Villa game (the last game played since the suspension of the league due to Covid-19). This will be referred to as the Current League Table (CT) and these points totals will be referred to as Current Points (C Pts). This will appear in gold colour in all tables.

Next, we’ll look at the Starting Handicap League Table (SHT). This table was created before the start of the 19/20 season and assigns each team X handicap points (H Pts) to add to their end of season tally. This will appear in light blue colour in all tables.

As we can see, Man City are assigned 0 H Pts (meaning they are expected to finish 1^st in the actual league table), Liverpool are assigned 4 H Pts (meaning  they are expected to finish 2^nd in the actual league table (and about 4 points behind City)), and Sheffield United are assigned 52 H Pts (meaning they are expected to finish 20^th in the actual league table (and about 52 points behind City and about 48 points behind Liverpool). 

This table gives us a good idea of what teams’ pre-season expectations looked like. Man City and Liverpool are the clear challengers for the title with City having a slight edge. Spurs, Chelsea, Man Utd and Arsenal know that they’ll really have to over-perform to make a challenge for the league title, but top 4 is a very reasonable expectation.

From this table, we can get a pretty good idea of how teams are faring so far this season. Most teams CT positioning is pretty near their SHT positioning. Where these numbers differ greatly is where we find the teams most over-achieving or under-achieving. Sheffield United were expecting to finish 20^th but are currently in 7^th – over-performing by 13 places. Spurs were expecting to finish in 3^rd place but are currently in 8^th – under-performing by 5 places.

It’s important to note that when drawing inferences using this method, it only accounts for league position and doesn’t factor in points. Let’s take three teams who are all under-performing by 5 places – Spurs, Watford and Bournemouth. Spurs expected to finish 3^rd but are 12 points off 3^rd place in the Current Table. Watford expected to finish 12^th but are 10 points off 12^th place in the Current Table. Bournemouth expected to finish 13^th but are 8 points off 13^th in the Current Table. So, although all three teams are 5 places below expectation, Spurs are actually the greatest number of points below expectation. What about Manchester City in 2^nd place? They are only under-performing by 1 place but are a whopping 25 points off their expected finish of 1^st place. 

It is intuitive that places in the top half of the table have bigger points differences than places in the bottom half of the table. The bottom half is more often bunched together and a team could jump 5 places picking up two wins in a row. However, it’s rare that two wins in a row would raise you 5 places in the top half of the table. 

Another small point to note is that in three cases (Chelsea & Man United; Southampton & Watford; Newcastle & Aston Villa) teams are assigned the same number of H Pts. In these cases, the team with the higher CT positioning was placed ahead in SHT positioning. 

The below graph gives a rough visual representation of how teams are doing versus their pre-season expectations. Teams above the blue line are outperforming their pre-season expected finishing position, and teams below the blue line are under-performing. The greater the gap between the blue and the orange line, the greater the gap in expectation.

Table positions can tell us a lot, but at this point of the season the points totals may be more telling. For this, we look at the Current Handicap League Table (CHT) aka “The Over-Expectation League Table”.

This table factors in Current Points and Handicap Points. This table tries to indicate what team is doing best versus their pre-season expectations. Remember, this is a table that every team would have had aspirations of topping at the beginning of the season.

Unsurprisingly, Sheffield United top the table. Tipped for 20^th place before the season started and expected to finish 52 points behind Man City, Sheffield United find themselves in 7^th place in the Current Table and just 14 points behind Man City. It’s very tough to argue that any other team is outperforming their pre-season expectations to such a degree.

Burnley, Leicester and Liverpool round out the top 4 in the Current Handicap League Table. Leicester and Liverpool have been talked about plenty this season but some may not have noticed Burnley’s ascent up the table. Sitting in 10^th place and guaranteed safety with ten games to spare is quite a lofty spot for a team that were expected to be relegated.

The other end of the table also makes for interesting reading. Interestingly, Spurs, City, Arsenal, Man Utd and Chelsea (supposed members of the “big six”) all find themselves rounding out the bottom five. When it comes to league position, Chelsea and Man Utd don’t seem to be under-performing at all. However, both teams are attaining far fewer points than they would’ve expected despite their acceptable league positions. This indicates that this season’s Premier League has been an extremely competitive league with the mid- and lower-table teams causing more upsets than would’ve been expected. Aside from Liverpool, it’s hard to argue that any of the big six are having ‘good seasons’.

Points vs Position

Chelsea and Man United are under-performing versus their points expectation but are doing fine versus their league expectation. Which is more important? At the end of the season, teams are awarded “merit payments” based on league finishing position. Titles, qualification for European competitions, and relegation are also decided by league positions – points totals are a mere statistic for the record books. So, the first factor people consider when judging how a team has performed in a season is their finishing position. This is fair and makes sense.

Scenarios can arise where a team is less incentivised to pick up points when it will have little effect on their overall place in the table. Man City are a reasonable example of this right now. With the league title out of reach, City likely won’t be too concerned about their overall points tally so long as they finish 2^nd place in the league.

Mid-table teams with Cup runs will also often under-perform versus points expectations. Once a team is safe from relegation, they may take their foot off the gas in league games and shift their focus to Cup games towards the end of the season. This could result in dropped points but have no significant impact on their league position.

Some alternative models

I firmly believe that the financially incentivised bookmakers prove to be the most reliable forecasters when it comes to these things but let’s take a look elsewhere and at some other data that could be used.

Final League table predictions serve as a reasonable look at a team’s pre-season expectations. For an aggregate view of fan predictions, BBC’s fan-voted 2019/20 predictions can be found here. Six NBC pundits made their predictions of a final league table here, and FourFourTwo did something similar here.

The trouble with league position predictions is that it doesn’t factor in points expectations. The quantitative and data centric site FiveThirtyEightproduced a simulation for the 2019/20 season which attempts to forecast goal difference and points total. This model is much more similar to the handicap method we used above.

Using these figures with the same method as above (so City are awarded 0 H Pts, Liverpool 5 H Pts, Chelsea 20 H Pts etc), it gives us a similar looking CHT:

Comparing both the 538 table and the Bet365 table, we see that both forecasts are quite similar. Sheffield United top both tables by a reasonable margin. The top 4 remains the same in both tables although the order is slightly different. According to the 538 model, Man City are actually performing worse than Spurs when it comes to pre-season points expectations. Man Utd jump six places from 17^th in the Bet365 table to 11^th in the 538 table. Clearly 538 had lower pre-season expectations for United than Bet365 had.

The future…

It’s important to note that these league tables are simply a snapshot in time. All teams have several remaining fixtures and this will impact their finishing positions in both the actual league table and the Handicap League Table.

Each team will have their own unique ambitions and expectations when they enter into a new season. Most teams’ aims will be clearly defined (“win the league”, “make top 4”, “avoid relegation”) but it can be difficult to compare teams with different aims. If Chelsea achieve their aim of finishing 4^th and Burnley achieve their aim of avoiding relegation, which team had a ‘better’ season? If Man Utd miss out on top 4 and Aston Villa get relegated, which team had a ‘worse’ season? It’s impossible to answer these questions with a high degree of certainty, but using these tables can give us a decent yardstick to compare every team in the league to each other.