With Premier League fixtures arriving thick and fast at the moment, I thought now would be a good time to revisit the Premier League Top 4 Predictions series to see if there are any changes to discuss. This predictive model isn't for betting purposes but simply looks for positive or negative result trends in comparison to the previous season. The model takes a basic assumption that teams will end up with the same results as those from the corresponding fixtures of the previous year and adjusts its final predictions as and when differences in results occur.

As noted in the first instalment of the series, relegated teams from the previous season are placed in order of highest to lowest points and replaced with promoted teams in the same order. In the case of this season:

Relegated/Promoted

  • Swansea/Wolves
  • Stoke/Cardiff
  • West Brom/Fulham

Therefore, when a team plays Wolves this season, the result will be compared with the home and away fixtures against Swansea last season.

Previous Top 4 Prediction

Note: 5th and 6th positions are included for reference.

  1. Manchester City: 101 
  2. Tottenham: 80
  3. Liverpool: 79
  4. Manchester United: 75
  5. Chelsea: 72
  6. Arsenal: 68

Previous Point Differences vs 2017/18 Season

  • Manchester City: +1
  • Liverpool: +4
  • Chelsea: +2
  • Arsenal: +5
  • Tottenham: +3
  • Manchester United: -6
Source

Updated Results

Manchester City - 2017/18 vs 2018/19

  • Southampton (h) - win/win: 0
  • Manchester United (h) - lose/win: +3
  • West Ham (a) - win/win: 0
  • Bournemouth (h) - win/win: 0
  • Watford (a) - win/win: 0

Points Difference: +3

Expected Points: 104

Liverpool - 2017/18 vs 2018/19

  • Arsenal (a) - draw/draw: 0
  • Fulham [West Brom] (h) - draw/win: +2
  • Watford (a) - draw/win: +2
  • Everton (h) - draw/win: +2
  • Burnley (a) - win/win: 0

Points Difference: +6

Expected Points: 85

Chelsea - 2017/18 vs 2018/19

  • Crystal Palace (h) - win/win: 0
  • Everton (h) - win/draw: -2
  • Tottenham (a) - win/loss: -3
  • Fulham [West Brom] (h) - win/win: 0
  • Wolves [Swansea] (a) - win/loss: -3

Points Difference: -8

Expected Points: 64

Arsenal - 2017/18 vs 2018/19

  • Liverpool (h) - draw/draw: 0
  • Wolves [Swansea] (h) - win/draw: -2
  • Bournemouth (a) - loss/win: +2
  • Tottenham (h) - win/win: 0
  • Manchester United (a) - loss/draw: +1

Points Difference: +1

Expected Points: 69

Tottenham - 2017/18 vs 2018/19

  • Wolves [Swansea] (a) - win/win: 0
  • Crystal Palace (a) - win/win: 0
  • Chelsea (h) - loss/win: +3
  • Arsenal (a) - win/loss: -3
  • Southampton (h) - win/win: 0

Points Difference: 0

Expected Points: 80

Manchester United - 2017/18 vs 2018/19

  • Bournemouth (a) - win/win: 0
  • Manchester City (a) - win/loss: -3
  • Crystal Palace (h) - win/draw: -2
  • Southampton (a) - win/draw: -2
  • Arsenal (h) - win/draw: -2

Points Difference: -9

Expected Points: 66

Source

Current Points Difference

  • Manchester City: +4
  • Liverpool: +10
  • Chelsea: -6
  • Arsenal: +6
  • Tottenham: +3
  • Manchester United: -15

Current Predicted Top 4

Note: 5th and 6th positions are included for reference.

  1. Manchester City: 104
  2. Liverpool: 85
  3. Tottenham: 80
  4. Arsenal: 69
  5. Manchester United: 66
  6. Chelsea: 64

Conclusions

The biggest issue with this kind of prediction is that it assumes the previous season's results should be considered standard for each team. This may be true of teams like Manchester City and Tottenham, with both clubs finding themselves with similar results so far in comparison to last season. However, Liverpool and Manchester United are outliers in this model for contrasting reasons.

Liverpool are managing to get results that they simply couldn't last season. This shows them as having 10 more points this time around compared to the same fixtures last season. The model doesn't account for this going forward, however, and therefore still has them predicted to finish 19 points behind Manchester City. On this season's form so far, that seems a bit farfetched and would take a monumental drop in form.

Likewise, Manchester United are no longer able to grind out results like they did last season. After just 15 games, they find themselves with 15 less points than they gained from the corresponding fixtures last season. The model still has them finishing in 5th place ahead of Chelsea simply because they managed to earn a lot of points last season, but it's difficult to see them managing to get anywhere near the top 5 this time around.

Ultimately, this model should be taken as a bit of fun. All it looks to do is make comparisons from season to season. What it does show us, however, is how teams are performing overall compared to last year, and I doubt anyone will be surprised to see just how far Manchester United have fallen at this stage.