Midseason Statistical Over Analysis (Part One)

As regular readers of my blog know, back in November I was doing reports on how Premier League teams historically finished their season depending on how many points they had earned after various numbers of games.  As I got busy I had less time to do such reports and, when we started losing, they became less fun.  Instead of doing a small report each week, I decided to try to finish a big report for the halfway point of the season.

Unfortunately, the project turned out to be far more time consuming than I expected, although not more time consuming than I should have expected.  As a result, I have decided to publish each part of the study as I finish it rather than waiting for the entire study to be complete.  It is possible that I will lose interest or run out of time before I can finish everything so it is better that something get published—if only to encourage me to keep going.

A few caveats:  my results are based on the table position and statistics for each club after it played exactly 19 games.  In some years not all clubs reached 19 games at the same time.  These totals apply to each club for their first 19 games of their season—not for one game against each opponent.  For example, in 2010-2011, the first five clubs hit 19 games on December 27, but the other teams had only played 17 or 18 games.  Manchester United played its 18^th game on December 29 and its 19^th game on January 1.  I counted both those games in my totals for Manchester United, but not for its opponents if they had already played 19 games.

This chart summarizes where clubs stood in the table after 19 games with the points and goal differential it took to get there as well as where the clubs ended the season and their points and goal differential at the end of the season.

AFTER 19 GAMES                                                         FINAL POSITION

Pos         AvgPts  PtsRng  GDAvg  GDRng                AvgPos PosRng AvgPts  PtsRng  GDAvg  GDRng

1              43.2        36-52     24.1        14-37              2.1          1-4         85.7     68-95    42.2     15-71

2              39.8        35-45     20.7        10-33              2.6          1-6         77.8     55-91    38.2       5-65

3              37.6        34-42     16.0        7-26                3.2          1-13       74.3     48-90    31.1       -12-44

4              34.3        30-37     12.1        1-21                4.7          2-8         66.4      53-77    20.1       -1-39

5              33.2        29-36     10.5        4-21                4.7          2-12      64.27    43-84    18.3       -15-51

6              31.6        28-34     9.7         -1-18                6.1         1-14       61.6      46-78    16.1       -6-37

7              29.7        26-34     3.3         -2-11                7.6         5-11       55.2      45-69     1.5       -11-26

8              27.8       24-33     0.7          -10-8              10.7        6-17       49.2       35-60     -5.0      -25-20

9              26.7       22-31     2.2          -5-10              10.2        6-16       49.8       39-61     -1.7      -23-23

10            25.2       24-28     -0.9        -6-2                10.9        6-16       49.1       37-63     -3.2      -20-28

11            24.1       21-27     -2.7        -11-3             10.5         6-19       49.7       39-61     -5.0    -20-26

12            22.9       21-25     -4.2        -11-1              11.3       4-18       47.9       34-68     -6.4       -25-21

13            22.3       20-25     -4.6        -15-4              11.8       6-20       45.9       33-58     -10.0    -34-5

14            21.3       18-25     -6.6        -16-1              14.4       7-19       41.8       30-54     -13.3     -40-6

15            19.9       17-22     -6.2        -16-8              13.4       9-17       44.3       36-52     -9.3       -24-9

16            18.6       17-21     -10.9      -23- -             15.5        8-20      40.2       25-52     -17.8      -42-2

17            17.7       15-20     -11.9      -20- -             15.7        8-19      39.6       32-50     -17.7      -35-6

18            16.5       14-20     -13.8      -22- -5           16.2       12-20     37.6       19-45     -21.9      -45- -1

19            15.0       10-18     -15.3      -26- -8           17.2       14-20     35.3       24-42     -23.7      -42-0

20            12.2       6-18       -20.6      -34-               19.2        14-20     28.4       11-42     -33.6      -69- -17

Pos:  The clubs’ position in the Table after 19 games

AvgPts:  Average number of points for clubs in this position after either 19 games or the full season, depending.

PtsRng:  The range of points for clubs in this position after either 19 games or the full season, depending.

GDAvg: The average goal differential for clubs in this position after either 19 games or the full season, depending.

GDRng:  The range of goal differentials for clubs in this position after either 19 games or the full season, depending.

AvgPos:  The average table position for clubs at the end of the season.

PosRng:  The range of table positions for clubs at the end of the season.

The most surprising (to me as a recent American fan of English football) thing I discovered in doing this project was the second half performance of the 2002-2003 Sunderland club.  I never would have guessed that a team that earned 18 points in their first 19 games would only earn 1 point is their second 19 games.  That must have been a miserable season.  I assume it was also the type of historic collapse that is still remembered and talked about.

At the halfway point of the season, the top of the table holds good teams playing well and average teams playing very well.  The bottom of the table contains bad teams playing poorly and average team playing very poorly.  The middle of the table contains average teams playing averagely, good teams playing poorly, and poor teams playing well.

Perhaps this seems obviously, but it explains why no team that finished the first nineteen games in 10^th place or better has ever been relegated and only one team has ever reached a Champions League spot (as currently defined) if they were not in the top six after nineteen games.  (Leeds in 2001-2002: Fourth place didn’t qualify them for a Champions League berth under the rules at that time.)  We will end the first half of the season between 3^rd and 6^th place with 29, 30, or 32 points. That puts us in great shape to avoid relegation and gives us a shot at the Champions League.  No one really could have hoped for more last summer.

Of course, we could let the “what ifs” run wild:  we should have beaten Liverpool, Aston Villa, Manchester United, West Brom, and Burnley.  We should have drawn with Arsenal and Tottenham.  That is 15 points dropped.  So we should be in second place with 44 points—already safe from relegation and virtually assured of a Champions League spot.  I suspect, however, that all our rivals have their own “what ifs.”

It is worth noting that in terms of points earned Chelsea is an above average 1^st place team, even if we beat them Sunday.  Manchester City is an above average 2^nd place team even if they lose Sunday. We, however, would be a below average 3^rd, 4^th, or 5^th place team and just an average 6^th place team.

Our goal differential paints a more optimistic picture, however.  Unless, we get clobbered by Chelsea, our goal differential will be well above average for whatever place we end the half season at.  The 19 game goal differential figure seems to be a pretty good indicator of team quality. If you look at the 8, 9, and 10 slots on the chart you can see that the otherwise inexplicable poor performance by teams that finish 19 games in 8^th place appears to be explained by their poor goal differential both at 19 games and over the whole season.  Of course, this is a fairly small sample size but I am convinced that goal differential is as good a predictor of future performance as most anything else.

On the other end of the table, the chart shows just how desperate Leicester is.  Unless, they beat Hull, they will be a below average 20^th place team which is further than average from safety.  Even if they beat Hull they are not in good shape.

That concludes the first part of my midseason statistical over analysis.  I hope you enjoyed it.

Edit:  While working on the next phase of the project, I noticed 45 minor errors.  I actually only made two minor errors but they infested about a quarter of the chart.  None of them were significant and I have fixed them.  Just to give one example, the goal difference average for 4th place after 19 games is 12.1 not 11.6 as originally posted.  Sorry.