With only six Eredivisie matches played, it is still very early days to be attempting to draw conclusions but Dutch TV described FC Utrecht’s start to the season as surprising and I thought it would be interesting to use my simple expected points model to see just how surprising Utrecht’s points total at this stage actually is.
After six matches, FC Utrecht were expected to have approximately eight points according to our simple expected points model. The club actually has 11 which is a 38% improvement on the bookmakers’ assessment of the club’s six matches. However, there are three other clubs who are doing even better.
Leaders FC Twente were expected to have amassed 13 points by this stage anyway as their opening six matches were relatively easy. The 2009/2010 champions have only played one of last season’s top seven clubs so far and the next few weeks will provide more evidence on their championship potential than the first six matches have. Twente have, however, managed to take six more points in these fixtures than the equivalent encounters with the same opposition last season.
Vitesse are another team who have started well but, like FC Twente, their schedule so far has been relatively light as only one of last season’s top seven, Feyenoord, has had to play the team from Arnhem. However, the 14 points Vitesse have accrued is around 40% more than than they were predicted to have at this stage. Perhaps more interesting is the fact that Vitesse took 15 points from these equivalent fixtures last season and are therefore performing no better than last term.
Finally ADO Den Haag whose total of nine points might not appear spectacular but it represents the biggest percentage improvement (over 42%) on the model’s prediction for points at this stage. This is because ADO have already played four of last season’s top half of the Eredivisie in their first six fixtures. Looking at ADO’s equivalent fixtures last season, they only managed to take three points from them as opposed to the nine this term. It will be interesting to keep an eye on them to see if they can challenge for European football again.
All of the above is based on far too small a sample to be concluding too much and I will return to this as the season wears on to discover which clubs really are over- and underperforming in comparison to expectations.
Scoreboard Journalism 
I understand what you are trying to do … the table is interesting, in a novelty kind of way, but there are a couple of huge flaws.
1. You have stated that you do not want to participate in rank/scoreboard journalism, but in fact that is exactly what you do, but with a slight twist. All your conclusions and statements are actually based on the rank a team has and has obtained over the past few years.
I’m not sure how the bookie numbers are exactly calculated, but we can assume that they take into account position/records over the previous years. I know you want to ban out “freak” results and by taking a ton of years into account the bookie odds probably aim to do that, but it is still what they are based on (I assume … I don’t think they incorporate intangibles like the perceived gain/loss of player quality, or change in coach, or change in tactics, type of referee assigned to the game, “morbo”, history of red/yellow cards in a specific game etc.).
2. Another issue is that basically this kind of analysis always looks “back” and not to the present or forward … this is difficult for me to explain and I really need to chat with a face-to-face with a friend or two about this to get my ideas clear and straight … an attempt … what I’m trying to get at is that … when analyzing a match, as it is being played, you can take into account the expected result to see if a team is over/underachieving, but you should also look at how the match evolves … how is the relationship of shots, possession, passing, opportunities etc. that are witnessed as the match is played, to what could be expected …
Example: I expect Vitesse to dominate 75% of their matches this season, as they continue to have an influx of quality … but in the first 5-9 games of the season I don’t expect them to amount to all that much. Why? Well, players have to adjust to a new trainer, adjust to each other etc. … but I can tell how well they are doing this by looking at some of those things I mentioned previously … maybe they are clearly out-shooting all their opponents, then as an individual game evolves I have higher expectation of the number of points they will take away. If I then want to make an analysis on how Vitesse are slowly becoming better, I take the same 5-9 games from the previous season and compare shooting, passing etc. Can I see more dominance over opponents? Is there a notable difference in balls won/tackling in certain areas of the pitch? Are there changes in fouling/cards received etc.
I am now taking the game out of the scoreboard environment and into the individual environment, taking the game on the merits of itself and not on what past scoreboards have told me what I should expect.
If I want to go even further I can see how Vitesse are improving as compared to the Dutch league in general … historically what is the expectation when dominating possession/shots/passing/whatever may have a statistical relevant connection to the possible outcome to a game (I believe shots are the best indicator of match outcomes)? Do Vitesse follow this trend or do they still deviate significantly etc. (thus allowing for more improvement or not)?
Scoreboard journalism, as Adriaanse mentioned, means looking at the matches individually and then tabulating an expectation … he means, don’t only look at what history has told you will be the outcome of a match (bookie odds), but also look at how the match evolved to see if the outcome was justified (lucky or not, etc.)
Yes, one can make a table of expected position related to actual position … but only AFTER the season is over … why? well, you want to eliminate luck from the table and can only do so by taking a significant number of games into account, this will only happen very late in the season, even now a lucky bounce and a win (3 points) will possibly rocket a team up the standings. Those 2 PSV “shock” results still have a huge effect on your predictions and the teams they played are still obviously high in the expected table as 3 points is still 16.7% of a possible total number after 6 games, but after 34 that’s down to 2.9%. RKC are high in your table due to exactly that, without that one game they are at expected level (which looking at their other results seems exactly correct) thus are they really overachieving, as your table seems to say? Or is it down to them being fired up for the first game of the season at home against a local rival/the big boy from Brabant and against the others playing to ability (funny enough, none of the other games are against local teams)? But here I am basing quality of a team on results without looking at the details … (see what I mean).
I hope you can expand your site and thoughts a bit more, that would be cool. I see you’re just starting out so plenty of time to improve etc. At least you don’t blab away like 95% of the fans/analysts etc. and are starting to look deeper behind results, keep up the good work …
Cheers.
I just thought of an interesting experiment …
I’m not the biggest believer in the stats revolution flying through football at the moment, as I still believe that such things as intangibles have an influence on the game and not everything can be boiled down to stats … so … how to incorporate intangibles into the team expectancy story …
I believe VI publishes “end of year” tables based on the opinions of a whole host of “experts” (I refuse to label W. Genee (sp?) anything close to an expert) … anyway, I’m almost sure none of them did some sort of complicated probability analysis on what the final table of this season is going to look like. They probably looked at trainers, players, financial situation, tactics, may even have qualified a team to a certain position because they support that team and many more intangible aspects.
Get the VI and calculate the net result of all their opinions and see where each team would be at the end of the season … then compare that with what the bookie odds have for end-of-year positioning.
How do the two tables compare? Do the experts value Vitesse higher than the bookies? Who do they see as going down? etc.
You would assume that the experts’ table is inclined to be based on rank journalism (intangibles) more than what the bookies say.
(My guess would be that at the end of the season the real table will be the average of the two. I don’t read or have VI so I’m not sure how many experts were consulted, hopefully it’ll be around 30-50, if only 10 or so then the stuff above doesn’t really work … we need as many opinions as possible, as much intangibility as possible.)
Also …
With the expected results in hand … two interesting games, seen this weekend (so far), to analyze would be the 3-3 (Heracles-ADO) and the 1-0 (Ajax-Twente) taking into account the game statistics like shooting, possession, corners, passing, aggressiveness, line-up/tactics etc.
Does what was expected corroborate with what was seen on the field? Or was there luck/lack of luck involved (as Adriaanse once stated)?
Cheers.