This time last year, Scoreboard Journalism embarked on an exercise to collect Eredivisie predictions from various sources to be assessed to discover how these predictions actually fared. A similar exercise has already begun in time for Friday’s big kick-off for 2014/2015 which you can take part in by contacting me via Twitter or here but what about last season?
As a reminder, the 13 participants last season were:
1: Euro Club Index (@euroclubindex). Based on results of last four years, weighted for recency.
2: Steve Lawrence (@SteveLawrence_): Based on transfer values of current squad members.
3: Terry Tipster (@terrytipster): Based on team/player strengths.
4: Soccermetrics (@soccermetrics): Based on projected statistical outcomes of current squad combined with Pythagorean.
5: Niels van de Ven (@nielsvandeven): Based on goal difference from last 3 years, weighted for recency and transfers (subjective component).
6: A mystery guest from the Dutch sports media who preferred to remain anonymous.
7: Michiel Jongsma (@jongsmajongsma) of BeNeFoot.net
8: Nikos Overheul (@noverheul) of catenaccio.nl
9: Arjan Blom (@arjan_blom)
10: DJ Van Zoen (@fcgodenzoon, twitter handle now defunct)
11: Will Burns (@willburns6) of TotalDutchFootball.com
12: Babette van Haaren (@babettevhaaren)
13: Martijn Staal (@m_staal)
An overview of the original forecasts from a year ago can be found at https://scoreboardjournalism.wordpress.com/2013/08/01/what-will-the-eredivisie-final-standings-look-like/. To assess their accuracy the standard deviation of the differences in the predicted points per team and actual points per team have been plotted by @jameswgrayson:
- Soccermetrics is the winner on this measure followed by Steve Lawrence and the Euro Club Index.
- All of the top three were based on statistical models and the median of the five models outperformed the median of the subjective picks.
- Nikos Overheul of Catenaccio.nl produced the best of the subjective forecasts and was also better than two of the statistical models.
- Babette van Haaren was the only other subjective forecaster to perform better than a simple baseline of just picking the 18 clubs to get the same points as in 2012/2013.
- All five of the statistical models were better than picking the clubs to get the same points as a season earlier.
- The median of the subjective forecasters outperformed them all but was not as good as three of the models. This was perhaps an illustration of “the wisdom of the crowd”.
- Two of the statistical models outperformed the median of the objective forecasters.
- Everyone thankfully did better than the dumbest baseline of all which was that all 18 clubs would finish on 47 points.
However, as I suggested last season, one season is not enough to be certain that forecasters are good or bad which is why I am hoping to have as many of last year’s participants enter again this time around. One of the 13 (@fcgodenzoon) has disappeared but I hope to have the rest back again in an expanded group predicting the final table for the 2014/2015 season. If you’d like to join them, please get in touch before Friday’s big kick-off. In the meantime, congratulations to Soccermetrics, last season’s best predictors of the Eredivisie.
I’d also like to thank Steve Lawrence (@stevelawrence_) for his sterling work throughout the season, tweeting overviews to show how the competition was progressing. Updates will be much more frequent this season I hope and I plan to integrate Steve’s work in this in addition to James’ plot.