Posted the final CT prediction correlations a while ago now the PT ones:
My PT prediction turned out to be the most accurate PT prediction of the season with a narrow margin ahead of matt.
Just as in the CT update most subjective predictions turned out better than formula based one so perhaps we all should be more brave and post our subjective thoughts over trying to create perfect formulas
Team by team prediction overview:
Doing well in correlations has a lot to do with avoiding big misses and both me and matt did that well. My biggest miss was to predict aker 7 spots above where they ended but that didnt hurt as much as other misses because of how close the upper table actually was. A notable prediction that both me and matt did was to predict Wiesenhof only just above relegation spots and without Taaramaes first week TDF struggle that prediction would have looked worse than it did.
Sorry abhishek, putting a promotion finisher as bottom of the table is not helping your algorithm to a high score.
Spoiler
The correlation of the different ranking updates:
0,493
#1
0,496
#2
0,560
#3
0,673
#4
0,774
#5
0,812
#6
0,934
#7
0,968
#8
0,983
#9
0,997
#10
Probably not too interesting for most of you but it serves as a reminder that early season rankings don't mean much. Good pre-season prediction correlate more with the final ranking than pre-TDF rankings do. Of course context about performances and season planning help estimating mid-season but too many conclusions are drawn mid-season.
