Forensic check on the 2012 season with the new streamlined Monte Carlo full season predictor: Managed to predict six of the top eight - Fremantle came from nowhere in my model (I predicted 14th and they managed 7th: well done purple haze) and North snuck into the eight where I had them at tenth with a 43% chance of making the eight. My big loosers/ poor performers of 2012 (apart from the Hawks loosing a seemingly straight forward statistical win in the last game of the season) were the Siants and Carlton who managed to end up 9th and 10th so I do not feel too bad about these.
The beginning of 2012 saw the Hawks facing a number of potentially tight games/ swing games that were flagged to define their season within the model, the Hawks still managed to be minor premiers despite the mid season Monte Carlo model predicting a third or fourth position - their very strong run of games in the second half of the season starting with the demolition of the Roos pushed them to the top (peaking too early?). Both Sydney and West Coast were flagged as strong contenders right from the start which i think was borne out - the Weagles coming very clsoe to a top two position - loosing the crucial swing games late in the season
Now having said how good the model predictions were for final ladder position it must be said that the order was all over the shop - only got one of the top four correct and as shown in the last post the top four is where the premier comes from 96.5% of the time in the model. As I have previously posted the model seems quite accurate for up to six rounds into the future. I am focusing on using a more robust (i.e. I need to improve the) model algorithm for the six rounds into the future then switch over to the simpler streamlined Monte Carlo version for the longer range predictions.
No comments:
Post a Comment