It should definitely be a credibility weighted prediction

Based on actual past data (eg league record to date) and a subjective view of the future remaining games (independent of league record) taking into account various factors like we've established are missing from the very simplistic model design, and then blend them to give a Bayesian posterior distribution.

The only attempt to take into account any form of subjectivity (independent of the games played so far this season, which is entirely captured in the current league table anyway) seems to be weighting home wins more than away wins.

I expected/had hoped for more from the site owner. They just seem to have build a Monte Carlo simulator which isn't exactly difficult to do. It's the model assumptions that really provide meaning to the predictions.

As for the "predicted" % chance of us getting relegated is concerned. The fact (coincidence) that it's not too dissimilar to the betting markets, could just be that, a coincidence. Two models could both be wrong but similar.....

Posted By: essexcanaryOTBC, Mar 23, 16:12:56