Interestingly, I was chatting to an acquaintance on the weekend who develops software for a few big rolling professional gamblers. AFL is one of the areas he has done this for. As you can imagine, it's all very statistically based, not emotionally. It's certainly not full proof. Like for example, he said last year the software was always about 2-3 weeks statistically behind the Doggies' real world progress. So didn't tip them as premiers at any point. But seems to be fairly close to form more often than it's not.
He said, for most of the 2017 season, the software was tipping Adelaide as most likely premier, but not as clearly as many in the media would have thought. It had Sydney as marginal second favourite for much of the second half of the season, then us, GWS and Geelong pretty much as equal third favourites.
However he said someting interesting happened in the late part of the season. The Geelong loss at Kardinia actually caused the programme to make us pull marginally clear of Geelong (and GWS). The reason being that the computer rates Geelong a 3-4 goal better side in Geelong, so us losing tightly was statistically a win (if that makes sense). It also factored in that we played a man short for most of the game and that some of our best players were down on their statistical averages and rated it an anomaly (it looks for longer term trends on individual player performance).
Then futhermore, the flogging of Fremantle was the first week that his programme had us as leapfrog Sydney and Adelaide into marginal premiership favouritism. Then pulling slightly further ahead with the big St.Kilda win. From that point onwards his software had us clearly as premiership favourites and Adelaide as a clear second. Think one of his clients cleaned up, putting a pretty big outright bet on us after the Freo win, others making marginal profits, having more of a hedge strategy.