Hi, Chinbok.

I assume your stats are using the default neural settings? If so, they are what can be expected for accurate ratings because the strike rate is what faves win, and the 14.2% LOT would more-or-less account for the TAB takeout. Therefore, if the neurals are accurate in selecting the winner would it not be reasonable to expect they are also accurate in selecting the false-fave? Or maybe this is only wishful thinking on my part?

Applying the selections to Betfair, and assuming you would have to lay them at a 33% premium to their TAB starting price, the payout on Betfair would have been $22.22 ($16.70 X 1.33). As there have been 38 selections, by my reckoning the system would be well in front?

So far, two of the three selections today have been beaten. Hopfully the third will also be beaten.