Actually, assuming that bets are respected up through one atomic blockchain block before pool closing, the operator can't ensure they're last-to-move, because o

Actually, assuming that bets are respected up through one atomic blockchain block before pool closing, the operator can't ensure they're last-to-move, because of the mechanics of Bitcoin-transactions-into-blocks. All bets are public, and all bets into the 'last block' before the pool closes would be equally valid. (Bitbook.ie could use some other 'we saw your transaction in time, even though it didn't make a block' approach... but that would be so open to abuse and customer complaint I doubt it'd fly.) So this again presents more problems for potential arbitrageurs as soon as there's more than one last-minute "sniper": not only are they unsure if their bet will be minted into a block in time, they can't be sure N others following the same strategy don't try the same thing, making the final odds awful. To the extent the operator has any discretion, it's just to make a final ruling on whether a block arrived "soon enough", or is the first block for which bets are a "no-go". There's a very, very slight abuse potential in such discretion: their decisions are publicly observable, and biased/arbitrary decisions would result in a lot of griping and lost credibility. On the other hand, large mining pools would have significant power to control the 'last money in', in the blocks just before an event closes. (If they see imbalanced odds they like, they could blockade other bets, and only allow their own confederates' late money in, and only to the extent that the odds remain favorable.) Maybe that's the real business model: kickbacks from side-betting mining pools!