I'd suspect it was less to do with the edge, which isn't especially huge especially over the short run, and more to do with the authors' unusual selection of bets entirely overlapping the bookies' ad hoc analysis of bets they've mispriced. Particularly since the bookies watch each others' odds (as do scripts sold to wannabe arbitrageurs on the internet). Having a subset of winning betters with an information advantage may even help bookies overall if they know who those people are, but they'll want to limit how much they pay out for that information. Especially if the limits are market specific, and the section of the paper highlighting betting limits shows a screenshot of an attempt to place bets on Australian youth association football...
But $900 is pocket money bookmakers are willing to give away: around the time this study collected it's first data points I made more from fewer bookmakers just from intentionally +ve expectation welcome bonuses (and that was after they'd responded to the first wave of people pocketing welcome bonuses by eliminating them... for people from Denmark)
But $900 is pocket money bookmakers are willing to give away: around the time this study collected it's first data points I made more from fewer bookmakers just from intentionally +ve expectation welcome bonuses (and that was after they'd responded to the first wave of people pocketing welcome bonuses by eliminating them... for people from Denmark)