"The best way to predict the future is to create it." I always thought the the problem with prediction markets is you have the incentive to make the prediction come true. And that gets ugly fast. But hell, I guess that's what all this is at the end of the day.
If players in the market have ways outside the market to significantly affect the outcome that the market is betting on, then it isn't a prediction market. The whole point of a prediction market is that there is no such feedback mechanism.
For example, the article discusses prediction market bets on Trump winning the election. The underlying assumption of such markets is that no player has more than one vote in the actual election, out of some 150 million or so votes, so each individual player's ability to affect the outcome is negligible.