Yes yes, it's zero sum less the house rake. I thought that was obvious enough to go without saying; apparently I was mistaken.
You are treating the house take as if it had a NET effect of zero; the only loss being a predictable constant. Suppose that the election created a roller-coaster in popularity for each candidate. That would generate a lot of buying and selling, no? Then the house's share of the total value would be large.
Or suppose everyone became convinced that a given candidate was 96% likely to win. According to a zero-sum game, the value of that candidate's stock would be 96. But suppose the house takes 10% of each transaction. Who would buy at 91, costing 9, when the most they could get to sell was 100? Even though the seller suspects that buying at 91 would probably result in a profit aside from the fees, he knows for a certainty that the profit could never surmount the cost of his trading fee.