I agree with everything but the bolded sentence. Bad methodology doesn't magically become good just because a better method is not currently practicable. His method is probably the best that you can do with this data without some hardcore math that I left behind in grad school
Are you saying you think there's a way to use both the Democrats' chances of winning the White House and Sanders's winning individual odds to assess his implied odds of winning the general? Like a weighted average or something?
The math is pretty straightforward if we assume that the probability of a generic D vs R is independent of the particular candidate. Certainly the candidate matters, but long term the mood of the electorate towards a particular party probably matters more. So, I'll make that assumption to simplify the math for the first part.
If the candidate selection in the primary is independent of the general election then the probability of an individual winning the general election is the product of their probability of winning the nomination times the probability that their party wins the general election. For example if Clinton has an 80% chance of winning the nomination and the Dems have a 60% chance of winning the general election, then Clinton would have a 48% chance of winning the general election as an individual (0.80*0.60=0.48).
Given both the probability of individuals winning the primary and general election I can flip the equation around and solve for the probability that their party wins the general election. So given Clinton at 87.7% to nominate and 54.1% to elect, that means the expectation is that the Dems win 61.7% in the general. For Sanders at 13.8% to nominate and 9.1% to elect the expectation is that the Dems win 65.9% in the general. Hence the claim that Sanders is placed to fare better than Clinton in the general election. Of course the probabilities are not independent so there is no expectation that the results would be identical.
More likely the effect is due to the traders themselves. As a contract gets lower there's more perceived upside relative to the risk. This could lead a trader to overvalue a Sanders contract to win the general election with respect to a Clinton contract. The general election contracts are lower than the primary contracts due to the co-mingling of both parties so the effect is compounded for Sanders compared to Clinton. The net effect is that Sanders appears to be more likely than Clinton to win the general in the minds of the Betfair traders.
