Conventional wisdom suggests the Tories could bleed Remain votes to the Lib Dems. Our detailed data analysis suggests this idea could be very wrong indeed
To find out, I teamed up with Martin Baxter, from Electoral Calculus. We wanted to take the best, most recent polling available on the voting intentions of remain and leave supporters, split by who they voted for in 2015, create a model of which voters are moving where, and apply that seat-by-seat to figure out what the major shifts are and what their impact could be in an election.
...
To find out, Other Martin used the transition matrix above to create a Markov chain style simulation. This approach takes a collection of possible states (i.e. being a Tory voter, being a Labour voter, and so on), and uses the probability that voters will hop between any two given states to simulate outcomes for a large population of voters. (For a superb visual explanation, check out this link.)
To make the model more realistic, we accounted for stickiness of each parties core vote, dividing voters into the stronger red til I die supporters and the weaker types more likely to swing. Put together, this provides a pretty good approximation of what would happen in an election if the polls were reasonably accurate.
...
Our model sees the Tories on 422 seats, with Labour reduced to just 150, and the Lib Dems declining from 9 to 6. The Conservative majority would be north of 190. Labour would be wiped out beyond what most people are currently predicting. Leadership candidates like Clive Lewis would no longer be leadership candidates, because they would no longer be MPs.
https://www.theguardian.com/science/2017/apr/27/lib-dems-shouldnt-count-on-remain-votes-the-data-looks-bleak
The problem being that the changed votes so often seem to happen in places where it benefits Tories - from Labour in places where they're in direct contention, or from UKIP where that combined the RW vote, while Labour transfers to LibDems aren't in the right places to elect a LibDem instead, but might enough to weaken the Labour candidate enough to let in a Tory.