It was introduced under the RAND name in 2012 which has been rebranded for LA Times, but the same team that developed it is still involved. Something like fourth most accurate for Romeny/Obama ??
https://alpdata.rand.org/?page=election2012
looks like their final numbers were: obama 49.5%, Romney 46.1%
The actual numbers were, of course: obama 51.1%, Romney 47.2%
So they were obviously right on the winner, their spread was 3.4% and the actual spread was 3.9%. That’s pretty close.
Of course they were off in that obama was polled under 50% in their final poll, but got 51. I guess their final poll still allowed more of an “undecided” than the real world election booth allowed. (I do think it is accepted that “undecideds” broke towards obama....unlike the conventional wisdom that “undecideds” break towards challengers...so maybe that accounts for obama’s lower numbers ...except Romney’s numbers are also lower...in any event, this is a minor point. The bottom line is that they were correct on winner, and pretty darn close on the spread.