Nate Cohn provides one:
As a general rule, the incumbent’s share of the vote is more predictive than the challenger’s peak, so if Romney doesn’t near 49 percent after his convention, that wouldn’t preclude him from such a showing in November.
Many voters with deep reservations about the president’s performance could justifiably hold out for more information about Romney before choosing to support him. On the other hand, if Obama can’t inch close to 49 percent, that could be a sign that a majority of voters have very deep reservations about giving him a second term. Again, it wouldn’t outright prevent him from winning, but it would contradict the view that the voters who like him but are disappointed with Obama’s performance are itching to join his cause in November. However, if one candidate gains a decidedly higher share of the vote at the peak of their bounce, that would be a very good sign for that candidate, even if they fall short of 49 percent. The candidate with the highest peak has won 10 of 12 presidential elections since 1964, including every election involving an incumbent president.
(Chart from Seth Masket)