Have we? Or was the rule book, call it conventional wisdom been maintained. Sure, you have all the Trumpers going around saying all the polls were wrong. Have they ever stopped to look at what was being polled. The popular vote. The polls said Clinton by 3 points on average, she won by two. The polls weren't wrong. Trump won because he squeaked wins in Pennsylvania, Michigan and Wisconsin.
Conventional wisdom, history if you will says a president will lose seats in his first midterm election. Trump has lost 39 with four still pending, not decided. That's an average of 23 seats since 1934, FDR through and including Obama. Trump lost 16 above average. Only three presidents, Obama, Bill Clinton and LBJ lost more than 39. Obama, Bill Clinton lost control of the House, LBJ didn't. LBJ had 295 to start off with, fell to 248 which still gave him a 61 seat majority. LBJ could easily afford to lose those 47 seats. In 1954, the only other president to lose control of the house in his first term, Eisenhower lost 18. IKE loss of 18 hurt him and the GOP much more than LBJ's 47.
It's true Trump's win was unique in that he became first candidate ever, or since Gallup and Pew Research began keeping track of favorable ratings with a favorable rating of below 51%. Trump at 36% set the record for the lowest favorable rating ever at 36%. Only three other presidential candidates since FDR had favorable ratings of below 50%, Hillary Clinton 38%, Barry Goldwater 43%, G.H.W. Bush 46% in 1992. Those three lost. But Trump won by having Hillary Clinton as his opponent, the second lowest ever candidate in favorable ratings. So yes, that is a unique election. 2016 just happened to pit the two candidate with the lowest favorable ratings ever against each other.
Other than that, conventional wisdom, history, the rule still applies. Of the two president who withdrew after being challenged, Trump had a 22% approval rating, LBJ 35% approval. The other who lost the general election, Ford had a 45% approval, Carter 37% and G.W.H. Bush at 31% on election day. Approval rating being different than how one views the person, favorable rating.