The final nail in the coffin for Sanders was the Latino vote in TX. Obviously Hillary has the black vote, and Bernie and her seem to trade off with the white vote depending on the state. So if it were true (as the exit poll from Nevada indicated) Sanders was winning with Latinos, then he had a shot. But actual precinct data put into doubt that NV exit poll and now with TX, it was clear that Sanders isn't winning with Latinos, which puts NM, AZ, FL and CA out of reach too and that is pretty much the ballgame, even if he could somehow pull off some wins in the rust belt.
He won Nevada Hispanics. People have said he's won Colorado Hispanics. I think he didn't do too badly with Massachusetts minorities.
Data on Nevada is ambiguous at best and been called into question. THere is no data from CO, so that is unknown. As for MA, again there is no Latino breakout, but he lost non-whites 61/39 and Latinos are the largest portion of non-whites in MA. And when you look at southern states where they breakout non-whites and blacks, the numbers are about the sames, meaning that Latinos, Asians and Other are not helping Sanders to offset losses with blacks. The one state where that isn't true is OK, and so the assumption is that he did well with the large native American population.
But again, there is only one state with unambiguous data on Latinos and that is TX, where he got crushed. And that is also a primary, not a caucus. Recent polling also shows Sanders doing poorly in NM, which has a good number or liberal whites, so again if he were winning with Latinos he should be ahead. Same is true for polling in FL.
This gets to my original point, if Sanders was winning Latinos, then he had a shot. So far, there is no evidence that is going to happen.