I don't think it's possible to perfectly predict. Sometimes you might get lucky enough to spot it but to lay out rules for that would be not accurate as it's not possible in my humble opinion.
I think so too. You can rely on TA, try to follow MACD, but these things work only when it's already crashing (and even then, not always). The recent correction was almost ideal when you were following fibonacci retracements. I'm always setting up my buys a bit above the predicted bottom and this time it was very easy to catch some cheap coins. I did well with ETH and BTC. LTC was being quite FUD proof this time and lost much less than people were hoping for.
I still think that the main factor is the political situation followed by the news. If bad news are followed by another wave of bad news, it's possible to break low supports and go deeper than the TA would suggest. At the same time if good news follow good news during a bull run you can expect a new high.