Wave analysis breaks down not on hard cases but on discipline. It has few rules and a great deal of freedom, and almost every mistake is a place where the freedom was used without looking.
Counting after the fact
On history the count always works out: the reversals are visible and the numbers fall into place. In real time that data does not exist — the last extreme is unconfirmed, and the current wave may turn out to be anything.
A test: if your count is flawless on history and falls apart at the fresh edge of the chart, that is neither skill nor bad luck. It is the difference between knowing the outcome and not.
Mixing degrees
"The fifth wave ended, but a lower-degree third is running" joins statements about different scales. They neither contradict each other nor add up: a fifth wave of the higher degree contains a full set of lower ones inside it.
Before drawing a conclusion, say out loud which degree you are talking about.
Ignoring the hard rules
There are only three, and they are not negotiable: wave 2 does not go beyond the start of wave 1, wave 4 does not enter wave 1's territory, wave 3 passes the end of wave 1.
The temptation to "bend one slightly" appears when a beautiful count fails on a single point. A count that breaks a rule is wrong — that is not pedantry, it is the only thing separating wave analysis from drawing lines to taste.
There is exactly one exception and it is part of the canon: in a diagonal, overlap between waves 1 and 4 is allowed, with its own rule in place of the suspended one.
One count instead of several
A situation admitting exactly one count is rare. Usually there are several, and they point in different directions.
Picking one and forgetting the rest feels good, and that is precisely what turns analysis into prediction. It is more useful to keep the alternatives and know which price move would cancel which.
Counting towards a forecast
The most common motivation and the worst method. If a count is built to reach a desired conclusion, one will always be found: there are many scales, many corrective forms, and a diagonal permits overlap.
The reverse order is sounder: first mark up what is visible, then see what continuations it admits — including the ones you do not like.
Expecting price to reach the zone
Fibonacci zones are typical proportions, not obligations. Price regularly falls short of them and regularly overshoots.
Failing to reach a zone does not make a count wrong. Only breaking a rule does.