1 / 4 · 6 min
Wave analysis and its inborn flaw
The idea is simple and elegant: a move runs in five waves, a correction in three. The flaw is that almost any chart fits several counts at once.
In the 1930s Ralph Elliott noticed that markets move not chaotically but in repeating sets: five waves in the direction of the main move, three against it. The observation is reasonable in itself — it describes what is visible without it: price moves in bursts with pullbacks, and bursts with the trend are longer than the pullbacks. The problems begin not with the idea but with what it turned into.
The main flaw: the alternative count
Almost any count has a spare. The fifth wave failed to confirm — then it was the third and we are in the fourth. That failed too — then the whole construction was wave one of a larger degree. A claim that always has a backup explanation cannot be refuted, and therefore cannot be confirmed. That is exactly why wave analysis looks so infallible in somebody's retelling and travels so badly to a live chart.
What is a rule and what is a guideline
| Statement | What it is | Gives an invalidation price |
|---|---|---|
| wave two does not pass the start of wave one | rule | yes |
| wave three is not the shortest of 1, 3, 5 | rule | yes |
| wave four does not enter wave one's territory | rule | yes |
| wave three is usually the longest | guideline | no |
| a correction usually reaches 0.618 | guideline | no |
| wave five is often equal to wave one | guideline | no |
Why that split decides everything
There are exactly three rules, and each gives a PRICE at which the count is declared wrong. That is the only thing turning waves from a story into a tool: you get a number you can put a stop on and by which you can count how often you were wrong. Guidelines give no such number. They are useful for expectations, not for decisions, and confusing them with rules means being left with no invalidation price at all.
Recounting after price went the other way
Almost everyone does this, and almost always without noticing. The count stopped fitting, you move the numbers, and now it fits again. Formally there is no error: no rule was broken. In substance you have just erased your own wrong forecast and replaced it with a correct one after the fact. There is one cure and it is unpleasant: write the count and the invalidation price BEFORE, as a separate list, and count how many times you had to change them. That number is the only honest measure of your wave analysis.
Looking for waves where there is no move
Inside a range anything can be labelled: price goes up and down and any five swings can be called an impulse. A wave count only makes sense where there is directional movement with clear structure — that is, in roughly a third of cases. The rest of the time the honest answer is "there is no count", and it is far more useful than a forced one.
What is genuinely valuable in waves
One thing other methods lack: they force you to think about PROPORTION and about where you are inside a move. The question "are we at the start of this run or at the end" is not asked at all by most approaches — they speak only of the present moment. Waves ask it directly, and even an approximate answer changes position size and expectation. It is for that question that the subject is worth knowing — with a clear understanding that the answer will be approximate.
There is exactly one way to test your count: in bar replay, labelling as candles appear and without seeing the future. The chart plays candle by candle and nothing is loaded ahead — not even into the indicators. If your count afterwards matches what you would have drawn on finished history, you have a working method. It usually does not match, and it is better to learn that without money.
The chart and bar replayTen counts with a written invalidation price
Take ten instruments and label the current move on each. For each write down TWO things: the number of the wave you believe you are in, and the price at which your count becomes wrong under one of the three rules. Come back a week later and count: how many times the invalidation price was reached, and how many times you wanted to "refine" the count instead of admitting it was wrong. The second number is usually larger than the first — and that is the thing to work on.
Why is the possibility of an alternative count a flaw rather than flexibility?
Because a forecast that comes true under any outcome tells you nothing. If you are right when price rises ("wave three is running") and also right when it falls ("so that was wave four"), then your count does not distinguish between future states of the world — which is precisely what a forecast is for. Flexibility is useful in explaining the past and harmful in predicting the future, and wave analysis confuses those two tasks more than any other method.
Three rules and a dozen guidelines — why can they not be treated the same?
Because only rules give an invalidation price. The guideline "wave three is usually the longest" does not tell you where to put a stop: the word "usually" does not convert into a number. The rule "wave three is not the shortest" does convert: the moment wave three becomes shorter than both one and five, the count is dead, and that is a specific price. Mixing them leaves a person with the confidence of guidelines and none of the protection of rules — the worst possible combination.