Because the same peak can honestly be different waves, and the rules permit it. The end of a fifth wave, the end of a third, the end of wave B inside a correction — each version is self-consistent, and the chart alone cannot choose between them.
We show everything that passed the rules and call them readings.
Where the ambiguity comes from
Markup requires choosing a scale: what counts as a wave and what as noise inside it. The data does not set that scale. Shift it one step and what was wave 1 becomes part of wave A, and a five turns into a three.
So markup is worth building at several scales rather than one, checking which of them yield a self-consistent result. There is usually more than one survivor.
How many there are
As many as passed the rules. Sometimes three, sometimes two, sometimes none. "Exactly three options" is a frequent case rather than a property of the method, and there is nothing to invent the missing ones from.
If no numbering passes the rules, the honest answer is that there is no markup.
Why we do not pick one
Because there would be nothing to pick with. Any preference criterion — "more likely", "more common", "looks better" — is either taste or an appeal to statistics that wave analysis does not have: nobody has assembled the body of marked-up cases such frequencies would have to be counted from.
Showing the readings side by side leaves the choice to the reader and states plainly what would otherwise stay hidden: this situation admits several interpretations. A single scenario on screen would look more confident exactly to the degree that it was less honest.
How readings differ from each other
Each shows which wave the peak is taken to be, where the preceding legs sit, which wave is running now, and where it could end by the geometry of its form.
Often the versions differ not only in numbering but in the direction of the next move — and that is information in itself: a situation is uninformative when two permissible readings point opposite ways.
What several readings do not mean
They do not mean one is correct and the rest are noise. All of them are equally permissible under the rules. And they do not mean a prediction is hiding among them: each describes the structure of the past and a permissible geometry, not a future price.