Consider the Elliott wave picture above where about an 80 point decline in the S&P 500 can be seen from C to 1. The reason why from the all-time high the wave count in color looks best is determined by considering all options for what the decline can be and weighing each of these options by how good (probable) they look.
A complete corrective pattern down to 1 is difficult to view. Obviously a complete corrective pattern is required if there is to be a new all-time high before moving to new lows. Because the retracement of [iv] is so massive, it is safe to say [iv] is not a 1st wave, wave 'a' of a zigzag, or wave 'w' of a double zigzag. On the other side, taking [ii] to be the high of the last wave makes little sense in the larger context (the structure of the rally from B) and due to the structure of what would be the wave higher to [ii] (it looks like a 3-wave pattern).
So in looking for a complete corrective wave down from C to 1, one can see that it is difficult to find something better than the wave count above in color since C. Using the impulse waves lower as 'a' or 'c' waves of zigzags does not fit into a larger pattern. Breaking them up to take perhaps C to b of (ii) as an 'a' wave for a zigzag 'w' to (iii) and double zigzag to 1 is not a good option due to the structure of 'a' of 'w'; the size of the corrections within 'a' are not at all similar and there is an uncommon overlap between 2nd and 4th waves. Changing the [ii] to (ii) region to be a pair of 1st and 2nd waves leaves the problem that a zigzag wave 'c' from [ii] to 1 is far larger than wave 1. Legs of zigzags and double zigzags tend to be similar in size. There are several other possibilities to describe a complete zigzag pattern lower, but they all have similar if not worse problems. Looking for an alternate pattern lower like a flat from [iii] or C would be even more strange, something I realy have never seen in years of tracking this market.
The reasoning behind an incomplete decline from the high is supported not only by the fact that the wave count for it looks sound, but larger time frames are also pointing to the same possibility. This is a topic that I have produced much material on recently and can be found on my blog or website.