With that, let's have a look at the month of October, 2018.
S&P500 Cash - Daily - High/Low Plotted in Order - October 2018 |
So, here you can pretty clearly see the contracting leading diagonal minute wave ((i)) as counted in near real time on this blog. The chart doesn't change anything, but it does tend to abstract the wave just a bit. It's very clear in this manner where the "big moves" are, but it's a little harder to see the internal detail. That being the case, let's focus on the most important detail in this chart. That is that wave minuet (iii) is shorter in time than wave minuet (i). As shown, if T1 is the time of minuet (i), then that arrow, copied over, shows that wave minuet (iii) is shorter than (i). T3 is shorter than T1. It should also be clear to all that wave (v) is shorter than wave (iii) in time, as expected.
Neely terms what we call a contracting diagonal a "1st Extension Impulse". I don't think the renaming is needed. I think it creates more confusion than it helps resolve. He is very clear in the so-named wave that there is overlap between parts of wave two, and wave four. Sounds like a simple diagonal to me!
Now let's move on to the month of November.
S&P500 Cash - Daily - High/Low Plotted in Order - November 2018 |
In the second chart I have included the most significant lows and highs to serve and an anchor for you to view, so the chart covers a little more than a month. Besides the pretty clear three-wave (w) wave, the other item to note on this chart is that the time signature of the (x) wave is much longer than that of of the (w) wave. So, T2 as shown, is much longer in time than T1. It is not disproportionately so, but it is enough to be noticeable - perhaps about a third more.
The (y) wave is then shorter in time that both the (w) wave and the (x) wave which represents nice time alternation within the correction. Plotted in this manner, the (y) wave failure is clearly noted.
I am focusing on the time signature as we move into the month of December, below.
S&P500 Cash - Daily - High/Low Plotted in Order - December 2018 |
So here we see the most recent data - always the most difficult to deal with because one doesn't know for sure that a move is over with. But let's focus on two of the time signatures. The first clear result is that wave minuet (iii) of minute ((iii)) is clearly longer in time than wave minuet (i). In other words T3 is clearly longer than T1. That is important. Why? Because it allows us to assess than minuet (iii) may be the extended wave in the sequence - not just by price, but also by time. So, we mark that wave as x (iii) - extended wave (iii) in the sequence.
The next time signature I will leave for you to measure, but, as far as I can tell, wave minuet (iv) is already longer in time than wave minuet (ii). And that is the usual state of affairs in wave counting, is it not? So, at this point, even though the pattern is expanding a bit, there are no overlaps, and we should consider this count as a valid one - provided one thing is true.
That one thing is that minuet (iii) of minute ((iii)) as a sub-wave, must be shorter than all of minute ((i)).
I did that measurement tonight. At the wave that ends at the point of the arrow, it is exactly shorter by two points! Whew! That to me was itself an astounding result. So, either the wave ends at the arrow or wave (iii) ended at the bottom of the wave on 12/20 - and there is a long b wave in a flat - which adds more cushion. Either way, the count works. We also know that level is the 1.618 Fibonacci extension on wave minuet (i) of minute ((iii)).
In Mastering Elliott Wave, on page 5-14, Glen Neely shows a picture much resembling the impulse wave shown in the December chart. The second wave is somewhat shallow, and the fourth wave is deeper in relationship to it - which is already the case. In reality, the pattern is not expanding. What will most likely happen is that wave (v), if it occurs would end roughly parallel to wave (i) once the first line of the parallel is redrawn from waves (ii) to wave (iv) and the parallel copy is placed on wave (i).
We also know, by measurement, that wave (v) will most likely not become the extended wave for the simple reason that at it's current length wave (iii) is as long as degree labeling allows. So, wave (v) should not become longer than it.
Given all of this, we should have some pretty good degree of confidence a full impulse will form. Of course, we'll have to wait to see if it happens.
In the mean time, rest up. Enjoy the celebrations, and have a great start to the New Year!
TraderJoe