DJIA textbook support


#1

At this point at least, looks like textbook support for DJIA from the 10w FLD. Tomorrow will determine if it holds. I thought the sell-off might be done around mid-day when sitting at the 5w FLD. Checked 15 min bars and saw there was more time left so I held onto my short until right before the close.


#2

Is this the trade that you complained of leaving 400 points on the table? It was still remarkably accurate, especially if Friday was the low, and for a weekly horizon.


#3

Yup. 29 trading day wave is dominant. Got jittery and bailed after 28 days with the Dow down over 700 points. Rallied weakly the next day (3/23) then crapped out in the last two hours of trading, just like 29 trading days before on 2/9. The high amplitude of the 29 day wave and the FLD interactions make this a textbook example but the wave is longer than a historical one of this magnitude by about 5 days. Still remarkably consistent lows of 11/15, 12/28, 2/9, and 3/23. Only a good manual phasing or a spectral filter (or possibly ST) would pick this up in real time.

I have been shocked at the accuracy of this stuff over much longer time horizons.


#4

Were you expecting more bearishness in DJIA this week based on the impending trough around Apr 30? Could the subdued reaction be indicative of a bullish bounce on Monday or Tuesday?


#5

I’ve learned the market does not care about my expectations. Drilling down to intraday futures, we are pretty much on schedule. Still expect a low of some type at the end of next week. It could be brief and violent if it hasn’t already happened. Looking like the 40w nominal trough will come in late May/early June which would make this a 5w trough. Don’t think the real heavy stuff is coming down until the second half of the year.


#6

Not sure about 40 week, but I expect nominal 20w cycle price low next Sunday. Forgot to mention, a bounce from the coming low will not impressive, probably flat into 1st week of June.


#7

Price through 4/21.


#8

I have a slightly different take due to the fact that my price waves are shorter. The 20 week price wave is cycling into a lower high. The 10 week price wave has already put in a second lower high, implying that the absolute high of the 4 year price wave is in. Sigma L hasn’t turned down yet but a lower 20 week low will most likely roll it over.


#9

Hi William,

My cleaned up version without the long cycles expects the 40w trough around 5/17-18. Since the 20w filter is extrapolating 10w of data, if we take a real digger over the next month could your filter restate the last 20w trough? You have five 20w price waves in the 18 months from 2/16 to 8/17. Do you still call the 8/17 low an 18m trough or do you look at the 18m filter output to determine this. The conversion from traditional nominal cycles to price waves is confusing to me. Maybe I should stop trying to compare the two.

Did you normalize price? How do you measure Sigma L with filters? High pass filter > than 4 yr wave?

Sorry for all the questions but you know me.

Curt


#10

The lag on the filter construction is small. Once a high or low “forms” and the filter catches up to it, its location in time does not change. The extrapolation through the lag period and into the future is based on prior oscillations and the current data. Therefore there is always going to be a very small amount of error at nowtime. However as you move further into the past into the lag period, the extrapolation error becomes less and less until the point of the filter output where there is no error. The most recent 20 week low depicted won’t change (maybe 1 day or so) regardless of the future price action. What is relevant to me is the fact that the filter output is phasing to the highs instead of the lows now.

I’m not sure what you mean by “normalize price.” According to Hurst Sigma L is the sum of all the waves longer than the 80 week price wave. Therefore it is the high pass band of the 80 week wave filter.


#12

I’m not familiar with the price of the index you are using. Added one more question about labeling troughs above.

Actually, my understanding of Sigma L is that it is the sum of all the longer cycles not identified by phasing. The reason I thought yours was estimated based on waves longer than 4 years is because that is the longest filter you use.


#13

Now I realize to what you were referring. That chart consists of the log values of the YM futures. Sorry.

Sigma L is just a generic description. I could just as easily plot the high pass band of the 20 week wave shown and it would represent the sum of the waves longer than the 20 week wave. The same is true of any wave. Therefore Sigma L does not refer to any specific wave but is merely a function of the frequencies included in its structure. The dashed line on the chart just happens to be the high pass band for the 80 week wave. The high pass band for the 4 year wave would be much “flatter” because the cutoff frequency is 3 times longer. I would also have to use a lot more data to get an accurate calculation.


#14

Dow weekly bars with 40w and 18m nominal FLDs plus Market Profile support and resistance levels. We might put up a little fight near the recent lows over the next few days but this thing probably has eyes for the 18m FLD over the next week or two. This would likely be the 40w low.


#15

Weekly composite looks as bad as ever, not like it matters, but I think we should get some bounce next week. .


#16

Looks right on the money, especially the largest power cycle (39 calendar days). We might get a bounce around 5/9-10 and then down again to around 5/17-18. Those dates coincide nicely with two troughs in the 18m FLD. TWT.


#17

Yes, 39 day is very much dominant, we should get a bounce from the cycle low. Also 39 day cycle to noise ratio is very high.


#18

It is possible this wave may have extended to 47-48 days over the past 6-7 weeks. Bartels confirms.


#19

Yes, I see it. Market went up to fast, and arguably one day earlier, needs a pullback. Now about nominal 39 day cycle, stretching it to 47-48 days is bit of ‘stretch’ in my book. Usually I look at the band of about 10% variation and than average. If we look at weekly and monthly mods, I would not be looking to buy, and if my time frame was say 19 minutes (used to trade 5x19x67 minutes) , I would be in and out inside of one day and probably would cash out most of the move on Friday. Strongest cycle with highest cycle/noise ratio for the dominant band 35 - 43 is 39 days, than and abruptly strength start to increase starting from 45 days up to 49. Normal model when tested with past data does better without 48 day cycle. if I am wrong, immediate move below last swing low on the daily chart, would confirm overall weekly and monthly bearishness.

44 days is a divide between what I as two distinct cycle bands


#20

Longer term, the 18 month wave is clearly dominant. Compare the current wave beginning 8/17 to the 40w wave beginning 2/16 and ending 11/16. Breadth also looks incomplete.


#21

If you are using standard Hurst model than low of 40 w 20 w … is around May 18. Try to find 40 w on the Bartels scale, not easy is it? Also Composite model does not test well. Best strategy is to ignore 400 points move and if there is low around May 18 than buy. If we go up from here in to the end of May, than short starts to look very appealing.