Spydertrader's Jack Hershey Equities Journal

[Spydertrader]

As requested.

 

[svrz]

Quote from makosgu:

Thanks Spyder. The posting seams to say that the indicator will smooth out the price graph. Thus I will still be looking at either a Price or relative Price value... Y/N? Can anyone enlighten? If in fact MACD is a smoothing function, perhaps we can look at the rate of change of MACD... Does anyone have any light?

Unfortunately this is not very useful. If you, say, plot the 6 period ROC of MACD(5,13), you will see a near straight line with a few peaks that correspond to nothing useful.

In a way, you can look at MACD as the velocity of price. When it crosses over during a strong trend, it tells you that the trend has slowed. It could also be a signal of the trend reversal. Ambiguity is the pitfall of these indicators.

When MACD is within 2*tick (i.e. 1 for NQ -> tick = 0.5), it tells you that the velocity is near zero which means there is no trend.

Perhaps one of the best MACD derived indicators is Jurike's MACD which is derived from Jurike's propietary moving average. I think he sells it for $200. If I'm not in error, one of the regulars on Jack's ES forum, Laziz, used to utilize this indicator quite a bit. From what I understand, it is very close to being a no lag indicator.

I have tried very hard to create a such an indicator from the various zero lag MAs. I have also attemtped to mix and match regular MAs with zero log MAs to create the perfect MACD. I have not been successful.

I don't think this answers your question fully, makosgu, but I hope it at least provided a small hint.

Take care

 

[nwbprop]

Quote from Spydertrader:

As requested.

Sorry. I meant End of Day chart.

 

[Spydertrader]

Quote from nwbprop:

Sorry. I meant End of Day chart.

My mistake. I hope this is what you were looking for. See attached.

- Spydertrader

 

[makosgu]

Quote from svrz:

Unfortunately this is not very useful. If you, say, plot the 6 period ROC of MACD(5,13), you will see a near straight line with a few peaks that correspond to nothing useful.

In a way, you can look at MACD as the velocity of price. When it crosses over during a strong trend, it tells you that the trend has slowed. It could also be a signal of the trend reversal. Ambiguity is the pitfall of these indicators.

When MACD is within 2*tick (i.e. 1 for NQ -> tick = 0.5), it tells you that the velocity is near zero which means there is no trend.

Perhaps one of the best MACD derived indicators is Jurike's MACD which is derived from Jurike's propietary moving average. I think he sells it for $200. If I'm not in error, one of the regulars on Jack's ES forum, Laziz, used to utilize this indicator quite a bit. From what I understand, it is very close to being a no lag indicator.

I have tried very hard to create a such an indicator from the various zero lag MAs. I have also attemtped to mix and match regular MAs with zero log MAs to create the perfect MACD. I have not been successful.

I don't think this answers your question fully, makosgu, but I hope it at least provided a small hint.

Take care

Thaks for your comments. I find them extremly interesting... I am particular curious about the reasons for a 6 period ROC of MACD as opposed to a single period? Intuitively, it seams to me that the purpose of MACD would be to minimize the identification of false price trends while maximum the ability to identify the correct price trend as soon as possible. I would assume that the optimal settings is a compromise between the shortest smoothing period that yeilds the highest degree of correctly identifying the type of price trend that is actually in place. Is there some sort of uncertainty principle boundary that I am fighting against???

As I see it, MACD is useful for resolving 2 criterias, 1- an acceptable degree of correctly identifying price trends, and 2 - notification of the prevailing trend as soon as possible, hence the calibration of MACD.

Since MACD is smoothing the series according to our preference, then we should only need to evaluate ROC on a single period of MACD. Assuming an EOD fractal, I would EXPECT a 6 day ROC to be nearly 0. Reason being is that our stocks, as someone had calculated in some posting of another site, has an UP cycle length of 6.6 days on average. However, closer inspection of the velocity during this up cycle period shows that the velocity of the vehicle goes from 0 to max velocity then back to 0. Thus a 6 day ROC would be comparing the rate change on days 1 and 6 and averaging this change over 6 days (nearly 0). Optimally we want to observe the ROC on every day, thus I would assume a single period ROC???

G33M4K the Newb

 

[Spydertrader]

Sevens - Ones - Zeros
VSEC - TASR - YHOO
PPDI - HTCH - UTHR
PENN - HDWR - JUPM
SIE - PETD - CNC
CREE - UPL - KNGT
CYTC - PMTI - CCBI
SINT - CRDN - ECSI
ADS - URBN - GILD
DDN - INSP - BSTE
TKC - CFC - EAGL

Hot List:

TASR
VSEC
PMTI
JUPM
ECSI
PETD

Watch List:

TASR - Du Cycle - DU5
HRT - Du Cycle - DU5 - Du10
NVEC - Du Cycle - DU5 - Du10 - Du20
IDSA - Du Cycle - DU5 - Du10
IDSY - Du Cycle - DU5 - Du10
CALM - Du Cycle - DU5 - Du10 - Du20
LWAY - Du Cycle - DU5 - Du10 - Du20
ANIK - Du Cycle - DU5 - Du10 - Du20
SWIR - Du Cycle - DU5
EVCI - Du Cycle - DU5
LIFC - Du Cycle - DU5 - Du10 - Du20
NAVR - Du Cycle
ENWV - Du Cycle
SMTS - Du Cycle - DU5
HLEX - Du Cycle - DU5
LSCP - Du Cycle - DU5

(HRT, VSEC, NVEC, IDSA & LWAY have float outside normal parameters)

Wealth lab Seven Hershey Equations:

ESMC - Du2
HANS - Du1 - Du2 - DU3
IDSA - DU1
SWIR - DU2 - Du3
TASR - Du1 - Du2 - Du3
TRMM - Du2 - Du3
UBET - Du1

Scores for above stocks:

ESMC - 0
HANS - 7
IDSA - 0
SWIR - 2
TASR - 6
TRMM - 0
UBET - 6

 

[makosgu]

Quote from makosgu:

...
Since MACD is smoothing the series according to our preference, then we should only need to evaluate ROC on a single period of MACD. Assuming an EOD fractal...

Alright, several more revelations fell into perspective but I'll share this when appropriate. I'll attempt to answer my own questions for those who may be having the same questions. So MACD is NOT a smoothed price curve. It is solely meant to INDICATE the existence of a trend. I should have heeded to the fact that it is recognized as an indicator. However, MACD has some of the constituents to smooth out prices. Doing some homework and thinking through the mechanisms will yield exactly what's needed to smooth out prices. Is or has anyone noticed this? If I am detracting from the thread, let me know and I will keep my thoughts to myself if my posts are not helping anyone. My only intention is to learn and contribute.

G33M4K the Newb

 

[Spydertrader]

Quote from makosgu:

If I am detracting from the thread, let me know and I will keep my thoughts to myself if my posts are not helping anyone. My only intention is to learn and contribute.

I both welcome and encourage contributions to this thread. Keep up the great work.

- Spydertrader

 

[Spydertrader]

I will be heading west to enjoy a last minute vacation in the mountains (and golf courses) of Arizona Sunday afternoon. As a result, I will not have access to a computer to post the daily lists until Saturday, October 23, 2004. I encourage continued involvement in the Journal and with the ongoing process of 'iterative refinement' in my absence. I will 'catch up' upon my return.

In regards to UBET, I have set quotetracker to email my mobile phone in the event of a 'stop loss' trigger or target price match. If neither event occurs, I will hold the stock.

Thanks again to everyone for their interest and participation in the Journal thus far. I appreciate the effort. Good trading this week, and I will post again upon my return.

- Spydertrader

 

[svrz]

Quote from makosgu:

Thaks for your comments. I find them extremly interesting... I am particular curious about the reasons for a 6 period ROC of MACD as opposed to a single period?

Well, by definition, MACD uses two MAs. Hence the reason for the double period and not single. Or perhaps I completely misunderstood your question?

Intuitively, it seams to me that the purpose of MACD would be to minimize the identification of false price trends while maximum the ability to identify the correct price trend as soon as possible.

That would be ideal but there are false crossovers. However Jack's version (5,13,6) does a very good job of aletring us as to when the trend has ended, in an expedient manner.

I would assume that the optimal settings is a compromise between the shortest smoothing period that yeilds the highest degree of correctly identifying the type of price trend that is actually in place.

Yes, exactly.

Is there some sort of uncertainty principle boundary that I am fighting against???

I'm not exactly sure what you mean by this statement.

As I see it, MACD is useful for resolving 2 criterias, 1- an acceptable degree of correctly identifying price trends, and 2 notification of the prevailing trend as soon as possible, hence the calibration of MACD.

Once again, you are correct.

Since MACD is smoothing the series according to our preference, then we should only need to evaluate ROC on a single period of MACD.

Well, as you well know MACD is the difference between two MAs. Perhaps you just want to calculate the ROC of one MA, no?

Assuming an EOD fractal, I would EXPECT a 6 day ROC to be nearly 0.

6 day ROC of an MA? MA of what period?

Reason being is that our stocks, as someone had calculated in some posting of another site, has an UP cycle length of 6.6 days on average.

Perhpas this is the reason that the signal line of Jack's MACD has the period as 6.

However, closer inspection of the velocity during this up cycle period shows that the velocity of the vehicle goes from 0 to max velocity then back to 0. Thus a 6 day ROC would be comparing the rate change on days 1 and 6 and averaging this change over 6 days (nearly 0). Optimally we want to observe the ROC on every day, thus I would assume a single period ROC???

ROC of price or ROC of MA of Price?

Take care.