Gold Customer
Joined: 1/25/2005 Posts: 29
|
This PCF is telling responding with 'Syntax error' ... HHHEEEELLLLPPPP! ((O >= C1 * 1.05 OR O1 >= C2 * 1.05 OR O2 >= C3 * 1.05 OR O3 >= C4 * 1.05 OR O4 >= C5 * 1.05 OR O5 >= C6 * 1.05 OR O6 >= C7 * 1.05 ) AND MS1 > 30 AND BOP1 > 30 AND (AVGC10 > (AVGC40 * 1.05))
|
|
Worden Trainer
Joined: 10/7/2004 Posts: 65,138
|
It appears to be a parentheses problem. At the very least, your parentheses were closed once less than they were opened. You may wish to try the following:
(O >= C1 * 1.05 OR O1 >= C2 * 1.05 OR O2 >= C3 * 1.05 OR O3 >= C4 * 1.05 OR O4 >= C5 * 1.05 OR O5 >= C6 * 1.05 OR O6 >= C7 * 1.05) AND MS1 > 30 AND BOP1 > 30 AND AVGC10 > AVGC40 * 1.05
-Bruce Personal Criteria Formulas TC2000 Support Articles
|
|
Registered User Joined: 1/1/2005 Posts: 2,645
|
whitepaw98,
Your syntax error is a missing parenthesis at the end. The proper syntax is:
((O >= C1 * 1.05 OR O1 >= C2 * 1.05 OR O2 >= C3 * 1.05 OR O3 >= C4 * 1.05 OR O4 >= C5 * 1.05 OR O5 >= C6 * 1.05 OR O6 >= C7 * 1.05 ) AND MS1 > 30 AND BOP1 > 30 AND (AVGC10 > (AVGC40 * 1.05)))
Unfortunately, no such meaningless PCF should ever be written. Whether it returns True or False depends on the other PCFs calculated at the same time. The MS1 > 30 makes no sense whatever.
Your source has a PCF not intended.
Thanks, Jim Murphy
|
|
Registered User Joined: 1/1/2005 Posts: 2,645
|
whitepaw98,
I did not see Bruce's post until after I posted. As a trainer, he is not in position to offer advice such as mine, but you will find that he will not disagree with what I said. (If he does, he will lose all of his trophies.)
Thanks, Jim Murphy
|
|
Worden Trainer
Joined: 10/7/2004 Posts: 65,138
|
bustermu, The trainers cannot give settings, interpretation or investment advice, but I can point out when something is mathematically meaningless. I should have noticed the MS1 > 30 portion of the formula and pointed it out.
whitepaw98, You may wish to review the following to see why MS1 > 30 doesn't make sense in the context of your Personal Criteria Formula:
Dealing with OBV & MS in PCF's - how to interp their "values" OBV Unstable?
-Bruce Personal Criteria Formulas TC2000 Support Articles
|
|
Registered User Joined: 1/1/2005 Posts: 2,645
|
The PCF:
MS1.0-MS1.1 > 30
is an example of one that does not make sense.
The PCF:
MS1
is equivalent to:
MS1.0-MS1.P
for some positive integer P which depends on the other PCFs calculated at the same time.
We might say that if:
MS1.0-MS1.1 > 30
does not make sense for trading purposes, then:
MS1 > 30
is an absurdity for trading purposes.
Thanks, Jim Murphy
|
|
Gold Customer
Joined: 1/25/2005 Posts: 29
|
In your opinion, what value for MS would make sense?
|
|
Worden Trainer
Joined: 10/7/2004 Posts: 65,138
|
whitepaw98, It's not obvious and probably a bit counter-intuitive, but there isn't a specific value that would make sense. The actual value of MoneyStream is meaningless because it is a Cumulative Indicator. MoneyStram values can only legitimately be compared to other values of MoneyStream for the same symbol calculated at the same time. So, while MS > 30 will evaluate, it doesn't actually make sense as a Personal Criteria Formula because MoneyStream is being compared to fixed value and not another MoneyStream value for that symbol. The topics referenced in my Wednesday, September 13, 2006 11:00:45 AM ET post explain this concept in more detail and provide examples of how to correctly use Cumulative Indicators in PCFs.
-Bruce Personal Criteria Formulas TC2000 Support Articles
|
|
Registered User Joined: 1/28/2005 Posts: 6,049
|
whitepaw98
Lets say stock A trades an average of 1 million shares a day.
Stock B trades an average of 200,000.
Even though it is true that stock A has 5 times more volume.
If they traded the same values tomorrow. It would not indicate anything. (they are both trading "normal" volume)
The only way you can tell where volume is relative to themselves.
Things like BOP,TSV,MS are in the same category. Their actual value being higher or lower than other stocks does not matter.
One possible solution is a scan where you pick a time period (10days,20days whatever) and ask:
Is TSV(today)>(TSV days ago) Is MS(today)>(MS days ago) Is BOP(today)>(BOP days ago)
Hope that helps diceman
|
|
Registered User Joined: 1/1/2005 Posts: 2,645
|
QUOTE (diceman) Things like BOP,TSV,MS are in the same category. Their actual value being higher or lower than other stocks does not matter.
One possible solution is a scan where you pick a time period (10days,20days whatever) and ask:
Is TSV(today)>(TSV days ago) Is MS(today)>(MS days ago) Is BOP(today)>(BOP days ago)
diceman,
I don't think you will get anyone to agree with the above.
As far as scale change and translation are concerned, each PCF in any of the three should satisfy the following:
1) A PCF in MS should be invariant with both scale change and translation.
2) A PCF in TSV should be invariant with scale change but not necessarily with translation.
3) A PCF in BOP need not be invariant with scale change nor translation.
Let PCF( X ) be a PCF involving the indicator X.
a) A PCF( X ) in X is said to be invariant with scale change in X if and only if:
PCF( a*X ) = PCF( X )
for all real a > 0.
b) A PCF( X ) in X is said to be invariant with translation in X if and only if:
PCF( X+b ) = PCF( X )
for all real b.
Examples which satisfy the requirements are:
1) MS1.0-MS1.1 > 0, (MS1.0-MIN(MS1,10))/(MAX(MS1,10)-MIN(MS1,10))
2) TSV18 > 0 TSV18/(MAX(TSV18,10)-MIN(TSV18,10))
3) BOP1 >30 BOP10
Examples which do not satisfy the reqirements are:
1) MS1 > 0 MS1.0-MS1.10
2) TSV18 > 30 TSV18.0-TSV18.10
3) There are none.
As can be seen, the three indicators require quite different forms within a PCF in order to be meaningful for trading purposes.
Thanks, Jim Murphy
|
|
Registered User Joined: 1/1/2005 Posts: 2,645
|
QUOTE (whitepaw98) In your opinion, what value for MS would make sense?
whitepaw98,
I suggest you contact Sir Lazy-Jumper directly and ask him what he intended in place of MS1 > 30. He is the only one that could possibly know the answer to that.
Thanks, Jim Murphy
|
|
Registered User Joined: 1/28/2005 Posts: 6,049
|
bustermu
Remember I stated one "possible" solution.
The way I see it people look for divergences in these indicators.
Therefore it would seem to be of some use to compare present values to previous values.
As I understand it if BOP was red yesterday and green today that is some type of improvement.
If TSV is trending upward that is some type of improvement.
If MS trends upward with price that is considered good.
It should be noted however that TSV and BOP appear to be in an oscillator format.
So in theory there values can be lower than a few days ago but still be viewed as positive.
Thanks diceman
|
|
Worden Trainer
Joined: 10/1/2004 Posts: 18,819
|
Another thing to keep in mind... unlike MS and TSV, BOP is a bound indicator. It cannot register values below -100 or above +100.
- Craig Here to Help!
|
|
Registered User Joined: 1/1/2005 Posts: 2,645
|
diceman,
Just because a PCF is acceptable according to the invariance criteria described above does not mean it is particularly useful. Let's examine some examples.
First, consider the PCFs:
1) BOP1 > 30 2) TSV1 > 30 3) MS1 > 30
Only the first is acceptable by the invariance criteria. But, each implies the following acceptable PCF, respectively:
1) BOP1 > 30 2) TSV1 > 0 3) MS1.0-MS1.K > 0
1) Implies BOP1 is green which is considered Bullish. 2) Implies there is share accumulation (as measured by TSV) today which is considered Bullish. 3) Implies MS has a net increase from K days ago to today which is considered Bullish. The problem is that we need to know which PCFs were calculated along with the original 3) in order to determine K.
Now, suppose we are interested in what each indicator has done over the last 10 days. Consider the following acceptable PCFs:
1) BOP18.0 > BOP18.10 2) TSV18.0 > TSV18.10 3) MS18.0 > MS18.10
1) This can return True with BOP1 red and decreasing each of the last 10 days. 2) This can return True with TSV1 negative and decreasing each of the last 10 days. 3) This can return True with MS decreasing more than the previous day's decrease for each of the last 10 days.
The last set of examples should preclude using these forms for determining positive divergences.
Note; If you run tests, you will find that the TSV PCF is the worst offender. You will find that in about 80% the trials that:
TSV18.0>TS18.10 AND C0<C10
returns True, the PCF:
TSV10 < 0
also returns True, i.e., there is net share distribution as measured by TSV over the last 10 days.
Thanks, Jim Murphy
|
|
Registered User Joined: 1/28/2005 Posts: 6,049
|
Bustermu
I don't use BOP,TSV and MS to often. When I do look at them they are usually just on the screen.
So I am not too familiar with their equations.
Since your equations (under the word "Now") have 18 in them and your next set have the "raw" value. I would assume you are saying that the average(18 periods) can be up even though the unsmoothed value has gone down.
Isn't that just the cost of doing business?
If we want to scan for upward sloping stocks and we add something like:
AVGC50>AVGC50.25
It is quite possible that a strong stock could have a nasty drop which would not be reflected in the equation yet.
I think whatever smoothing we decide to use. We must put up with what ever pro's and con's it brings to the table.
Of course I may be just missing your point.
Thanks diceman
|
|
Registered User Joined: 1/1/2005 Posts: 2,645
|
QUOTE (diceman) If we want to scan for upward sloping stocks and we add something like:
AVGC50 > AVGC50.25
diceman,
Just let me say that a PCF of that form is probably the most popular way of determining an uptrend of smoothed data over the 25 days and at the same time the worst way that I have ever observed. The PCF can return True with C decreasing more than the previous day's decrease for each of the last 25 days.
We do not have to endure such circumstances. One alternative would be to require that the Linear Regression Slope of Period 25 be positive. Another alternative is a favorate of Craig's:
AVGC12 > AVGC12.13
One can find no criticism of the two alternatives such as that above of the original.
QUOTE (diceman) Isn't that just the cost of doing business?
No!
QUOTE (diceman) I think whatever smoothing we decide to use. We must put up with what ever pro's and con's it brings to the table.
If a smoothing filter has con's such as the original above, do not use it. You cannot get "smoother" than the straight line output of the Linear Regression Filter. (It filters, or, transforms, the data points over its period into a straight line.)
Thanks, Jim Murphy
|
|
Registered User Joined: 1/28/2005 Posts: 6,049
|
Bustermu
Realize that my point to whitepaw98 was that MS,TSV,BOP may be better used in a scan as they relate to themselves. I did not offer a method.
"1) BOP18.0 > BOP18.10 2) TSV18.0 > TSV18.10 3) MS18.0 > MS18.10"
Were methods offered by you. If there are "flaws" involved using the above that can be changed but the basic point still remains this is probably a more productive road to go down than: MS>30.
--------------------------------------------------------------------------------------------------
I believe there is a misunderstanding of terms here. When I talk about the "cost of doing business". I talk about in as it relates to the stock market. Our scans can be perfect. It does not mean our results will be. -------------------------------------------------------------------------------------------------- Realize that: AVGC50>AVGC50.25 was specifically chosen by me to be a "flawed" filter.
You claim that:
AVGC12>AVGC12.13 is an improved version. That may be true mathematically but there is no evidence that it will yield better stocks.
There can be no guarantee that a stock screened with: AVGC12>AVGC.13 will perform better than one screened with: AVGC50>AVGC50.25.
-------------------------------------------------------------------------------------------------
Every scan and every filter we use by definition eliminates certain stocks. That is the "cost of doing business".
If I create a scan:
C>C30 C>5 C<30 V>10000
I can find stocks that meet this criteria with certainty.
However it does not mean there cant be a stock where
C<C30 C>30 V<10000
that will perform better.
Every decision we make in the market has a cost associated with it.
Thanks diceman
|
|
Registered User Joined: 1/1/2005 Posts: 2,645
|
diceman,
I apologize. I interpreted your statement:
--------------- If we want to scan for upward sloping stocks and we add something like:
AVGC50 > AVGC50.25 _______________
to mean that we might use the PCF to find stocks that have moved up over the last 25 days.
You are correct. The above PCF and any I proposed will find stocks that will perform very well in the future and will find stocks that will perform very poorly in the future.
Thanks, Jim Murphy
|
|
Registered User Joined: 1/1/2005 Posts: 2,645
|
diceman,
I ran preliminary tests on the two PCFs:
1) AVGC50 > AVGC50.25
2) AVGC12 > AVGC12.13
1) 61.93% of those selected were profitable after 5 days.
2) 64.37% of those selected were profitable after 5 days.
The All Stocks WatchList was used with the 5 days ending on 09/13/06.
Of course, additional tests must be run to show which is the better predictor of the future if either.
It was not my original understanding that the purpose of either was to predict future behavior of prices.
Thanks, Jim Murphy
|
|
Guest-1 |