Logical Error in the Strategy Based on Open Types

[alex_laxya 10]

Hi,

I am facing some logical error in below code

 

inputs:timeopen(0945);
vars:  od(0);
od = (h-l)*4/5 ;
condition1 = (h-l) >= ((o - l) + od) ;
condition2 = close>open and (h-l)> average((h-l),10);
condition3 = low> low[1] + ((h[1]-l[1])/2) and c >c[1];
if time < timeopen then 
begin
if condition1 and condition2 and condition3 then buy 
this bar at close;
end;
if time<time[1] then begin
value1 = barssinceentry(1);
if value1 >=4 and value1<8 and high> highest(h,3)[1] then buy 
("fe") 2 shares next bar at market; 
end;
setexitonclose;

 

Logic is, first three conditions defines open type ( in this code it is open drive) in first half an hour after session open.

so far so good, no problem,

Now problem arises with second part of the logic, I want to buy highest high of last 5 bars ( tf is 15 min ) after 1 hr of the session start when above three conditions ( condition1, condition2 and condition3) are true.

Ex. Market opene at 9 am by 9.30 we get signal as open drive. We wait for an hour then afterwords wait for the opportunity to buy when high is greater than highest high of last 5 bars {( h >(h,5)[1]} but MC cant let it happen

because among previous three conditions there is one parameter which says time <0930 ( i.e. time is less than 9.30 am ) so all the code stop working after 9.30am

I have tried barssince entry, barcount = barcount+1 etc but it's somehow not working, I can compile the code but it doesnt reflect on chart as per logic.

posting the chart for reference..

 

 

Uploaded with ImageShack.us

[Tams 20]

Hi,

I am facing some logical error in below code

 

inputs:timeopen(0945);
vars:  od(0);
od = (h-l)*4/5 ;
condition1 = (h-l) >= ((o - l) + od) ;
condition2 = close>open and (h-l)> average((h-l),10);
condition3 = low> low[1] + ((h[1]-l[1])/2) and c >c[1];
if time < timeopen then 
begin
if condition1 and condition2 and condition3 then buy 
this bar at close;
end;
if time<time[1] then begin
value1 = barssinceentry(1);
if value1 >=4 and value1<8 and high> highest(h,3)[1] then buy 
("fe") 2 shares next bar at market; 
end;
setexitonclose;

 

Logic is, first three conditions defines open type ( in this code it is open drive) in first half an hour after session open.

so far so good, no problem,

Now problem arises with second part of the logic, I want to buy highest high of last 5 bars ( tf is 15 min ) after 1 hr of the session start when above three conditions ( condition1, condition2 and condition3) are true.

Ex. Market opene at 9 am by 9.30 we get signal as open drive. We wait for an hour then afterwords wait for the opportunity to buy when high is greater than highest high of last 5 bars {( h >(h,5)[1]} but MC cant let it happen

because among previous three conditions there is one parameter which says time <0930 ( i.e. time is less than 9.30 am ) so all the code stop working after 9.30am

I have tried barssince entry, barcount = barcount+1 etc but it's somehow not working, I can compile the code but it doesnt reflect on chart as per logic.

posting the chart for reference..

 

 

Uploaded with ImageShack.us

 

tips:

 

1. draw a flow chart, map out ALL the permutations !!!

 

2. write out your "logic" in short sentences

-- one thought at a time,

-- one thought per sentence, (if you have more than 1 thought in a sentence, split it up)

-- one action per sentence, (if you have more than 1 action in a sentence, split it up)

-- create/use variables/counters generously, they are free.

 

GL

[alex_laxya 10]

Thanks Tams for the advice.. Well I fragmented the ideas step by step and also tried to print the signals from MC to text file.

Here I found something problematic.

 

I was trying to debug the code one structure at a time, and seems like problem lies in time< time[1] ( or date<> date[1] ) line. Here is an experiment..

Run this code without time<time[1] ( or date<> date[1] )

vars: highesthi(0);
if time>0900 and time<1500 {and date<>date[1]}
then begin
highesthi = highest(high[1],5);

if high>highesthi then buy next bar at market;
end;
setexitonclose;

 

you will get accurate signals

now add time>time[1] or date<>date[1]

 

vars: highesthi(0);
if time>0900 and time<1500 and date<>date[1]
then begin
highesthi = highest(high[1],5);

if high>highesthi then buy next bar at market;
end;
setexitonclose;

 

now see the chart, it give erroneous signals.

I can't understand why this is happening.

[onesmith 10]

If you would have clearly stated what you wanted to do then someone would have coded it by now. The problem as I see it is I do not understand what you want. That doesn't mean I haven't enjoyed taking apart your code, simplifying it, and adding what I thought you might want.

 

Do not take this as me making a committment to code the strategy or offer further help. I often code things I see on the internet, to take my mind off other things, but I hope someday I won't feel the desire for these types of distractions.

 

I assure you the problems you're experiencing are insignificant and easily solved. I suggest following Tams advice of breaking this into segments that acomplish specific goals. After you've done that and have a smaller objective that's more easily defined, then request a solution that achieves a smaller step toward your larger objective. By breaking this into subsets of the larger project it will be easier (at least for me) to understand what you want this code to do.

 

At that point, it is best if you post a specific test case including the symbol, timeframe, date and time of an expected result. It is always easier to solve a specific case of an unexpected result. It is best if you tell us exactly which bar you want something to happen on instead of thinking I can figure out what you want via process of elimination of all the possibilities by using the info you've already posted.

 

The more specific you can be when describing what you want, the more specific and functional the solution will be.

[alex_laxya 10]

Thanks Smith for the advice.

The logic is

 

Buy the break out of opening range. Time frame for opening range is one hour. After one hour fifteen minute if Opening Range has not been broken then strategy would look for break out of highest high of last 4 bars ( time frame is 15 min through out the chart ) provided the open type is either open drive, open rejection reverse up or open test drive up ( Market profile logic, I have already coded those open types and those codes give signal within first half hour, then it stop calculating, this part has no issues )

 

I have already posted the code for open drive in first post ( OD = Open Drive )

 

Attached the sample data with this post.

 

 psudo Code is 
// code for open drive 
time > 0900 and time < 0930 then begin 
type of open  = Open drive 
end;

//opening range break out
if time > 1000 and time < 1015 then begin 
if high > opening range then buy next bar at market; 
end; 
if marketposition = 0 then begin
high > highest ( high,5)[1] and open type is open drive 
end;
setexitonclose;

 

This is the intra day strategy and problem is if I start the code as date <> date[1] or time < time[1] then MC gives error. That is the major obstacle for me and I can't figure it out why is this happening.

data for test.txt

[Tams 20]

Thanks Smith for the advice.

The logic is

 

Buy the break out of opening range. Time frame for opening range is one hour. After one hour fifteen minute if Opening Range has not been broken then strategy would look for break out of highest high of last 4 bars ( time frame is 15 min through out the chart ) provided the open type is either open drive, open rejection reverse up or open test drive up ( Market profile logic, I have already coded those open types and those codes give signal within first half hour, then it stop calculating, this part has no issues )

 

I have already posted the code for open drive in first post ( OD = Open Drive )

 

Attached the sample data with this post.

...

 

What you have written is a blob.

You might think it makes sense.

Evidently it did no help you to see through the logic.

 

try to reformat your "blob" into simpler thoughts:

 

 

Buy the break out of opening range.

Time frame for opening range is one hour.

After one hour fifteen minute

.... if Opening Range has not been broken

.... then

....strategy would look for

.... ....break out of highest high of last 4 bars

.... ....( time frame is 15 min through out the chart )

.... ....provided

.... .... .... the open type is either

.... .... .... .... open drive,

.... .... .... .... open rejection reverse up

.... .... .... .... or open test drive up

 

 

note how you breakdown your blob into

one thought per line,

one action per line

one line per sentence...

 

 

if you are attempting anything half sophisticated,

you cannot escape the effort of a flow chart.

Edited January 25, 2012 by Tams

[onesmith 10]

The complexity of, for instance your conditions 1, 2, and 3 is unnecessary. The purpose those conditions serve is to trigger true or false. So simplify that (because you said they're working ok?)

 

simplify it to

if t=1100 then myCondition=true;

 

You can replace that simplicity with your complex conditions after you solve the parts you've been unable to figure out. When you get that figured out and add the complexity back then sometimes it's necessary to make minor adjustments because the simplification might not have taken into consideration some specific quality of your real conditions, but this is always a minor adjustment, and it shouldn't be necessary, because it's easy to pick the correct simplication. Although in the case of your specific problem, you've failed to convey what that problem is so you can expect further debugging will be required if you're unable to pick up on how this debugging process works, and make the appropriate adjustment to the simplification. So adding the complexity back into the strategy could bring in a new issue that wasn't built-in to the simplification.

 

The secondary stuff about exceeding highest H can often be simplifed to something like

if c>c[1] then buy next bar market;

 

The bulk of the problem that still exists is I do not know what you are having a problem with other than you want to issue a buy order, or mulitple buys, and the wrapper (if d<>d[1]) that controls access to a gate that resets some values .. prevents issuing those buy orders. If that's the problem then it's simple to set a new variable to true, and use that new variable to control access to a new branch, which is where that specific case of buy order is located.

 

Do you have access to the emini S&P futures contract named ES? If you do then format the time zone to Exchange_Time. Or the US stock symbol for Microsoft? MSFT? Otherwise pick a symbol so we are on the same page. I mention this because I'm unable to load external data files.

 

When we've solved the issue of being on the same page as for symbol and timezone then please pick the simplist strategy logic you can, and post the entire (complete) code, up to the point where your first question begins. Then ask how to solve a specific test case. For instance, date and time when you expect your code to buy and it fails to buy.

 

Telling me MC gives an error does not give me any information that will enable me to help you. Please do not refer back to opening drives, other posts or market profile because for instance your request to look back at od=(H-L)*4/5 just tells me there's multiple layers of bugs that you are not seeing.

 

I recommend avoiding further description of the overall logic, until you have a template that can reset a switch to false at the beginning of the day, and set it to true via a simple c>c[1] event; From that point it will be preferrable if you speak in terms of events that open gates to different segments of the code, or an event that closes a gate and prevent access to a specific branch that needs to be accessed, because otherwise a buy does not occur on the bar following t=1100; Swapping out the actual events that open a logic gate, by substituting a simplified event in it's place will make it easy to distinguish which branches are open or closed.

Edited January 26, 2012 by onesmith

[alex_laxya 10]

Thanks a lot Tams and Smith for your time.

I don't have access to the s&p, emini or microsoft scrips on MC. I use NT for it, sorry for that, if possible you please send me the text file Smith so I could import it in to my MC.

 

Today was holiday for Indian markets so all day was busy in coding and simplifying it as Tams had suggested. ( even downloaded Mind Manager s/w for thought tree drawings to make a template. ) So far able to resolve flew glitches and few remains.

You are absolutely right Smith, I have already removed Market Profile part for the sake of simplicity and code one step at a time, I would test it and afterwords add other conditions. Will get back in a while with refined and much simpler code.

[alex_laxya 10]

Well, finally solved the riddle. As you guys suggested I go through step by step and add one condition at a time. System is working fine. When now I turn back and look at final strategy I can't help but think that It has become helluva complicated system.

The main issue of generating intraday signals using date<>date[1] has been solved by this logic

date<> date[1] then

begin

numberofbars = numberofbars +1 ;

if numberofbars > waittime/ barinterval then begin ( waittime is input value of 60 and barinterval is 30 so num of bars to wait are 60/30 = 2 )

 

Attached the .pla code to demonstrate the logic flow ( code wont work for others as functions for VAH,VAL, AND POC are not disclosed, as this thread is only about logical error in coding )

Thanks again Tams and Smith for kind support and patience.

_mpsystem.rar