ProBacktest: strange behavior

Viewing 13 posts - 1 through 13 (of 13 total)

Author

Posts
07/20/2019 at 10:29 PM #103032 quote
othello
Participant

Senior
Hi guys.
I have written here some lines of code that I believe generate abnormal behavior.
This is the program.
```
p = 0.01

if d <=0 then
M = max(M,close)
endif

PP = (M - close)/ M
NP = PP >= p

if NP then
M = 0
d = 1
endif

once L = close
if d >=0 then
L = min(L,close)
endif

VP = (close - L) / L
NV = VP >= p

if NV then
L = close*100
d = -1
endif

buy at -close limit
graph d
graph M
graph L
graph PP
graph VP
graph NP
graph NV
```
The attached figure shows the value of the variables I am monitoring at the first candle.

The instrument is the bund (CFD) with time frame at one hour.

My question is:
why isn’t the division by zero error reported?
If we look at the code, the variable M, initially, is zero.
Thus, the PP variable should have the denominator at zero. However, the system does not report it.
Why?

Thank you.

fig1-2.png
07/20/2019 at 10:41 PM #103034 quote

Vonasi
Moderator

Master

On the first run through the code PP and VP are zero divided by zero. Perhaps zero divided by zero is an acceptable calculation whereas one divided by zero is not?

07/20/2019 at 10:48 PM #103036 quote

Nicolas
Keymaster

Master

M is equal to Close at first run because of line 4.

07/20/2019 at 11:09 PM #103039 quote

Vonasi
Moderator

Master

Nicolas wrote:

M is equal to Close at first run because of line 4.

Ooops. Missed that important bit. So the calculation is (close – close) / close and so there is no division by zero.

The wine I’m drinking must be even cheaper than the wine Nicolas is drinking!

07/21/2019 at 12:57 AM #103040 quote

Nicolas
Keymaster

Master

Dunno about the price but it’s untasteful. 😆

07/21/2019 at 8:08 AM #103050 quote

othello
Participant

Senior

Sorry, you’re right: it’s not a division by zero just because of the line of code number 4.

Obviously the wine I’m drinking is the worst! 🙂

07/21/2019 at 1:31 PM #103064 quote

othello
Participant

Senior

About to see the value contained in a given variable, by means of the “graph” instruction, I was wondering:
why the PP variable is not shown?
Initially I thought it could be due to some limitation of the “graph” instruction.
In the manual I have concerning ProBacktest and ProOrder (version 5.1.3 – 20170718) there is written in a note on page. 35, in fact, that up to 5 variables can be simultaneously displayed on the same graph.
However, referring to the online documentation concerning the v. 10.2 of ProRealTime, traceable to this link:https://www.prorealcode.com/documentation/graph/

no limitation is mentioned.

And besides, in the image previously posted, the value of 6 variables is displayed.

Investigating even more on this question, I also observed that the value of the PP variable is displayed starting from the candle at 13:00 on 16 September 2016 (I remember, for those wishing to replicate this situation, that the instrument is the Bund (CFD ) time frame 1 hour, 10,000 units) as illustrated by the image I attach here.

I can’t understand the reason. Some idea?
Thank you.

fig2-1.png

07/21/2019 at 2:44 PM #103067 quote

othello
Participant

Senior

I realized, only now, that with the last question I have crossed the boundaries of the thread (defined by the title of the thread itself).
Do I ask the moderators, therefore, that it is necessary to open another thread with a more appropriate title and relevant theme?
Sorry, in all ways.

07/21/2019 at 4:02 PM #103071 quote

robertogozzi
Moderator

Master

I don’t know what version you have.

I have v10.3 and it allows unlimited variables with GRAPH e GRAPHONPRICE 0n Demo accounts, while only 5 are allowed on Real accounts.

07/21/2019 at 4:27 PM #103072 quote

othello
Participant

Senior

Yes, I also have version 10.3. In this case the problem encountered remains (failure to display the PP variable)
Thanks Roberto.

07/21/2019 at 5:02 PM #103073 quote

othello
Participant

Senior

There are two other problems in the code. Or, at least, they seem so to me.

1. Why – again at the first candle – the value of the variable M is equal to zero?
If we follow the code we can see that in row 4 the variable M is given the value of the close of the candle (16504.7).
The PP variable to which the zero value is assigned is then calculated. Then, to the logical variable NP, the value zero (false) is assigned: this because PP is not greater than or equal to p (0.01). Therefore the cycle reported on lines 10-13 cannot be executed and, therefore, the variable M must remain at the value it had previously taken (16504.7).
From that line to the end of the program there is no longer any instruction that can cancel the value of M. But then, why is this null?
2. Why does the variable L, always after the first candle, take the value of 16450.1?
On line 15 there is an instruction that assigns to L the value of the close of the candle (16504.7). Then the cycle is executed (lines 16-18) which assigns to L the minimum value between L and the close. Therefore, the variable L should not change its value.
Subsequently the variables VP and NV are calculated and, by reason of their values, the cycle on lines 23-26 is not executed.
But then, why does L take on this value?

07/22/2019 at 2:15 PM #103110 quote

robertogozzi
Moderator

Master

On the same day and hour I can see ALL variables, you need to use your mouse to shrink windows 1 & 2 and enlarge window 1 then reducing its scale as much as possible (you should almost see a flat line for variables).

x-14.jpg

08/06/2019 at 6:53 PM #104309 quote

othello
Participant

Senior

I followed your suggestions and got the desired result. With a small difference, as can be seen from the attached image (the seventh variable, L, is not aligned with the others), but irrelevant.
Thanks Roberto.

fig1.png
Author

Posts