Last few bars dissapear on some NEWBAR events.

[mbuta]

Hi,
This is an issue I raised before. Sometimes [apparently when there is a tick righ on NEWBAR] last few bars [price and plots] dissapear for a brief moment, then they re-appear.

It looks exactly like I plot with GetSeries in external interval [when EFS2 engine keep the entire external bar in sync], but I am not. I do not use any external time interval with my studies [or plots]. The only reference to external intervals is to 1min NEWBAR [which I use to filter out executions].

I hope I explained it clear enough.

Can anyone give me some clues what can cause this?

Thank you.
Mihai

[ACM]

Mihai
Post a sample script that illustrates the issue
Alex

[mbuta]

Alex,

I already said that the problem does not happen all the time. I can't expect anybody to be able to reproduce my problem, I have to digg it myself. I just need some clues to what to look for.

Thanks.
Mihai

[ACM]

Mihai
Difficult to offer any guidance without having some common reference.
Alex

[mbuta]

So what do I do?
Nobody can make any suggestion?

Thank you.

[mbuta]

Last few bars dissapear on some NEWBAR events [2]

Hi,

I repost the question hoping to get some help.

Here is the problem:
Sometimes [apparently on NEWBAR event of an external timeinterval] the plots for the last few bars [I am pretty sure they match the number of bars of the external time interval] dissapear rbiefly, then reappear.
It is like I plot and external interval study with getSeries [which keeps them in sync], but I am not.

Can anybody give me some clues what can generate this?

Note: Please understand that it is an intermittent bug, so I doubt anybody can reproduce it.

All I am asking is some idea what [in priciple] can generate such behaviour.

Thank you.

[eSignal_JasonK]

Hello Mihai,

When a series object is being returned, which does not only happen with getSeries(), the EFS2 engine will be synchronizing the plot for a higher time frame series. If you are initializing a series object within main() and it's not inside an initialization routine that is being limited to one execution, returning that object name in the return statement will have the same affect as getSeries(). This process redraws the previous bar value to match the current value of the higher time frame bar. This behavior was created so that upon a reload of the chart, the historical values remain the same (or synched). If you do not want this behavior, return the .getValue(0) of the series object in question. Understand, however, that upon a reload of the formula, the history values will be changed, or synched, with the higher time frame because the formula can only process the completed bar information of the higher time frame.

If this does not help explain the behavior you are seeing, please post a code sample that illustrates the problem and I'll investigate further.

[mbuta]

Jason,

Thank you.
I will check again, but [1] all my studies are declared at ALLBARS time and [2] none of them are declared in external interval [I emulate my own, precicely because I could not keep them synchronized, as I "normalize" all of them].

I do calculate the exertnal intervals, but in premain.
I also use NEWBAR state of 1min, which I use to filter ticks/executions [ols story, as you know].

Problem is, does not happen all that often [maybe a few times a day] and only in real time [Tick Replay never caught it].

Thanks again.

[eSignal_JasonK]

Hello Mihai,

At this point, the only way to tell what the issue may be is to see some code. Please post a complete code sample that illustrates the problem so I can see the problem first hand on my end. Please give me some symbol/interval combinations you use it with and any custom time template settings that you may be using.