CloseD Series Function

[koganam]

- hope so but just making sure.

but thanks for helping out.

so say I have 500 5 minute sessions to backtest each one requiring a 100 day moving average - how many 5 minute bars do I need to load?

That is certainly information that you can calculate, but is pointless. Just load 100 days. OK. maybe 105 days, for a cushion, and be about it. Leave it to the platform to calculate and load however many 5-minute bars that is. Why do you need to know?

If you insist on knowing, as you are using only 5-minute bars, no matter how many sessions, it would not matter how many bars you have to load: it will be the same number, regardless of how many sessions you are running, albeit each session will have to load the bars for itself. It is still just the same data, being read from the same data files, so the bars will just be loaded from a cache, either by NinjaTrader or by the operating system.

[futurenets]

I realize this may be frustrating for you but please bear with me, this discussion has been very useful ...

I just want to be aware of approx. how many 5 minute bars would actually need to be downloaded

EURUSD 24 hours session has 24*60/5 = 288 5 minute bars

daily MA at session 0 would need session 1 to session 100 bars = 288 * 100 = 28,800
daily MA at session 1 would need session 2 to session 101 bars = 28,800 + (288 * 1) = 29088
daily MA at session 2 would need session 3 to session 102 bars = 28,800 + (288 * 2) = 29376
...
daily MA at session 100 would need session 101 to session 200 bars = 28,800 + (288 * 100) = 57600
daily MA at session 200 would need session 201 to session 300 bars = 28,800 + (288 * 200) = 86400
etc.

is this correct? not even considering calculating daily MA Length > 100 (eg 200 day MA) or MA weekly/monthly yet.

also if I have a daily chart that's running alongside my 5 minute chart, could the 5 minute chart reference the daily chart? if this was possible I could possibly create an equivalent CloseD series function?

previously from NT

You would create a script that runs on the other chart (the daily in this example) and then expose that data series - making sure it is synced to the primary series in the script and set to the desired value.

but not sure how to do this?

[koganam]

I realize this may be frustrating for you but please bear with me, this discussion has been very useful ...

I just want to be aware of approx. how many 5 minute bars would actually need to be downloaded

EURUSD 24 hours session has 24*60/5 = 288 5 minute bars

daily MA at session 0 would need session 1 to session 100 bars = 288 * 100 = 28,800
daily MA at session 1 would need session 2 to session 101 bars = 28,800 + (288 * 1) = 29088
daily MA at session 2 would need session 3 to session 102 bars = 28,800 + (288 * 2) = 29376
...
daily MA at session 100 would need session 101 to session 200 bars = 28,800 + (288 * 100) = 57600
daily MA at session 200 would need session 201 to session 300 bars = 28,800 + (288 * 200) = 86400
etc.

is this correct? not even considering calculating daily MA Length > 100 (eg 200 day MA) or MA weekly/monthly yet.

I do not understand what you are asking. Why would you have overlapping sessions? What information would require you to have 300 overlapping charts? I cannot answer, because I do not understand what you are trying to do.

also if I have a daily chart that's running alongside my 5 minute chart, could the 5 minute chart reference the daily chart? if this was possible I could possibly create an equivalent CloseD series function?

previously from NT but not sure how to do this?

You can use some undocumented methods to have an indicator on one chart read data from a different chart, but why would you? You would have introduced another slow resource into your data processing, and have mandated that that slow resource must exist before you can even process your data.

This same idea has been brought up before. Here is the thread. http://ninjatrader.com/support/forum...494#post427494

[futurenets]

ok yes just seen it, give me 5 minutes

[futurenets]

sorry about that mix up ...

daily MA at session 0 (day 0) would need 288 bars = 288
daily MA at session 1 (day 1) would need = 288 * 2 = 576
daily MA at session 2 (day 2) would need = 288 * 3 = 864
...
daily MA at session 100 (day 100) would need = 288 * 101 = 29,088
daily MA at session 200 (day 200) would need = 288 * 201 = 57,888
etc.

is this the correct number of 5 minute bars required to plot a daily MA?

[futurenets]

this is what I need

This same idea has been brought up before. Here is the thread. http://ninjatrader.com/support/forum...494#post427494

thanks for your help koganam

any sample code?

[futurenets]

if you have time koganam I'd like to know your thoughts on running a daily chart alongside a 5 minute chart with an array outside the namespace to hold previous 200 (or something) daily closing values. I understand this can be referenced by other charts.

Again thanks for all your help (and patience), really useful.

[koganam]

this is what I need



thanks for your help koganam

any sample code?

There is no sample code from me, because, as I clearly say, it is a silly idea to process data in a slow, constrained manner, just so that one can read back that data. I write code to process data as efficiently as I can, not code that runs in a slow, and constrained manner.

Other than for visual purposes of confirmation, I do not care one hoot about displaying data on a chart, and I would most certainly never put data on a chart for the sole purpose of reading it back. It is beyond my ability to comprehend why anyone wants to do so, but to each one's own.

[koganam]

if you have time koganam I'd like to know your thoughts on running a daily chart alongside a 5 minute chart with an array outside the namespace to hold previous 200 (or something) daily closing values. I understand this can be referenced by other charts.

Again thanks for all your help (and patience), really useful.

See my other response.

ref: http://ninjatrader.com/support/forum...d.php?p=451755

[koganam]

sorry about that mix up ...

daily MA at session 0 (day 0) would need 288 bars = 288
daily MA at session 1 (day 1) would need = 288 * 2 = 576
daily MA at session 2 (day 2) would need = 288 * 3 = 864
...
daily MA at session 100 (day 100) would need = 288 * 101 = 29,088
daily MA at session 200 (day 200) would need = 288 * 201 = 57,888
etc.

is this the correct number of 5 minute bars required to plot a daily MA?

Not really. No matter how you slice and dice it, you are going to need a minimum of 100 days loaded in order to calculate a correct value for the 100 day SMA. All your other calculations mean nothing. Once you hit that threshold, you will always require a previous number of bars equal to 100 days worth of data in order to calculate SMA(100): it is irrelevant if there are more than that many on the chart, though as a matter of fact, and as I have already said, NT, like most other charting packages uses a sliding window. That means that, generally speaking any data older than the days specified, will be seamlessly dropped from the chart, so incurring no further resource penalty.

[futurenets]

sorry I disagree,

inside or outside OOP protocol there are better ways than this. accessing a single value (in this case previous days closing price) on another chart once during each trading session is perfectly ok (albeit unsupported by NT).

As far as I'm concerned this is my equivalent of CloseD with very little effort (and much better than relying on 57,000 bars arriving without any problems).

I'll probably create a GLOBALS indicator until such time a formal solution is introduced by NT.

[koganam]

sorry I disagree,

inside or outside OOP protocol there are better ways than this. accessing a single value (in this case previous days closing price) on another chart once during each trading session is perfectly ok (albeit unsupported by NT).

As far as I'm concerned this is my equivalent of CloseD with very little effort (and much better than relying on 57,000 bars arriving without any problems).

I'll probably create a GLOBALS indicator until such time a formal solution is introduced by NT.

Your statement is not true, as an unreliable data stream on ANY time frame has exactly the same problem: unreliable results. If there are any errors in those 57,000 bars, they are also present in the daily bars. In any case, I have explained to you multiple times that at any data point only the last 28,800, 5-minute bars would be relevant to a 100 day calculation. It is beyond me why you persist in presenting faulty arithmetic. So whence the irrelevant data beyond 100 days?

However, to each his/her own. I shall now exeunt, as there is little, indeed nothing, that I can contribute further to this. You seem to think that this is a matter of opinion rather than analysis. I present analysis, with reasoning. You prefer to hand out an edict of your opinion. So mote it be.

All the best.

[futurenets]

your statement is incorrect, possibly over simplified (but understandable). thanks koganam for your input. my solution is working fine so no need to discuss further. find me and send your cv.

[futurenets]

for future reference:

How to make two charts share variable values

Why would I wish to do this? Lets say I have a Weekly chart with a 50EMA on it and I wish to know the current value to plot on my 100 tick chart. I do not wish of course to take the usual course of adding at least 50 weeks of data onto my 100 tick chart ...