Is this a known bug?

[joanNT]

Hi!

I've readed that you found a bug relating with "Unfortunately we are still experiencing issues with proper session detection based on various combinations of session definitions and time zones. This could result in incorrect time stamps of daily bars as well as inaccurate construction of a real-time daily bar."

And I was working with an strategy and found this, please see on the image attached.

The Timestamp in both data series are the same, but the prices, low, close... not. I must add that both data series have the same session.

It's only to confirm that this is what you're team are working on, or this is another issue.

Thanks

[NinjaTrader_Bertrand]

joanNT, thanks for reporting - the issue you cited is related to daily bar building.

I tried reproducing your behavior seen here on ZenFire with 5 sec and 10 min bars but was unable to unfortunately - do you only see this happening when accessing MultiSeries OHLC info from the finer series (i.e. the 5 sec in your combo)?

[joanNT]

Hi Bertrand,

Yes I only see this behavior accessing to historical data, I tried with AMP-zen fire and I get the same results.

I simply open a new chart with instrument ES 03-10 data between 11/12/2009 and 17/03/2010) and select a session I created (GMT+1)that goes:

Start Day Start Time End Day End Time
Monday 01/01/2000 Monday 01/01/2000 23:30
Tuesday 01/01/2000 Tuesday 01/01/2000 23:30
Wednesday 01/01/2000 Wednesday 01/01/2000 23:30
Thursday 01/01/2000 Thursday 01/01/2000 23:30
Friday 01/01/2000 Friday 01/01/2000 23:30

Then if you look for 23/02/2010 (16Hs --> GMT+1)

[NinjaTrader_Bertrand]

joanNT, thanks - which exact custom session were you using? In the one you posted, I don't see the starttime...is this only happening if you're using the custom session template you setup?

Thanks

[joanNT]

I use for start time 00:00 at every day

I tried with the instrument setting session (default) and the same differences, you can see them on the picture attached.

[NinjaTrader_Bertrand]

I see thanks - please try going down to a second chart to see clearly which tick should be included in the closing higher timeframe candle.

As NinjaTrader uses the closing bar time to 'stamp' it, the tick coming in at for example 16:00:00 would actually be in the new candle forming.

[joanNT]

Yes, I understand, but don't you think is confusing two bars with the same timestamp and different close price values? I mean, the 5 seconds bar has a close "lower" than the 10 minutes bar, both have the same timestamp, then there is something wrong.

Even, I've done what you said, and the right displayed bar is the 10 minutes one, cause the tick chart shows that the last tick close was at 15:59:59 and at 1106 price, while the 5 seconds bar says it's at 1102,5!!.

There is something wrong. I've attached the last tick picture.

Thanks

[NinjaTrader_Bertrand]

joanNT, if you compare all three charts (10 min, 5 sec and 1 tic) you see a flush of ticks with the same timestamp at exactly this period / candle (16:00 yr time, 8:00 AM MST). Your 5 sec bar is build from tick data and thus it uses the last tick received with 8:00:04 timestamp which has the price value of 1102.50 as per the HDM.

[joanNT]

Bertrand, I don't understand, please, let me explain with my own words

You are saying or I'm understanding:

On 10 minute Bar with TimeStamp 8:00:00 AM => Last Tick is from 7:59:59

On 5 seconds Bar with TimeStamp 8:00:00 AM => Last Tick is from 8:00:04

Is this true? Why?

Thank you

[NinjaTrader_Bertrand]

joanNT, difference might be related to using time data vs tick data to build the charts - upon further research into the issues you reported I found an inconsistency we're checking into. Thanks

[joanNT]

Ok, let me now when finished.

Thanks