Charts changing when re-opened

[PKFFW]

Hi All,

I've had a wierd thing happen a few times and I'm hoping someone can explain it.

I use a 100 tick chart and I have an entry filter that uses a crossing of an EMA over the centre line of a Bolinger band. Basically a EMA cross really.

Now last night is a prime example. In real time at approximately 0911hrs the EMA crossed from above the bollinger band channel to below the midline of that channel. It then crossed back above the midline at approximately 0914hrs.

I closed down the charts to go to bed about 10 minutes later. Now when I opened the charts this morning(I'm in Australia by the way) everything looked the same.

Then I logged onto my Mirus account with the Zenfire feed and updated all the data to the present time. After doing so I went back over the charts and found the EMA appears to never have gotten closer than about 1 point to the bollinger band channel.

This problem has occurred quite a few times but never with such a huge difference.

Is it a problem with my feed or the charting or is it something I'm not understanding about the way the EMA and bollinger bands are calculated?

Any help greatly appreciated.

Cheers,
PKFFW
P.S: I've attached a picture of how it looked after I had updated the data but I didn't think to take a picture at the time in "real time"

[NinjaTrader_Ray]

This is very possible. Here is a link that may shed some light.

http://www.ninjatrader-support.com/HelpGuideV6/HowDoesNinjaTraderBuildChartBars.html

[PKFFW]

Hi Ray,

Firstly, thanks for the prompt reply.

Just going through the page you linked this is what I've come up with....

1: Can't have anything to do with different data feeds as I'm using only the one data feed. That is Zenfire from Mirus.

2: Can't be the real time tick filter as I don't have that function enabled

3: Therefore it must have something to do with the time stamping.

So a few things about the time stamping......

1: My PC clock seems to be set correctly for the time of AEST. Obviously my PC clock will be a different time to the Mirus Zenfire feed as that is coming from the US and my PC is in Australia. Not sure that should make any difference otherwise only people in the US could trade using the Zenfire feed. Or am I missing something there?

2: 50 ms delay there is nothing I can do about and will just have to live with it.

3: It seems to me that if it is an internet lag or time stamp problem then this thing about a tick being added to the next bar when in real time(because it came in late for example) and then being adjusted to be put into the proper bar when you reload the charts would be an intermitent and infrequent occurrance.

A tick coming in a micro second after NT has closed the bar when it should have come in a micro second before(due to lag or whatever) would seem to be something that would be very random and certainly this kind of overlap would not be occuring on every bar.

So that being the case, even if the occassional bar had this problem I can't see how it would cause my EMA to appear to cross in real time but then when the charts are reloaded the EMA is a full point away from the bollinger bands.

Occassionally the ema's have crossed in real time and then are a tiny bit away from crossing when the charts are reloaded. Other times price may have closed about a tick away from the bollinger band channel in real time but then when the charts are reloaded it looks like it is only half a tick away. These kind of discrepencies I can understand. But to have the EMA and bollinger bands being a whole point apart and not crossing until about 8 minutes and about 15 bars later is a bit much to be putting down to the occassional tick going into the wrong bar I would think.

Any other suggestions I can check on? Or am I missing something here and the occassional tick in the wrong bar could have this kind of seriously warping effect on the charts?

Cheers,
PKFFW

[NinjaTrader_JoshP]

In realtime indicators are constantly being calculated with the latest bar if you have CalculateOnBarClose = false. As the bar progresses the bar is changing and the value of the indicator is constantly changing. This is generally the difference between real-time and historical data. It bounces up and down and indicators will bounce up and down along with it.

[PKFFW]

In realtime indicators are constantly being calculated with the latest bar if you have CalculateOnBarClose = false. As the bar progresses the bar is changing and the value of the indicator is constantly changing. This is generally the difference between real-time and historical data. It bounces up and down and indicators will bounce up and down along with it.

Hi Josh,

I am aware of this issue however I don't think this is what is causing it.

In real time the EMA crossed through the BB channel from above to below and then about 2 minutes and 3 or 4 bars later it crossed from back again from below to above.

I then continued running NT for about another 10 minutes or so with many more bars printing off. These two crosses of the EMA and BB channel remained and were clearly visible.

Then when I started NT up again and logged into my data feed, thus updating the data from when I closed NT down to the present time, that is when the discrepancy occurred. That is when the EMA reprinted and never came closer than 4 ticks from the BB channel until 8 minutes later and about 15 bars later.

Cheers,
PKFFW

[PKFFW]

Ray,

After replying to Josh another thought occurred to me. I just wanted to make sure I was understanding this time stamping problem thing correctly.

In real time, NT uses my computer clock for printing off bars and the data feed suppliers use their own clock for time stamping.

This can lead to the problem of a tick that got stamped by the data feed supplier as say 0900hrs and then due to lag arrived at my computer at 0901hrs would be printed in real time in the bar that occurred at 0901hrs.

Then upon reloading the historical data at a later time, NT would use the time stamp supplied by the data feed supplier and so this tick would then be placed in the bar that occurred at 0900hrs rather than the bar it was originally placed in.

So, would not this problem of the EMA being printed differently to how it printed in real time occur as soon as I start up NT again and reload the chart?

Have I understood the time stamping thing correctly?

In this case, when I load up NT and reload the chart it prints exactly like it did in real time. It is not until I reconnect to the data feed and get all the data from when I turned of NT to the present time that the EMA is then printed differently.

Hope this clarifies the issue and helps to narrow down what might be causing it.

Cheers,
PKFFW

[NinjaTrader_JoshP]

Regardless of how the timestamps are if your bars are the same your indicators should produce the same results. The only possible reason why your indicator would have a different result on historical bars are if your bars have changed. (e.g. having different data when you redownloaded it).

[PKFFW]

Regardless of how the timestamps are if your bars are the same your indicators should produce the same results. The only possible reason why your indicator would have a different result on historical bars are if your bars have changed. (e.g. having different data when you redownloaded it).

This is what I would have thought. Hence my query.

I have not changed data feeds so I'm wondering why the bars should be any different when I open them up and connect to the data feed again.

Any further suggestions from someone in the know would be greatly appreciated.

Cheers,
PKFFW

[NinjaTrader_JoshP]

Only thing that comes to mind is if when you reconnected you ended up redownloading data. The redownloaded data might be different depending on data source. Some providers may have a different feed for historical data vs real-time data.