Does NinjaTrader drop ticks real-time ?

[NinjaTrader_Bertrand]

Guys, please be assured there's no finger pointing involved - we will test Gomi's sample next week on live data and then post our conclusions here.

Thanks,

[playafh69]

Thank you..

[timmyb]

THANKS very happy

Guys, please be assured there's no finger pointing involved - we will test Gomi's sample next week on live data and then post our conclusions here.

Thanks,

[NinjaTrader_Bertrand]

Hi All: today I tested the sample indicator discussed here on the ES 12-10 1 tick chart with a session ranging from open (9:30 EST) to 15 mins later (9:45 EST). I then compared the ticks and timestamps received from the SteamWriter text file from our OnMarketDate() with the ticks recorded in the NT database....summed all relevant ones for this 15 min session (9:30:00 to 9:44:59) and in both cases I got identical results to the tick : 174921 ticks.

[GoldStandard]

Hi All: today I tested the sample indicator discussed here on the ES 12-10 1 tick chart with a session ranging from open (9:30 EST) to 15 mins later (9:45 EST). I then compared the ticks and timestamps received from the SteamWriter text file from our OnMarketDate() with the ticks recorded in the NT database....summed all relevant ones for this 15 min session (9:30:00 to 9:44:59) and in both cases I got identical results to the tick : 174921 ticks.

Finding 15 minutes without errors doesn't prove much either way does it?

[NinjaTrader_Bertrand]

It clearly shows that no ticks are being lost or dropped by NT, they are all processed. Gomi's original requst was for 5 min, I compared about 39K ticks or 15 min on a very liquid futures contract and found no anomaly.

[NinjaTrader_Bertrand]

For more clarification - here's the exact route I took to reproduce and look into -

1. Setup custom session template from 9:30 - 9:45 EST
2. Under Tools > Options > Data used 'Save chart data as historical' and unchecked 'Get data from server'
3. Connect to Kinetick only
4. Open up 1 tick chart with custom session as defined in 1. and 0 bars lookback.
5. Apply Gomi's sample OnMarketData() indicator
6. After the 15 min was finished, I imported both the custom textfile from the indicator and Historical Data Manager tick export to Excel and summed the ticks received from 9:30:00 to 9:44:59, taking into consideration that the indicator would not be called for the first 20 bars, the unstable period.

In both tick recordings I arrived at the exact same summation value.

Please let me know if you were doing something different then my above steps or if you would see different results following them exactly as outlined.

Thanks,

[Acqua]

It clearly shows that no ticks are being lost or dropped by NT, they are all processed. Gomi's original requst was for 5 min, I compared about 39K ticks or 15 min on a very liquid futures contract and found no anomaly.

Could you run the test another couple of times and compare the results? It's rather awkward that on your very first test not even 1 tick was dropped whereas Gomi's and other people testing there are quite a few ticks dropped.
Just to make sure your finding is not a one-time event and also to have more data to support your side of the argument.

[gomifromparis]

Here's my shot at yesterday :
At least 1 tick dropped (chart 1)
Lots of timestamp differences, so why does the e.Time in OnMarketData is different that timestamp stored in historical manager ?
Will upgrade results for today, but I'm waiting for the ticks to be written to the database.

[NinjaTrader_Ray]

gomifromparis,

We for sure want to investigate and are doing everything we can to understand what may be happening. Bert and our development team is at the ready to work through this. What I suggest right now is that we all get to a controlled test environment so both sides are looking at the exact same thing. We can then tweak things step by step together in hopes that we finally are able to see what you see.

If you agree, we will then provide exact set up steps along with a data account that you can use.

[gomifromparis]

Thanks Ray,

For one thing I don't have

"2. Under Tools > Options > Data used 'Save chart data as historical' and unchecked 'Get data from server'"

But the opposite. I'll change them to yours and see what happens. It might explain a lot of discrepancies.

[NinjaTrader_Bertrand]

Ok Gomi, please keep us posted on your results seen with the settings used instead then.

Thanks

[gomifromparis]

Here's what I got today :
in real-time : everything OK.
I press F5 multiple times => no problem.
then I disconnect and reconnect esignal to force historical download => 2500 contracts lost...

I guess I have solved the problem : I have 4 files : the files from the recording indicator I sent, the file form my gomrecorder, the file from the ninja export and the file from qcollector that downloaded esignal data.

The files from the recording indicator and the file from GomRecorder match
Files from ninja export and from qcollector match.

So basically, what happens is that esignal has sh*t on its historical servers, and they don't hold every tick. So real-time everything is OK, but as soon as historical data is downloaded , discrepancies appear.

Sorry for the inconvenience, I didn't realize you had to click somewhere to have your real time data actually saved.

[NinjaTrader_Bertrand]

Thanks for posting your testing results on the changed settings and great to hear the mystery is solved. The setting to not save realtime data is offered for improving performance when you work with a connection that would support a backfill.

[playafh69]

Thank you for getting to the bottom of this. Im glad that everything is working fine.