My problem or eSignals - multi-interval series objects

[gspiker]

Hi,

I have reduced a problem I have been seeing to a minimal case...see the attached file.

The problem seems to be limited to multi-interval studies using series objects. In this case the called EFS is a lower timeframe than the calling EFS.

When run in RT, I get a slightly different answer than what I get when I reload the study using historical data.

The lines patterns look very similar in both cases, but there are minor differences in value. Due to the way I use this data in my strategy formula, this means I get different signals in RT than I do when I reload.

To reproduce, use the attached EFS and load in on a 60 minute ES #F chart and let it run for a many hours during market hours. Then load a new instance of the EFS and compare the values for the current day.

I'm not sure if I'm doing something wrong, or if this is related to
this problem or if there is something else wrong.

Notice that in this EFS I am multiplying two series values together in the called
EFS, but you can see the same issue (though not as obvious because the delta is even less) if you follow the instructions in the EFS on commenting out and uncommenting speicfic lines of code.

Thanks for any pointers or suggestions...

Garth

[gspiker]

Not to lose sight of the original question, but I wanted to verify that I understand things correctly.

Is it true that all OHLC data from an instance of eSignal comes from the same server?

Is it true that if I reload an EFS that calls a series using a lower interval, that the sub-interval OHLC data will be reloaded? Or will it use cached historical data for the sub-interval?

If it reloads the sub-interval data, then I should be able to verify if the actual data is changing for the called function in my previously posted EFS by running a 30 minute chart, doing a data export and then reloading the 30 minute chart.

Thanks!

Garth

[ACM]

Garth

Is it true that all OHLC data from an instance of eSignal comes from the same server?

In real time it comes from the same server, but on historical bars daily and intraday data come each from a different server.

Is it true that if I reload an EFS that calls a series using a lower interval, that the sub-interval OHLC data will be reloaded? Or will it use cached historical data for the sub-interval?

My understanding is that it uses the cached data.
Alex

Originally posted by gspiker
Not to lose sight of the original question, but I wanted to verify that I understand things correctly.

Is it true that all OHLC data from an instance of eSignal comes from the same server?

Is it true that if I reload an EFS that calls a series using a lower interval, that the sub-interval OHLC data will be reloaded? Or will it use cached historical data for the sub-interval?

If it reloads the sub-interval data, then I should be able to verify if the actual data is changing for the called function in my previously posted EFS by running a 30 minute chart, doing a data export and then reloading the 30 minute chart.

Thanks!

Garth

[gspiker]

Alex,

Thanks for the reply.
Now I'm even more confused about my original issue...


Garth

[eSignal_JasonK]

Hello Garth,

We're testing this with a 2-min vs 1-min scenario to get more data points to check against. We'll have more info for you soon.

[gspiker]

Thanks Jason.

I haven't tried it with that small an interval...and since the delta was so small using 60 minute bars I was worried that the issue may not appear at all when the bars interval (and therefore price delta per bar) was reduced.

But if you can repro the issue with that small an interval, that's even better!

Garth

[eSignal_JasonK]

Hello Garth,

Well, I didn't see any differences in the indicator data after loading a second instance of the study. They matched exactly for several hours of 2-min data. I'll try this again tomorrow with some higher interval combinations.



The row of zeros at the top is the total sum of the differences between the two instances of the study. The second instance was not loaded until the end of the day.

[gspiker]

Jason,

Thanks for looking into this. I ran this EFS for several days in a row, and when I loaded a second instance at the end of the day, it always had different values.

I'm going to try to do some more work on the EFS this weekend so that I capture all the values in a file. That should at least tell me where the problem lies..

It may be that for short intervals and only displaying two decimals that the issue can't be seen. I guess multiplying the retuns by 1000 may help if that is the case...

Garth

[eSignal_JasonK]

Hello Garth,

Again I didn't see any differences today (30 vs 15 minutes).



Are you saying that the differences you're seeing are less than .01 resolution? It is possible that you're seeing some differences due to floating point representation of numbers in javascript. I wasn't testing this on ES, however. That may be why I'm not seeing it. Also, (not sure this matters) do you have Compact Price display option enabled in Preferences?

I'll try this again with ES.

[gspiker]

Jason,

Thanks for looking at this again today.

In the case where I run the EFS as I sent it (with

nMySMIDiff = ( ( ( nClose * nOpen ) ) );

being uncommented, I would see lines that were very similar in form for both the instance of the EFS that ran in RT and the instance I would load at about 3:30pm to 4:00pm PST. However, the actual values were different, and not just after the decimal.

If I ran the script with the line that only looks at close, the results for some bars would match early in the day, but as the day progressed there were deltas...sometimes only after the decimal place.

The charts only display two digits after the decimal, so that is all the resolution I can comment on right now. The other was just speculation, based on the fact that the issue seems to "compound" the longer the study is run, and the fact that I have to multiple values together to see the issue on every bar in the day.

Like I said I will try to isolate this down, and if I can I will post the data here.

Garth

[gspiker]

Hi Jason,

Here is a snapshot of the RT results of my formula vs. one that was just loaded. Notice the different values. It will also show you what my time template looks like.

Here is a dump of values from one bar back on each NEWBAR. Notice that the values are different on the last set of bars. I will post another set on Tuesday - since the delta's are not always on the last few bars.

I'm not sure why the nOpen and nClose results seen to be one bar off between the RT and Historical results...that is odd.

RT:

Time:9:0 nOpen-1: 1214.25 nClose-1: 1213.5

Time:9:0 nEMASMIDiff: 1509249.342084505 nSMIDiffBBUpper: 1557428.8646918356 nSMIDiffBBLower: 1513285.329527036 nSMIDiffBBMiddle: 1536254.7173821882
Time:9:0 nEMASMIDiff-1: 1509249.342084505 nSMIDiffBBUpper-1: 1557428.8646918356 nSMIDiffBBLower-1: 1513285.329527036 nSMIDiffBBMiddle-1: 1536254.7173821882
------------------------------------------------------------------------------------------------------------------------------

Time:10:0 nOpen-1: 1213.5 nClose-1: 1214.25

Time:10:0 nEMASMIDiff: 1507518.322536819 nSMIDiffBBUpper: 1557869.7848922638 nSMIDiffBBLower: 1510626.596189503 nSMIDiffBBMiddle: 1535536.2992880817
Time:10:0 nEMASMIDiff-1: 1507518.322536819 nSMIDiffBBUpper-1: 1557869.7848922638 nSMIDiffBBLower-1: 1510626.596189503 nSMIDiffBBMiddle-1: 1535536.2992880817
------------------------------------------------------------------------------------------------------------------------------

Time:11:0 nOpen-1: 1218 nClose-1: 1226.5

Time:11:0 nEMASMIDiff: 1507947.7969632864 nSMIDiffBBUpper: 1558338.860681817 nSMIDiffBBLower: 1508188.7836278493 nSMIDiffBBMiddle: 1534783.8840766621
Time:11:0 nEMASMIDiff-1: 1507947.7969632864 nSMIDiffBBUpper-1: 1558338.860681817 nSMIDiffBBLower-1: 1508188.7836278493 nSMIDiffBBMiddle-1: 1534783.8840766621
------------------------------------------------------------------------------------------------------------------------------

Time:12:0 nOpen-1: 1235.25 nClose-1: 1225.75

Time:12:0 nEMASMIDiff: 1507266.5799939558 nSMIDiffBBUpper: 1558938.89813874 nSMIDiffBBLower: 1505792.8538338172 nSMIDiffBBMiddle: 1533965.3692722814
Time:12:0 nEMASMIDiff-1: 1507266.5799939558 nSMIDiffBBUpper-1: 1558938.89813874 nSMIDiffBBLower-1: 1505792.8538338172 nSMIDiffBBMiddle-1: 1533965.3692722814
------------------------------------------------------------------------------------------------------------------------------
Time:13:0 nOpen-1: 1220 nClose-1: 1227

Time:13:0 nEMASMIDiff: 1507219.8754228726 nSMIDiffBBUpper: 1559558.770080679 nSMIDiffBBLower: 1503536.0830083631 nSMIDiffBBMiddle: 1533099.423906302
Time:13:0 nEMASMIDiff-1: 1507219.8754228726 nSMIDiffBBUpper-1: 1559558.770080679 nSMIDiffBBLower-1: 1503536.0830083631 nSMIDiffBBMiddle-1: 1533099.423906302
------------------------------------------------------------------------------------------------------------------------------

Time:14:0 nOpen-1: 1227.75 nClose-1: 1224.5

Time:14:0 nEMASMIDiff: 1506440.6753987086 nSMIDiffBBUpper: 1560195.2998343129 nSMIDiffBBLower: 1501326.498471949 nSMIDiffBBMiddle: 1532166.6105024125
Time:14:0 nEMASMIDiff-1: 1506440.6753987086 nSMIDiffBBUpper-1: 1560195.2998343129 nSMIDiffBBLower-1: 1501326.498471949 nSMIDiffBBMiddle-1: 1532166.6105024125
------------------------------------------------------------------------------------------------------------------------------

Time:15:0 nOpen-1: 1219.5 nClose-1: 1220.5

Time:15:0 nEMASMIDiff: 1505043.6153759253 nSMIDiffBBUpper: 1560845.8960399255 nSMIDiffBBLower: 1499102.8451278757 nSMIDiffBBMiddle: 1531152.8437888294
Time:15:0 nEMASMIDiff-1: 1505043.6153759253 nSMIDiffBBUpper-1: 1560845.8960399255 nSMIDiffBBLower-1: 1499102.8451278757 nSMIDiffBBMiddle-1: 1531152.8437888294
------------------------------------------------------------------------------------------------------------------------------

Time:16:0 nOpen-1: 1214.25 nClose-1: 1204.5

Time:16:0 nEMASMIDiff: 1501618.2016401582 nSMIDiffBBUpper: 1561609.4834163748 nSMIDiffBBLower: 1496635.7769359776 nSMIDiffBBMiddle: 1529983.9142698038
Time:16:0 nEMASMIDiff-1: 1501618.2016401582 nSMIDiffBBUpper-1: 1561609.4834163748 nSMIDiffBBLower-1: 1496635.7769359776 nSMIDiffBBMiddle-1: 1529983.9142698038
------------------------------------------------------------------------------------------------------------------------------

Time:17:0 nOpen-1: 1200 nClose-1: 1196

Time:17:0 nEMASMIDiff: 1497515.3329750062 nSMIDiffBBUpper: 1562499.919518801 nSMIDiffBBLower: 1493875.3044194793 nSMIDiffBBMiddle: 1528645.6391266554
Time:17:0 nEMASMIDiff-1: 1497515.3329750062 nSMIDiffBBUpper-1: 1562499.919518801 nSMIDiffBBLower-1: 1493875.3044194793 nSMIDiffBBMiddle-1: 1528645.6391266554

Historical:

Time:9:0 nOpen-1: 1213.5 nClose-1: 1214.25

Time:9:0 nEMASMIDiff: 1509249.342084505 nSMIDiffBBUpper: 1557428.8646918356 nSMIDiffBBLower: 1513285.329527036 nSMIDiffBBMiddle: 1536254.7173821882
Time:9:0 nEMASMIDiff-1: 1509249.342084505 nSMIDiffBBUpper-1: 1557428.8646918356 nSMIDiffBBLower-1: 1513285.329527036 nSMIDiffBBMiddle-1: 1536254.7173821882
------------------------------------------------------------------------------------------------------------------------------

Time:10:0 nOpen-1: 1218 nClose-1: 1226.5

Time:10:0 nEMASMIDiff: 1507518.322536819 nSMIDiffBBUpper: 1557869.7848922638 nSMIDiffBBLower: 1510626.5961895015 nSMIDiffBBMiddle: 1535536.29928808
Time:10:0 nEMASMIDiff-1: 1507518.322536819 nSMIDiffBBUpper-1: 1557869.7848922638 nSMIDiffBBLower-1: 1510626.5961895015 nSMIDiffBBMiddle-1: 1535536.29928808
------------------------------------------------------------------------------------------------------------------------------

Time:11:0 nOpen-1: 1235.25 nClose-1: 1225.75

Time:11:0 nEMASMIDiff: 1507947.7969632864 nSMIDiffBBUpper: 1558338.8606818167 nSMIDiffBBLower: 1508188.7836278477 nSMIDiffBBMiddle: 1534783.8840766605
Time:11:0 nEMASMIDiff-1: 1507947.7969632864 nSMIDiffBBUpper-1: 1558338.8606818167 nSMIDiffBBLower-1: 1508188.7836278477 nSMIDiffBBMiddle-1: 1534783.8840766605
------------------------------------------------------------------------------------------------------------------------------

Time:12:0 nOpen-1: 1220 nClose-1: 1227

Time:12:0 nEMASMIDiff: 1507266.5799939558 nSMIDiffBBUpper: 1558938.89813874 nSMIDiffBBLower: 1505792.8538338155 nSMIDiffBBMiddle: 1533965.3692722798
Time:12:0 nEMASMIDiff-1: 1507266.5799939558 nSMIDiffBBUpper-1: 1558938.89813874 nSMIDiffBBLower-1: 1505792.8538338155 nSMIDiffBBMiddle-1: 1533965.3692722798
------------------------------------------------------------------------------------------------------------------------------

Time:13:0 nOpen-1: 1227.75 nClose-1: 1224.5

Time:13:0 nEMASMIDiff: 1507219.8754228726 nSMIDiffBBUpper: 1559558.770080679 nSMIDiffBBLower: 1503536.083008362 nSMIDiffBBMiddle: 1533099.423906301
Time:13:0 nEMASMIDiff-1: 1507219.8754228726 nSMIDiffBBUpper-1: 1559558.770080679 nSMIDiffBBLower-1: 1503536.083008362 nSMIDiffBBMiddle-1: 1533099.423906301
------------------------------------------------------------------------------------------------------------------------------

Time:14:0 nOpen-1: 1219.5 nClose-1: 1220.5

Time:14:0 nEMASMIDiff: 1506423.1825415655 nSMIDiffBBUpper: 1560196.4230796308 nSMIDiffBBLower: 1501324.2667470365 nSMIDiffBBMiddle: 1532166.096006613
Time:14:0 nEMASMIDiff-1: 1506423.1825415655 nSMIDiffBBUpper-1: 1560196.4230796308 nSMIDiffBBLower-1: 1501324.2667470365 nSMIDiffBBMiddle-1: 1532166.096006613
------------------------------------------------------------------------------------------------------------------------------

Time:15:0 nOpen-1: 1214.25 nClose-1: 1204.5

Time:15:0 nEMASMIDiff: 1505027.122110619 nSMIDiffBBUpper: 1560847.9765272094 nSMIDiffBBLower: 1499098.677169469 nSMIDiffBBMiddle: 1531151.8441969915
Time:15:0 nEMASMIDiff-1: 1505027.122110619 nSMIDiffBBUpper-1: 1560847.9765272094 nSMIDiffBBLower-1: 1499098.677169469 nSMIDiffBBMiddle-1: 1531151.8441969915
------------------------------------------------------------------------------------------------------------------------------

Time:16:0 nOpen-1: 1200 nClose-1: 1196

Time:16:0 nEMASMIDiff: 1501602.6508471551 nSMIDiffBBUpper: 1561612.4234567236 nSMIDiffBBLower: 1496629.8587113493 nSMIDiffBBMiddle: 1529982.457301701
Time:16:0 nEMASMIDiff-1: 1501602.6508471551 nSMIDiffBBUpper-1: 1561612.4234567236 nSMIDiffBBLower-1: 1496629.8587113493 nSMIDiffBBMiddle-1: 1529982.457301701
------------------------------------------------------------------------------------------------------------------------------

Time:17:0 nOpen-1: 1195.5 nClose-1: 1196.25

Time:17:0 nEMASMIDiff: 1497517.749370175 nSMIDiffBBUpper: 1562502.353719731 nSMIDiffBBLower: 1493870.118472171 nSMIDiffBBMiddle: 1528644.2532289987
Time:17:0 nEMASMIDiff-1: 1497517.749370175 nSMIDiffBBUpper-1: 1562502.353719731 nSMIDiffBBLower-1: 1493870.118472171 nSMIDiffBBMiddle-1: 1528644.2532289987
--------------------------------------------------
----------------------------------------------------------------------------
------------------------------------------------------------------------------------------------------------------------------

[eSignal_JasonK]

Hello Garth,

Are you using an EFS for this output? If so, would you mind posting it?

Thanks,

[gspiker]

Hi Jason,

Yes it is.

Quick update. I had noticed before the weekend that while I had 10.1 it was a version about 1 month older than what was up on the download page.

Unfortunately I forgot to download it prior to running my test on Monday. So I downloaded the latest Monday night and ran with that today.

Historical and RT lined up perfectly today. I'm not convinced the issue is fixed, but I will run the test script for a few more days and see what happens.

This was encouraging though...I have never seen it line up before and I have run it about a 12 days prior to this, and many more times with the real study (the one this is a very simplified version of).

I'll report back on Friday if all is well, and before that if it isn't ;-)

Sorry about not checking the release dates first...I wonder what changed between the two releases?

G

[eSignal_JasonK]

Hello Garth,

Thanks for the update. I should have asked for your build number. Anyway, there was a maintenance release for 10.1 shortly after the gold release to fix a couple bugs. There wasn't anything related to the problem we're investigating here, however. I'll look forward to your results.

[gspiker]

Jason,

Well, after two days of no issues (a first since I have noticed this) today the lines were different today between historical and real-time. The issue started at the 12:00 (noon) bar and continued through the rest of the day.

Questions:

1) Can this be related to a specific server?
2) Is there ever a case where a session could be bounced from one server to another without the user doing anything?

If it could be server related, I can start tracking the servers I'm connecting to.

I'll post the EFS I am using to print the data in my next message.

Thanks for any idea's on how to proceed with this.

Garth