BUG in v7.4r1 with price filtering?

[bfry5282]

I look only at INTC (Intel) 15s candlesticks. As such, I use advanced chart tools/price filtering of 0.1% in both fields, and when I click OK, it does indeed color the spikes with the "filtered color" as expected (I use white). However, as the chart continues to run along, I see spikes which are not colored with the "filtered color", and so I wonder whether price filtering is working... ? When I then change both fields from an existing 0.1 to a new more liberal 0.11 (just to make a change which is even "wider" than the previous filter), and click OK, then I see the filtered color as the chart updates, for the previously erroneously colored spikes which should have appeared filtered on the chart.

So, one of two things is going wrong here:

1) Either the chart is simply not coloring the candlestick spikes with the "filtered color" as price runs along continuously, or, much worse...

2) Price filtering is not working on an ongoing basis, as it should be.

I suspect that it's a charting problem only, and that underlying filtering is working, because my Parabolic SAR triggers malfunction when outlying spikes appear in price. I need tight underlying price filtering to avoid that happening, and INTC trades tightly with a penny spread most of the time.

Please look into this as part of the 7.5 release. I'll post a reply if I see evidence which of the two hypothesis is true. THANKS!

[bfry5282]

yes, this is a BUG imho

I now confirm that this is an apparent BUG, which affects the underlying formula price feed. The filter is correctly applied for existing historical chart data, but then on an ongoing continuous feed basis, it is not working properly. Thanks for looking into this for 7.5.

[bfry5282]

candlestick appearance changes

The candlestick appearance tells the tale. The charting itself is working properly, and the candlestick apearance changes drastically, when the data is refiltered, vs the ongoing feed when spikes appear. I didn't want to get into a discussion of my EFS studies, because that makes things much too complex. Looks like a real underlying filtering algorithm bug to me, judging from candlestick appearance only on the 15s charting. Again, when filtering is applied, the charting is correct. Then, as it continues along, spikes appear which are not colored correctly, and/or candlesticks are taller than they should be. When filtering is again applied, the historical chart appearance is corrected, but subsequent data is apparently not filtered correctly.

[fabrizio]

BUG in v7.4r1 with price filtering? bfry5282 eSignal 7.x 2 7 09-15-2003 08:01 PM
by bfry5282

Yes, it looks like a price filtering BUG bfry5282 eSignal 7.x 1 4 09-15-2003 07:22 PM
by bfry5282

Sub-minute interval chart CPU efficiency bfry5282 eSignal 7.x 10 166 09-09-2003 07:40 PM
by bfry5282

Excessive CPU consumption bfry5282 7 Series Beta 22 293 09-05-2003 07:56 PM
by bfry5282

High CPU usage fix! realm1 eSignal 7.x 6 119 09-05-2003 07:06 PM
by realm1

custom price filtering for efs studies? bfry5282 EFS Studies 0 14 09-05-2003 03:23 PM
by bfry5282

parabolic_sar.efs price study code?? bfry5282 EFS Studies 0 18 09-04-2003 07:25 PM
by bfry5282

compare notes? bfry5282 eSignal 7.x 0 20 08-29-2003 04:30 PM
by bfry5282

Tick replay calls efs only on interval completion? bfry5282 eSignal 7.x 1 26 08-21-2003 10:38 PM
by Alexis C. Montenegro

copy to clipboard? bfry5282 EFS Studies 0 14 08-02-2003 01:51 AM
by bfry5282

Tick Replay dialog box unusable bfry5282 7 Series Beta 2 60 07-31-2003 06:28 PM
by mattgundersen

Designated main(args) inhibit reload of study instance? bfry5282 7 Series Beta 0 13 07-23-2003 07:41 PM
by bfry5282

Limit 5 symbols per formula instance? bfry5282 7 Series Beta 10 70 07-16-2003 10:21 PM
by JayF

Excessive CPU consumption bfry5282 7 Series Beta 2 120 07-16-2003 09:26 PM
by bfry5282

Args to main() not seen by Edit Studies? bfry5282 7 Series Beta 4 27 07-16-2003 04:47 PM
by bfry5282

Build 606 prob w/null formula main params? bfry5282 7 Series Beta 10 95 07-10-2003 11:26 PM
by JasonK

TICK resolution charting bugs? bfry5282 EFS Studies 4 46 07-02-2003 06:18 PM
by bfry5282

Tick advanced chart study alignment

bfry5282

I beg your pardon, but sincerely I do not understand and maybe you can explain me.

Of all your posts in this community the 90% I found are criticizing something or stressing that something doesn't work or is wrong.

I am a simple customer and I questioning myself: should I throw out this platform of which I am - wrongly? - satisfied?

Or Is it a dream the one that I'm living?

I mean, Should I pinch myself and discover that I did not have profits because the good performance of ESIGNAL 7.4 but I'm actually broken and I'm already living on a carton box at the 57th at 6th green light?

Winter is coming...... who will feed my kids?

[JayF]

Re: yes, this is a BUG imho

Originally posted by bfry5282
I now confirm that this is an apparent BUG, which affects the underlying formula price feed. The filter is correctly applied for existing historical chart data, but then on an ongoing continuous feed basis, it is not working properly. Thanks for looking into this for 7.5.

Price Filtering only currently works on historical data received from the Tick and History Server. The reason for this is that the code was taken from Advanced Get's EOD software, where there really was no filtering code necessary for streaming quotes.

I pass this along as a suggestion for a future update to our software.

[bfry5282]

continuous price filtering please

OK, now I understand that it works only on the historical data, and is not an ongoing filter. That didn't appear to me to be documented anywhere, since the dialog box for price filtering has a box checked "enable", and I naturally assumed it was continuous underlying price filtering. (silly me)

For me, this highlights the need for a filter which works continuously, or a way for us EFS programmers to "hook" into the fundamental price feed in order to define custom filters on it, either globally or on a chart by chart basis, before price data is fed to studies.

I would appreciate it greatly if continuous price filtering could be implemented in the v7.5 release, as many backtesting algorithms or automated trading algorithms depend heavily upon having a "cleaned up" price feed, and I don't see how to clean it up right now, other than to use the price filtering feature currently partially implemented.

Thanks!

P.S. eSignal is the best platform available. fabrizio, I think every serious trader should use eSignal as the very best method of understanding the market, and we can all make a good living !!!! ;-) We operate under pressure, and we all want to contribute positive feedback to improve this great product !!

[JayF]

Thanks bfry for your suggestions. I'm unsure if this can be done in 7.5, however as we just released the beta yesterday, I suspect that it won't fit into the time frame for this release.

As far as tapping into the raw feed with EFS, this isn't really possible. The only way to do this currently, is with the eSignal Desktop API (Active X), and even then an entire application would have to be built (not a feasible option for > 99.99% of users.)

I have submitted the suggestion to change our price filtering to include streaming prices.