HIGH DEFINITION CHARTING ...BUGS OR WHAT???

Questions about MultiCharts and user contributed studies.
Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

HIGH DEFINITION CHARTING ...BUGS OR WHAT???

Postby Tresor » 25 Jan 2009

All below presented cases were made with MC Version 5.0 Beta 2 (Build 1781) and TransAct data feed.
Last edited by Tresor on 25 Jan 2009, edited 1 time in total.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Volume bars

Postby Tresor » 25 Jan 2009

My MC fails to plot Volume Bars correctly. It fails to split the data into equal volume bars. Please see the screenshot attached.

Regards
Attachments
volume bars.jpg
(83.49 KiB) Downloaded 5873 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Point Original bars

Postby Tresor » 25 Jan 2009

MY MC fails to plot point original bars correctly. Please see the screenshot.
Attachments
point original.jpg
(62.84 KiB) Downloaded 5832 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Is it really a regular session?

Postby Tresor » 25 Jan 2009

The attached screenshot explains better than words. I set NQ as ''CME: Equity Index Futures [e-mini] (Globex)'' in QM + I set sessions to Regular in MC yet MC still plots non-regular session bars. Really confusing.
Attachments
not really regular.jpg
(121.62 KiB) Downloaded 5824 times

brodnicki steven
Posts: 407
Joined: 01 Jan 2008
Been thanked: 3 times

Re: Is it really a regular session?

Postby brodnicki steven » 25 Jan 2009

The attached screenshot explains better than words. I set NQ as ''CME: Equity Index Futures [e-mini] (Globex)'' in QM + I set sessions to Regular in MC yet MC still plots non-regular session bars. Really confusing.
* In QM did you set "custom" hours for the regular session ? (you MUST do that)

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 25 Jan 2009

My MC seems to plot candles badly with Transact data feed for all available resolutions (except for time resolutions). The attached screenshot explains it better.

Regards
Attachments
MC resolutions.jpg
(165.79 KiB) Downloaded 5810 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Re: Is it really a regular session?

Postby Tresor » 25 Jan 2009

The attached screenshot explains better than words. I set NQ as ''CME: Equity Index Futures [e-mini] (Globex)'' in QM + I set sessions to Regular in MC yet MC still plots non-regular session bars. Really confusing.
* In QM did you set "custom" hours for the regular session ? (you MUST do that)
Hi Steven,

In QM I set the symbol according to the custom template for Globex availabe in QM. Please see the screenshot.

Regards
Attachments
NQ QM settings.jpg
(74.32 KiB) Downloaded 5819 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Vanishing daily bars

Postby Tresor » 25 Jan 2009

The other issue I have with MC charting are vanishing daily bars that I documented with screenshots in this thread: http://forum.tssupport.com/viewtopic.php?t=5911

Regards

brodnicki steven
Posts: 407
Joined: 01 Jan 2008
Been thanked: 3 times

Re: Is it really a regular session?

Postby brodnicki steven » 25 Jan 2009

The attached screenshot explains better than words. I set NQ as ''CME: Equity Index Futures [e-mini] (Globex)'' in QM + I set sessions to Regular in MC yet MC still plots non-regular session bars. Really confusing.
* In QM did you set "custom" hours for the regular session ? (you MUST do that)
Hi Steven,

In QM I set the symbol according to the custom template for Globex availabe in QM. Please see the screenshot.

Regards
** I DON'T use pre-made templates like that, set you own days and times, then it will work. (does on mine)

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Re: Volume bars

Postby Marina Pashkova » 26 Jan 2009

My MC fails to plot Volume Bars correctly. It fails to split the data into equal volume bars. Please see the screenshot attached.

Regards
Hi Tresor,

If you look closely at your screenshot, you will notice that the subchart with the volume histogram has a red line at the bottom that exactly equals the hight of the 100 volume bars. That line prevents us from seeing if there are any bars covered by it but they are probably there. In this case the problem that you are describing is caused by the fact that some ticks resulted in volume spikes. MultiCharts has the following mechanism of plotting volume bars:

If a tick arrives whose volume is too large for the chosen resolution, this volume is split between a number of bars. However, to prevent memory consumption in the case of huge volumes, the maximum number of splits is 5. Whatever is left after such splits, is 'dumped' into the last bar in a group. In other words, you will have 4 bars of equal hight corresponding your chosen volume resolution and a bigger or smaller (depending on whatever is left after the devisions) fifth bar. This is exactly what could be seen in your sreenshot but for that red line.

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Re: Point Original bars

Postby Marina Pashkova » 26 Jan 2009

MY MC fails to plot point original bars correctly. Please see the screenshot.
Hi Tresor,

Point (original) is a rudiment of earlier stages of MC development. It was the very first implementation of point bars which subsequently turned out to be incorrect. It was removed from the list of available resolutions but then was returned for some of the users wanted to have this particular resolution as opposed to the new (and correct) point bars.

We will remove this option from the list of available resolutions. Meanwhile, I would recommend avoiding plotting point (original) charts.

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Postby Marina Pashkova » 26 Jan 2009

My MC seems to plot candles badly with Transact data feed for all available resolutions (except for time resolutions). The attached screenshot explains it better.

Regards
The logic of plotting change bars is the same as that of volume bars (and point bars). See one of my posts above.

Also, the accuracy of such kinds of charts depends on price scale and min movement set for the symbol in question. In QuoteManager -> Edit Symbol -> Settings please check if you are using the correct settings. If they are wrong, enable Use Custom Settings and enter the correct values.

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Re: Is it really a regular session?

Postby Marina Pashkova » 26 Jan 2009

The attached screenshot explains better than words. I set NQ as ''CME: Equity Index Futures [e-mini] (Globex)'' in QM + I set sessions to Regular in MC yet MC still plots non-regular session bars. Really confusing.
Steven is right. To have adequate regular sessions, you need to first specify them in QuoteManager -> Edit Symbol -> Sessions -> Use Custom Sessions.

Once the changes have been made, reload or re-create the chart.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Re: Is it really a regular session?

Postby Tresor » 26 Jan 2009


** I DON'T use pre-made templates like that, set you own days and times, then it will work. (does on mine)
Neither DID I (I always used my custom session settings) untill a few days ago when I tried to experiment with MC on a wider scale. Thanks for the advice. I will quit using the templates as they are useless.

Regards

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Re: Is it really a regular session?

Postby Tresor » 26 Jan 2009


Steven is right. To have adequate regular sessions, you need to first specify them in QuoteManager -> Edit Symbol -> Sessions -> Use Custom Sessions.

Once the changes have been made, reload or re-create the chart.
Marina,

The issue of trading hours has been a pain in the ass for many of MC users (search forum for this). People ask: should I set 24h in QM or rather something else in QM to get what I want on the chart in MC? Etc.

Wouldn't it be a much elegant solution to allow QM intake all 24/7 data and then the user would only need to manouvre the hours one wants to see on their charts through MC - Format Symbol?

Regards

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Re: Volume bars

Postby Tresor » 26 Jan 2009

Hi Tresor,

If you look closely at your screenshot, you will notice that the subchart with the volume histogram has a red line at the bottom that exactly equals the hight of the 100 volume bars.
Marina, I did have a look a the screenshot and what I noticed was:
(i) a majority of bars comply to the crtiterion of being completed at 100 contracts; and
(ii) a minrity of bars fails to comply to the criterion of being completed at 100 contracts (MC complets some bars at 300+ contracts and some bars at 70-) - this is not acceptable for someone who assumes all bars will be equal in terms of volume generated.
MultiCharts has the following mechanism of plotting volume bars:

If a tick arrives whose volume is too large for the chosen resolution, this volume is split between a number of bars. However, to prevent memory consumption in the case of huge volumes, the maximum number of splits is 5.
You explained the reason for inroducing the mechanism that is responsible for not plotting the volume bars correctly under certain conditions, namely to prevent the memory consumption. That was okay a year or two years ago. It has no logical justification today. It only confuses.

Personal computers now-a-days are more powerfull than industrial computers at the time TSS was introducing this mechanism into MC. There is no need to keep this mechanism alive any more, even if this means that during high volume spikes I will receive same looking 10 / 20 / whatever horizontal bars in a row.

As I mentioned MC also plots volume bars that are not only over-complete, but are also under-complete (not the ones at the end of day), i.e. when you set resolution of 100 contracts, MC plots some bars of less than 100 contracts during the day. This should also be dealt with.

Regards

PS. This is how computing power of CPUs changed over the last 2 years: http://www.cpubenchmark.net/high_end_cpus.html
Last edited by Tresor on 26 Jan 2009, edited 1 time in total.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Re: Point Original bars

Postby Tresor » 26 Jan 2009

Hi Tresor,

Point (original) is a rudiment of earlier stages of MC development. It was the very first implementation of point bars which subsequently turned out to be incorrect. It was removed from the list of available resolutions but then was returned for some of the users wanted to have this particular resolution as opposed to the new (and correct) point bars.

We will remove this option from the list of available resolutions. Meanwhile, I would recommend avoiding plotting point (original) charts.
Sure, get rid of it if it can't work properly.

Regards

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Re: Volume bars

Postby Marina Pashkova » 27 Jan 2009

Hi Tresor,

If you look closely at your screenshot, you will notice that the subchart with the volume histogram has a red line at the bottom that exactly equals the hight of the 100 volume bars.
Marina, I did have a look a the screenshot and what I noticed was:
(i) a majority of bars comply to the crtiterion of being completed at 100 contracts; and
(ii) a minrity of bars fails to comply to the criterion of being completed at 100 contracts (MC complets some bars at 300+ contracts and some bars at 70-) - this is not acceptable for someone who assumes all bars will be equal in terms of volume generated.
MultiCharts has the following mechanism of plotting volume bars:

If a tick arrives whose volume is too large for the chosen resolution, this volume is split between a number of bars. However, to prevent memory consumption in the case of huge volumes, the maximum number of splits is 5.
You explained the reason for inroducing the mechanism that is responsible for not plotting the volume bars correctly under certain conditions, namely to prevent the memory consumption. That was okay a year or two years ago. It has no logical justification today. It only confuses.

Personal computers now-a-days are more powerfull than industrial computers at the time TSS was introducing this mechanism into MC. There is no need to keep this mechanism alive any more, even if this means that during high volume spikes I will receive same looking 10 / 20 / whatever horizontal bars in a row.

As I mentioned MC also plots volume bars that are not only over-complete, but are also under-complete (not the ones at the end of day), i.e. when you set resolution of 100 contracts, MC plots some bars of less than 100 contracts during the day. This should also be dealt with.

Regards

PS. This is how computing power of CPUs changed over the last 2 years: http://www.cpubenchmark.net/high_end_cpus.html
Hi Tresor,

This reasoning is not actually correct. If you request an adequate volume resolution for a symbol, occasional spikes in volume won't harm your bars. However, some users choose resolutions that have nothing to do with volumes that are actually traded for a particular symbol. This actually happens quite a lot. It is not uncommon for somebody to request a 2 volume chart for MCFX which is traded in thousands of contracts. As you understand, trying to split such volumes into 2 volume bars will generate enormous number of bars. Given that each process under x32 bit systems cannot use more than 2 GB of memory, there is no way a computer can cope with those volumes. Hence the limitation on breaking up bars.

P.S. Smaller bars are a result of the same splits that result in large bars. For example, after all the splits you have not a 300 contract leftover, but a 5 contract leftover. It will be plotted separately.

P.P.S. I hope the above explanation makes sense.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Re: Volume bars

Postby Tresor » 27 Jan 2009

P.P.S. I hope the above explanation makes sense.
All I undestood is that MC will continue plotting volume bars incorrectly.

Instead of 20 bars of 100 contracts MC will, under heavy volume spikes (i.e. when the correct price action plotting is especially needed) MC will plot e.g. 5 bars of 100 contracts, then it will plot 1 bar of 400 contracts, then it will plot again 5 bars of 100 contracts, and then it can very well plot a bar of 700 contracts, etc.

This makes no sense for someone who wants to see all bars have the same resolution. This makes no sense to backtest / optimize strategies based on volume resolution and subsequently it makes no sense to trade strategies on volume resolutions, unless one wants to get broke.

Regards

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Re: Volume bars

Postby Tresor » 27 Jan 2009

This reasoning is not actually correct. If you request an adequate volume resolution for a symbol, occasional spikes in volume won't harm your bars. However, some users choose resolutions that have nothing to do with volumes that are actually traded for a particular symbol. This actually happens quite a lot. It is not uncommon for somebody to request a 2 volume chart for MCFX which is traded in thousands of contracts. As you understand, trying to split such volumes into 2 volume bars will generate enormous number of bars.
Marina,

Please let people decide if they want to sacrifice the software performance at the expense of having bars plotted properly.

Have a look at the attached screenshot. As a result of your mechanism, MC is unable to plot 1,000 volume bars for ES.

Please stop taking care of idiots who wants to have small volume bars and please do not worry about their computers. Please take care of those of your clients who want to have bars plotted in a reliable way, especially during volume spikes. Such spikes will have no adverse effect on the performance of MC on our machines.

My RAM usage NEVER is higher than 1 GB with numerous charts opened and numerous other applications and proccesses running.

Please get rid of this mechanism.

Regards
Attachments
1000 contract resolution.jpg
(116.79 KiB) Downloaded 5816 times

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Postby Marina Pashkova » 28 Jan 2009

Hi Tresor,

The volume spikes in your screenshot look abnormal. The bar with the volume of 2,781 is huge and it could have resulted only after the original tick was split into 5 bars: 4 having the volume of 1,000 and 1 having the volume of 2,781.

I checked tick data in TS for the whole of 01/26/2009 and the largest spick there was slightly over 1,000. In other words, volume data provided by TransAct probably contains abnormal values.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 28 Jan 2009

Hi Tresor,

The volume spikes in your screenshot look abnormal. The bar with the volume of 2,781 is huge and it could have resulted only after the original tick was split into 5 bars: 4 having the volume of 1,000 and 1 having the volume of 2,781.

I checked tick data in TS for the whole of 01/26/2009 and the largest spick there was slightly over 1,000. In other words, volume data provided by TransAct probably contains abnormal values.
Hi Marina,

TS is widely known for bad quality (filtered) data, while Transact provides quality unfiltered data.

Rafal

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Postby Marina Pashkova » 28 Jan 2009

Hi Tresor,

We ran another check using eSignal's data. The results were consistent what TS showed.

As for the logic of plotting volume bars, it has been proven correct many times in the past and for now we are not planning to change it.

Regards.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 28 Jan 2009

Hi Tresor,

We ran another check using eSignal's data. The results were consistent what TS showed.
Marina, eSignal is the same filtered crap like TS. You know it for sure from this and many other forums. Please check it with high quality unfiltered data feed, like Transact.

As for the logic of plotting volume bars, it has been proven correct many times in the past and for now we are not planning to change it.
Marina, the best scientific method of arriving to any conclusion is by trying to falsificate the thesis. This means that one should not look for arguments that support one's thesis, like ''it has been proven correct many times in the past'' (because there may be thousands of favouring evidences); instead one should focus on finding at least one argument that can falsificte / nagate one's thesis (which is cheaper and less time consuming than looking for aguments that support the thesis).

The screenshot provided in one of the earlier posts clearly shows that the logic of plotting volume bars is incorrect. WHY? BECAUSE IT SHOWS THAT MC FAILS TO PLOT BARS OF EQUAL VOLUME! One simple screenshot has thrown all other evidences that you were gathering all those years.

Regards

RWDickinson
Posts: 43
Joined: 01 Dec 2008
Has thanked: 2 times
Been thanked: 2 times

Postby RWDickinson » 28 Jan 2009

I have no information about Transact, but I do know from painful experience that "unfiltered" data from the exchanges have many error ticks in them. "Filtering" (the thing you despise) is what corrects those errors. I don't know where you got the idea that "unfiltered" is higher quality than "filtered", but you've got it just backwards. eSignal and TS provide much higher quality data than the unfiltered vendors, precisely because of their filtering.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 28 Jan 2009

I have no information about Transact, but I do know from painful experience that "unfiltered" data from the exchanges have many error ticks in them. "Filtering" (the thing you despise) is what corrects those errors. I don't know where you got the idea that "unfiltered" is higher quality than "filtered", but you've got it just backwards. eSignal and TS provide much higher quality data than the unfiltered vendors, precisely because of their filtering.
I do not want to argue with you on what quality means with regard to data feed. The fact is Marina used other than Transact data feed to check if MC plots bars correctly. Obviously, she could not have arrive at the same conclusion as mine, as she did not use the same data.

Had she used the same data, the discussion would have a slight sense.

Regards

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 29 Jan 2009

Another issue:

- there are differences in price levels between daily HHs and LLs when comparing them to hourly HHs and hourly LLs. Both daily and hourly charts' sessions time were set to regular in MC.

- in this screenshot I also documented vanishing daily bars (2 bars are missing during one week).

Regards
Attachments
diff prices + vanishing daily bars.jpg
(115.99 KiB) Downloaded 5809 times

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Postby Marina Pashkova » 30 Jan 2009

- there are differences in price levels between daily HHs and LLs when comparing them to hourly HHs and hourly LLs. Both daily and hourly charts' sessions time were set to regular in MC.
We have not been able to reproduce the above problem. Try reloading charts. If that does not help, please contact us either via LiveChat or by phone for a remote connection session.
- in this screenshot I also documented vanishing daily bars (2 bars are missing during one week).
There is a bug on providers who, like TransAct do not provide daily bars. Will be fixed in future MC versions.

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Postby Marina Pashkova » 24 Feb 2009

Hi Tresor,

We ran another check using eSignal's data. The results were consistent what TS showed.
Marina, eSignal is the same filtered crap like TS. You know it for sure from this and many other forums. Please check it with high quality unfiltered data feed, like Transact.

As for the logic of plotting volume bars, it has been proven correct many times in the past and for now we are not planning to change it.
Marina, the best scientific method of arriving to any conclusion is by trying to falsificate the thesis. This means that one should not look for arguments that support one's thesis, like ''it has been proven correct many times in the past'' (because there may be thousands of favouring evidences); instead one should focus on finding at least one argument that can falsificte / nagate one's thesis (which is cheaper and less time consuming than looking for aguments that support the thesis).

The screenshot provided in one of the earlier posts clearly shows that the logic of plotting volume bars is incorrect. WHY? BECAUSE IT SHOWS THAT MC FAILS TO PLOT BARS OF EQUAL VOLUME! One simple screenshot has thrown all other evidences that you were gathering all those years.

Regards
Tresor,

As was explained on a number of occasions in the past, THE ALGORITHM OF PLOTTING VOUME BARS IN MC IS PERFECTLY CORRECT.

Your screenshot showed 100 volume bars with a volume indicator applied in a subchart. That indicator showed several spikes. Spikes are the result of the following:

MultiCharts plots volume bars by dividing volume into equal portions, in your case, 100 contracts. Now, if a tick arrives whose volume is more than 100 contracts it is divided into equal 100 contract portions. But the number of possible divisions is limited to 5. Thus if a tick arrives whose volume is, say, 800 contracts, you will get four 100-contract bars and one 400-contract bar. This mechanism has been implemented as a protection against cases when users try plotting volumes way smaller than normal volume values for a symbol. If the number of divisions were not limited, your memory would have been eaten up by profusely multiplying bars. You can only imagine what would happen if somebody tries plotting 1-contract bars for a symbol that is traded in thousands of contracts – which isn’t out of the realm of possibility.

The spikes in your screenshot that appear against the background of numerous perfectly equal 100-contract bars are a result of huge volume values transmitted by TransAct. MultiCharts divided the volume received in a tick into four 100-contract bars and dumped the rest of the volume into the 5 bar. As I mentioned earlier, neither TS nor eSignal have revealed any similar spikes over the period of time in question. You are saying that the quality of data by TS or eSignal is inferior to that of TransAct. I would not want to comment on that in this particular thread.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 24 Feb 2009

Tresor,

As was explained on a number of occasions in the past, THE ALGORITHM OF PLOTTING VOUME BARS IN MC IS PERFECTLY CORRECT.

Your screenshot showed 100 volume bars with a volume indicator applied in a subchart. That indicator showed several spikes.
Marina,

My screenshot is showing 1,000 contract volume bars, not 100 volume bars! Please see the screenshot once again - the status bar.

Your algorithm may work well with filtered data feeds, but fails with unfiltered data feeds.

Regards.
Attachments
1000 contract resolution.jpg
(40.11 KiB) Downloaded 5815 times

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Postby TJ » 24 Feb 2009

...But the number of possible divisions is limited to 5...
I think it is reasonable to have a limit.
But 5 might be too small a number...

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Postby Marina Pashkova » 24 Feb 2009

Tresor,

As was explained on a number of occasions in the past, THE ALGORITHM OF PLOTTING VOUME BARS IN MC IS PERFECTLY CORRECT.

Your screenshot showed 100 volume bars with a volume indicator applied in a subchart. That indicator showed several spikes.
Marina,

My screenshot is showing 1,000 contract volume bars, not 100 volume bars! Please see the screenshot once again - the status bar.

Your algorithm may work well with filtered data feeds, but fails with unfiltered data feeds.

Regards.
Tresor,

My previous comments were based on this screenshot http://forum.tssupport.com/download.php?id=1902

Therefore I used exactly the same screenshot to make my point again.

If you want I can do the same once again for your other screenshot again.

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Postby Marina Pashkova » 24 Feb 2009

Tresor,

As was explained on a number of occasions in the past, THE ALGORITHM OF PLOTTING VOUME BARS IN MC IS PERFECTLY CORRECT.

Your screenshot showed 100 volume bars with a volume indicator applied in a subchart. That indicator showed several spikes.
Marina,

My screenshot is showing 1,000 contract volume bars, not 100 volume bars! Please see the screenshot once again - the status bar.

Your algorithm may work well with filtered data feeds, but fails with unfiltered data feeds.

Regards.
Tresor,

I was referring to http://forum.tssupport.com/download.php?id=1902
Looks like 100 volume bars to me.

P.S. I have explained how the algorithm of plotting volume bars works as well as under what conditions you can get spikes. As for the datafeeds, the program doesn't care if the data is filtered or not. We are about to release a new MC version with the new TransAct API implemented. Let us see if there are any spikes there. If there are no more spikes, it would mean that the problem was in TransAct's old API. If spikes are still there, we'll start investigating those furhter.

P.P.S. As of now, we do not have plans to change the logic of creating volume bars.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 24 Feb 2009

Tresor,

I was referring to http://forum.tssupport.com/download.php?id=1902
Looks like 100 volume bars to me.
The chart you were refering to shows that MC, due to the existing logic of plotting volume bars, is unable to create volume bars of 100 contracts.

And I was referring to this chart http://forum.tssupport.com/download.php?id=1920 in which it was shown that MC, due to the existing logic of plotting volume bars, is unable to achieve equal volume bars of 1,000 contracts in ES.

P.S. I have explained how the algorithm of plotting volume bars works as well as under what conditions you can get spikes.
Marina, your explanations were clear enough for a computer layman as myself to understand what algorithm you use when plotting volume bars. I do understand the logic of this algorithm very well and ... I do not like it.
P.P.S. As of now, we do not have plans to change the logic of creating volume bars.
Sure, I am the only person in this forum who does not like this logic (for reasons explained earlier). There wouldn't be much sense to introduce the changes in MC just for one customer. I suggest that we stop this thread in this point.

Regards

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Postby Marina Pashkova » 24 Feb 2009

Tresor,

If data that you based your charts on were correct and such volume spikes were nor an abnormality, it would really mean that we would need to change the algorithm of plotting volume bars. We would need to at least increase the number of allowed divisions from 5 to 10.

However, I have a number of reasons to question the accuracy of your data.

1. I compared the data from one of your screenshots with data over exactly the same period in TS and in eSignal. Neither TS nor eSignal had spikes similar to what you saw with TransAct's data. At the same time, when compared to each other, TS's and eSignal's data was very close. Thus, we have evidence from two well-established feeds against data from TransAct. I would also say that eSignal's data is anything but 'crap' as you called it. eSignal provides very good quality data with very good coverage. Their data is widely used by institutional customers who would not tolerate poor quality.

2. You are saying that TransAct's data is unfiltered as opposed to that of TS or eSignal. I do not know if TS or eSignal filter ticks. However, what I do know is that some exchanges already provide aggregated ticks and not raw ticks. So no matter what data feed streams those quotes, they will not be raw by definition. If I remember correctly, some data providers announced that CME, for example, provides aggregated ticks. I am going to contact guys from CME to find out what kind of data they give access to for data providers.

3. Even if ticks are provided as snapshots every n seconds for example, the volume will be aggregated as well. This volume is not lost.

It is not a matter or whether you are the only user with this problem or not. If there is really a problem we will fix that. It is just that so far we do not see any problem at all.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 24 Feb 2009

If data that you based your charts on were correct and such volume spikes were nor an abnormality, it would really mean that we would need to change the algorithm of plotting volume bars. We would need to at least increase the number of allowed divisions from 5 to 10.
The increase of allowed divisions from 5 to 10 might do good job for 1,000 contracts bars, but again it might not be satisfactory for e.g. 100 contracts bars.

In my opinion, what would make sense - but is unacceptable for TSS and maybe for a majority of MC users at the moment for reasons you explained earlier (it might occasionally slow MC when many charts are opened) - is to get rid of the division mechanism at all.
However, I have a number of reasons to question the accuracy of your data.

1. I compared ..................

2. ...................................

3. ....................... ........... This volume is not lost.
Marina, if MC gets deprived of the ''division mechanism'', MC would plot equal volume bars, regardless of data accuracy / inaccuracy. The downside of course would be that MC might work slowly. With the introduction of the ''division mechanism'' you showed you care about the customers who do not have state-of-the-art computers (a majority of us) and watch numerous charts with symbols of small resolutions.

I would wholeheartidly welcome, if in some point in time from now you'll decide that computers reached the capacity that allows for MC to be the ''division mechanism''-free.

It is not a matter or whether you are the only user with this problem or not. If there is really a problem we will fix that. It is just that so far we do not see any problem at all.
I will check this once MC with the new Transact data feed support is released and will report if I encounter a behaviour that would be specifically painfull.

Whether you decide to so something with this or not, I believe the worst thing to do would be to increase the number of allowed divisions. In half a year another member will ask for another increase from 10 to 15, in two years from now someone may ask to increase from 15 to 30. Vicious circle.

Regards

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Postby TJ » 01 Mar 2009

Tresor,

As was explained on a number of occasions in the past, THE ALGORITHM OF PLOTTING VOUME BARS IN MC IS PERFECTLY CORRECT.

Your screenshot showed 100 volume bars with a volume indicator applied in a subchart. That indicator showed several spikes. Spikes are the result of the following:

MultiCharts plots volume bars by dividing volume into equal portions, in your case, 100 contracts. Now, if a tick arrives whose volume is more than 100 contracts it is divided into equal 100 contract portions. But the number of possible divisions is limited to 5. Thus if a tick arrives whose volume is, say, 800 contracts, you will get four 100-contract bars and one 400-contract bar. This mechanism has been implemented as a protection against cases when users try plotting volumes way smaller than normal volume values for a symbol. If the number of divisions were not limited, your memory would have been eaten up by profusely multiplying bars. You can only imagine what would happen if somebody tries plotting 1-contract bars for a symbol that is traded in thousands of contracts – which isn’t out of the realm of possibility.

The spikes in your screenshot that appear against the background of numerous perfectly equal 100-contract bars are a result of huge volume values transmitted by TransAct. MultiCharts divided the volume received in a tick into four 100-contract bars and dumped the rest of the volume into the 5 bar. As I mentioned earlier, neither TS nor eSignal have revealed any similar spikes over the period of time in question. You are saying that the quality of data by TS or eSignal is inferior to that of TransAct. I would not want to comment on that in this particular thread.
Marina:

regarding:
"MultiCharts plots volume bars by dividing volume into equal portions, in your case, 100 contracts. Now, if a tick arrives whose volume is more than 100 contracts it is divided into equal 100 contract portions. But the number of possible divisions is limited to 5. "

I have given this a lot of thoughts in the past few days.

I must say, under the concept of "High Definition Charting", I cannot agree to this short cut.

If a customer were to study CONSTANT volume... then he needs to plot precise constant volume bars.
Not "some kind of" volume bars.
Not short cut volume bars.

If he wants to study a large sample of data, and he runs out of memory, then he needs to fix either the sample size, sample frequency, or the memory size.
MC should also have error handling routine to warn user if his memory leak might crash the system.

Capping the bars to 5 defeats the High Definition Charting concept.

I urge TSS to reverse this short cut.

regards
TJ

fs

Postby fs » 01 Mar 2009

This issue about volume charts was reported last year already in this thread:

http://forum.tssupport.com/viewtopic.php?t=5533

This was also shown that NT does not have any issues with plotting Volume Charts correctly. Looking in the NT support forums I couldn't find any complains about performance issues due to plotting volume charts, so the reason that it is done due to performance issues is not a valid excuse in my mind for not doing it correctly.

The claim from TSSupport that "THE ALGORITHM OF PLOTTING VOUME BARS IN MC IS PERFECTLY CORRECT" is incorrect. What is correct is that MC plots volume charts perfectly according to their algorithm, but that doesn't mean the algorithm is correct and in this case it isn't. The definition of Constant Volume Bars is that ALL bars have the same volume and not that most of the bars have the same volume as MC does it.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 02 Mar 2009

We do understand that constant volume bars must be constant. We will increase number of splits to 20. However in some cases it will not work. So we create a number of splits as registry key for advanced users who want to have full control.

You should understand that 5 bar limitation has been created as protection measure from beginners who don't understand how to use the volume bars properly.
I've created the same problematic situation in NT. I just set MSFT 1 volume bars and my computer died. See attached.
Experienced traders will not do it, but it is clear that it is only a few percent of such people. So we must think about majority and protect them from such accidents.

I hope my solution will work for all groups of users.
Attachments
lowMemory.PNG
(101.62 KiB) Downloaded 5813 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 02 Mar 2009

This issue about volume charts was reported last year already in this thread:

http://forum.tssupport.com/viewtopic.php?t=5533
Hello fs,

I didn't realize this issue had already been reported. Thanks for the link.

Regards

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 02 Mar 2009

You should understand that 5 bar limitation has been created as protection measure from beginners who don't understand how to use the volume bars properly.

..............................

I hope my solution will work for all groups of users.
Hi Andrew,

I do not understand written Russian (although I understand when it is spoken) so I have problems with understanding what the Windows warning was. In this thread http://forum.tssupport.com/viewtopic.ph ... 6&start=25 Marina said that MC uses only 1 core for the purpose of calculating / drawing price bars.

Maybe the best solution solution would be to allow MC to use all available cores for calculating / drawing price bars? Instead of one core, MC would be using two or four cores. If so, the problem with CPU's high usage might go.

Regards

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 03 Mar 2009

With a few next posts I would like to discuss issues that I have with all of Multicharts’ resolutions. After I discovered that Multicharts plots volume bars incorrectly I made extensive research on the quality of the bars’ plotting engine in MC. I believe that volume bar resolution quality is a small problem when compared to other resolutions. Other resolutions are even worse.
I will present the cases in the following order (the same order as in Format Symbol – Settings):
Tick
Contract
Point
Point (original)
Change
Time (second / minute / hour / day / week / month/ quarter / year)

On each subchart there is a study called Counter which tells if the resolution was plotted correctly or not. The study works this way:
(i) if the red bar stops in the yellow dot, this means that the bar was calculated and plotted correctly
(ii) if the red bar does not reach the yellow dot and remains partially blue, this means that the bar was closed before reaching the assumed resolution – such a bar is not complete
(iii) if the bar exceeds the yellow dot and plots above it, this means that the bar wasn’t closed on time – such a bar is over-complete

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 03 Mar 2009

TICK RESOLUTION

With some of the supported data feeds MC plots tick resolutions bars properly, but with some data feeds MC plots tick bars improperly. Screenshots with good and bad tick resolution plotting attached.

On one screenshot MC plots equal 100 tick bars – very good! On the other screenshot MC does not plot 100 ticks properly. Some of the bars are 200+, some are 400+ tick bars. This is far from very good!

One should not trade, apply studies, backtest and optimize strategies based on tick resolution unless one is sure that his data feed is processed properly by MC. Tick resolution is miscalculated and badly displayed. Tick resolution should be removed from MC or fixed.


[/u]

P.S. 6th March 2009 I was wrong about tick bars. They are calculated and displayed CORRECTLY. One can safely trade, apply studies, backtest and optimize strategies based on tick resolution.
Attachments
Tick resolution bad.jpg
(119.56 KiB) Downloaded 5825 times
Tick resolution good.jpg
(82.91 KiB) Downloaded 5822 times
Last edited by Tresor on 06 Mar 2009, edited 2 times in total.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 03 Mar 2009

VOLUME (CONTRACT) RESOLUTION - BUILD VOLUME ON: TRADE VOLUME

This was already discussed earlier in this and earlier posts. With all (not just some!), with all of the supported data feeds MC plots volume resolutions bars improperly (because of the ‘’# of allowed division mechanism’’). The volume bars can be either over-completed and under-completed.

Screenshot attached.

One should not trade, apply studies, backtest and optimize strategies based on volume resolution (Build Volume on: Trade Volume). Volume resolution is miscalculated (using ‘’properly working algorithm’’) and badly displayed. Either volume resolution should be removed from MC or division mechanism should be removed from MC.
Attachments
Contract resolution Trade Volume.jpg
(107.63 KiB) Downloaded 5806 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 03 Mar 2009

VOLUME (CONTRACT) RESOLUTION - BUILD VOLUME ON: TICK COUNT


With all (not just some!) of the supported data feeds MC plots volume resolutions bars improperly (because of the ‘’# of allowed division mechanism’’).

In fact when the ‘’Tick Count’’ option is selected, the bars are plotted identical way as with the ‘’Trade Volume’’ option. I do not know if it is a bug or not. But if it is a bug then the volume bars should be made of equal volume bars calculated on tick count, not trade volume. At themoment all bars are under-complete.

Please judge it based on the attached screenshot.

One should not trade, apply studies, backtest and optimize strategies based on volume resolution with volume built on Tick Count.
Attachments
Contract resolution Tick Count.jpg
(106.6 KiB) Downloaded 5823 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 03 Mar 2009

POINT RESOLUTION


With all of the supported data feeds MC calculates and plots Range bars incorrectly! Screenshot attached. Please see the differences in ranges.

One should not trade, apply studies, backtest and optimize strategies based on Point resolution. Point resolution is miscalculated and badly displayed. Point resolution should be removed from MC or fixed.
Attachments
Point resolution.jpg
(75.81 KiB) Downloaded 5800 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 03 Mar 2009

POINT (ORIGINAL) RESOLUTION

With all of the supported data feeds MC plots Point (original) bars incorrectly! Screenshot attached. Please see the differences in ranges. Marina already admitted that this resolution is not calculated and displayed properly.

One should not trade, apply studies, backtest and optimize strategies based on Point (original) resolution. Point (original) resolution is miscalculated and badly displayed. Point (original) resolution should be removed from MC or fixed.
Attachments
Point original resolution.jpg
(101.74 KiB) Downloaded 5789 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 03 Mar 2009

There are two resolutions left to be discussed: Change resolution and Time resolutions. As they are my favourite ones and I am still mastering the stop-watch, the quality of these resolutions will be discussed tomorrow :mrgreen:

Regards

miltonc4
Posts: 150
Joined: 14 Apr 2006
Has thanked: 1 time
Been thanked: 4 times

Postby miltonc4 » 04 Mar 2009

TS Support
Thought I would take a look at Point charts as well,and note the attached screenshot shows 8 Point Bars ( cyan bars)
The open for next bar is 1 point below last bars close for a down bar, and 1 point above last close for an up bar
Should the open of a new bar be the same as the close of the previous bar?

Thereon is a 2 point chart (Black bars) inserted and same situation, this time the 1 point gaps are more visible

Please advise
Milton[/img]
Attachments
Point chart.gif
(205.31 KiB) Downloaded 5786 times

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Postby TJ » 04 Mar 2009

Constant Volume Bar "Holes"

another problem with MC's constant volume bar logic:

it seems that when the tick has more contracts than the bar specification,
MC simply divide the number by 5.

e.g.
if the chart calls for a 100 contract volume bar
and the tick has 250 contracts,
MC would simply divide the tick into 100, 100, and 50.

you will see:
the first bar with 100 contracts,
the second bar with 100 contracts,
the third bar with 50 contracts !!!

whatever comes after is NOT appended to the third bar, but start anew in the fourth bar.

please see attached for illustration.

THIS IS NOT HIGH DEFINATION CHARTING
Attachments
constant_volume_bar_holes.gif
(54.37 KiB) Downloaded 5785 times

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 04 Mar 2009

You should understand that 5 bar limitation has been created as protection measure from beginners who don't understand how to use the volume bars properly.

..............................

I hope my solution will work for all groups of users.
Hi Andrew,

I do not understand written Russian (although I understand when it is spoken) so I have problems with understanding what the Windows warning was. In this thread http://forum.tssupport.com/viewtopic.ph ... 6&start=25 Marina said that MC uses only 1 core for the purpose of calculating / drawing price bars.

Maybe the best solution solution would be to allow MC to use all available cores for calculating / drawing price bars? Instead of one core, MC would be using two or four cores. If so, the problem with CPU's high usage might go.

Regards
The screenshot illustrates that if there is no limit of splits for volume bars any program will consume all available memory and finally will crash.
We will increase the number of splits and all reported problems will be closed.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 04 Mar 2009

With a few next posts I would like to discuss issues that I have with all of Multicharts’ resolutions. After I discovered that Multicharts plots volume bars incorrectly I made extensive research on the quality of the bars’ plotting engine in MC. I believe that volume bar resolution quality is a small problem when compared to other resolutions. Other resolutions are even worse.
I will present the cases in the following order (the same order as in Format Symbol – Settings):
Tick
Contract
Point
Point (original)
Change
Time (second / minute / hour / day / week / month/ quarter / year)

On each subchart there is a study called Counter which tells if the resolution was plotted correctly or not. The study works this way:
(i) if the red bar stops in the yellow dot, this means that the bar was calculated and plotted correctly
(ii) if the red bar does not reach the yellow dot and remains partially blue, this means that the bar was closed before reaching the assumed resolution – such a bar is not complete
(iii) if the bar exceeds the yellow dot and plots above it, this means that the bar wasn’t closed on time – such a bar is over-complete
Please publish the counter study. We need to see how it is calculated.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 04 Mar 2009

Constant Volume Bar "Holes"

another problem with MC's constant volume bar logic:

it seems that when the tick has more contracts than the bar specification,
MC simply divide the number by 5.

e.g.
if the chart calls for a 100 contract volume bar
and the tick has 250 contracts,
MC would simply divide the tick into 100, 100, and 50.

you will see:
the first bar with 100 contracts,
the second bar with 100 contracts,
the third bar with 50 contracts !!!

whatever comes after is NOT appended to the third bar, but start anew in the fourth bar.

please see attached for illustration.

THIS IS NOT HIGH DEFINATION CHARTING

TJ,
What version are you running?
The bug you are illustrating has been fixed in MultiCharts.4.0.1724.400_Release.exe.
According to visual signs you are running 4.0, but it could be beta which doesn't have the fix. Please let me know.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 04 Mar 2009

Constant Volume Bar "Holes"

another problem with MC's constant volume bar logic:

it seems that when the tick has more contracts than the bar specification,
MC simply divide the number by 5.

e.g.
if the chart calls for a 100 contract volume bar
and the tick has 250 contracts,
MC would simply divide the tick into 100, 100, and 50.

you will see:
the first bar with 100 contracts,
the second bar with 100 contracts,
the third bar with 50 contracts !!!

whatever comes after is NOT appended to the third bar, but start anew in the fourth bar.

please see attached for illustration.

THIS IS NOT HIGH DEFINATION CHARTING

TJ,
What version are you running?
The bug you are illustrating has been fixed in MultiCharts.4.0.1724.400_Release.exe.
According to visual signs you are running 4.0, but it could be beta which doesn't have the fix. Please let me know.
Hi Andrew,

This is also the case with Version 5.0 Beta 2 (Build 1781)

Regards

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 04 Mar 2009

Please publish the counter study. We need to see how it is calculated.
Hello Andrew,

I would be reluctant to publish the study. At least not now for two reasons:

1. It is still a Beta (it uses the computer’s time to measure time resolution’s correctness while it should use EL Time&Sales, which my programmer could not get from MC nor I could get a straight answer in this forum if Time&Sales is available in MC http://forum.tssupport.com/viewtopic.php?t=5985 ) – that’s why I bought watch-stop yesterday to verify this study. If I publish this study, some less experienced users might be confused.

2. I paid for the study a few hundred bucks + it contains a number of features not shown on the screenshots

Anyway, whether I decide to publish the study or not, you can EASILY see the study does it job, i.e. it detects bars that were incorrectly calculated and incorrectly plotted like in this example http://forum.tssupport.com/download.php?id=1986 or like in this example http://forum.tssupport.com/download.php?id=1987

GENERAL COMMENTS

I bought MC 1.5 years ago. Since then, despite I was not very happy about the way of the product is being developed (focusing on bells and whistles), I tried to support MC the way a newbie can (mostly by reporting bugs). I even started a thread devoted to MC on kreslik.com for the benefit of MC users on this forum: http://kreslik.com/forums/viewtopic.php?t=1141

Since you are nearing to releasing MC Gold, I think you need to get the basic issues in MC fixed (once and for all) before you release MC Gold or with the release of MC Gold. I know you need to keep on adding new eye-catching features to have a constant stream of revenues generated by new customers.

But with all those candy-like new features, based on bad charting and bad data management (and bad math / algorithms that support charting and data management) MC looks like a giant on clay legs. I think you should add some strength and flexibility to these legs. Simply stop working on whatever bell and whistle you are working now and make basic features (charting and data management) rock-solid.

I never posted any negative comments about MC on any internet forum. I was only being ironic on this forum from time to time. I did not rate MC on elitetrader.com because I thought MC was not ready for my positive rating yet.

I will hopefully rate MC very positive on elitetrader.com is when you release Gold version. My rating will be very positive if you only fix charting and data management.

Regards

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Postby TJ » 04 Mar 2009

Constant Volume Bar "Holes"
another problem with MC's constant volume bar logic:
it seems that when the tick has more contracts than the bar specification,
MC simply divide the number by 5.
e.g.
if the chart calls for a 100 contract volume bar
and the tick has 250 contracts,
MC would simply divide the tick into 100, 100, and 50.
you will see:
the first bar with 100 contracts,
the second bar with 100 contracts,
the third bar with 50 contracts !!!
whatever comes after is NOT appended to the third bar, but start anew in the fourth bar.
please see attached for illustration.
THIS IS NOT HIGH DEFINATION CHARTING

TJ,
What version are you running?
The bug you are illustrating has been fixed in MultiCharts.4.0.1724.400_Release.exe.
According to visual signs you are running 4.0, but it could be beta which doesn't have the fix. Please let me know.
Andrew:

I am using v4 beta.

I am glad to hear this "WAS" a bug, and it has been fixed. Thank you !

best regards
TJ

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 04 Mar 2009

Please publish the counter study. We need to see how it is calculated.
Hello Andrew,

I would be reluctant to publish the study. At least not now for two reasons:

1. It is still a Beta (it uses the computer’s time to measure time resolution’s correctness while it should use EL Time&Sales, which my programmer could not get from MC nor I could get a straight answer in this forum if Time&Sales is available in MC http://forum.tssupport.com/viewtopic.php?t=5985 ) – that’s why I bought watch-stop yesterday to verify this study. If I publish this study, some less experienced users might be confused.

2. I paid for the study a few hundred bucks + it contains a number of features not shown on the screenshots

Anyway, whether I decide to publish the study or not, you can EASILY see the study does it job, i.e. it detects bars that were incorrectly calculated and incorrectly plotted like in this example http://forum.tssupport.com/download.php?id=1986 or like in this example http://forum.tssupport.com/download.php?id=1987

GENERAL COMMENTS

I bought MC 1.5 years ago. Since then, despite I was not very happy about the way of the product is being developed (focusing on bells and whistles), I tried to support MC the way a newbie can (mostly by reporting bugs). I even started a thread devoted to MC on kreslik.com for the benefit of MC users on this forum: http://kreslik.com/forums/viewtopic.php?t=1141

Since you are nearing to releasing MC Gold, I think you need to get the basic issues in MC fixed (once and for all) before you release MC Gold or with the release of MC Gold. I know you need to keep on adding new eye-catching features to have a constant stream of revenues generated by new customers.

But with all those candy-like new features, based on bad charting and bad data management (and bad math / algorithms that support charting and data management) MC looks like a giant on clay legs. I think you should add some strength and flexibility to these legs. Simply stop working on whatever bell and whistle you are working now and make basic features (charting and data management) rock-solid.

I never posted any negative comments about MC on any internet forum. I was only being ironic on this forum from time to time. I did not rate MC on elitetrader.com because I thought MC was not ready for my positive rating yet.

I will hopefully rate MC very positive on elitetrader.com is when you release Gold version. My rating will be very positive if you only fix charting and data management.

Regards

Tresor,
You should understand that if I don't have the indicator code I never know how you calculate it. Please use the standard volume indicator or disclose the code. Otherwise the screenshots don’t allow us to be 100% that it is true.

I've just checked range bars on DAX symbol and they are all identical. I can send screenshots to prove. I need to know your settings in QuoteManager for this symbol. If they are wrong the bars will be spitted improperly.

I've created a new DLL that splits volume up to 30 times instead of 5.
I'm attaching it. The DLL will work with MC 5 Beta 4 only.
Please test how it works let us know.
Attachments
tsShaper.zip
(318.64 KiB) Downloaded 228 times
tsShaper.zip
(318.64 KiB) Downloaded 217 times

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 04 Mar 2009

TICK RESOLUTION

With some of the supported data feeds MC plots tick resolutions bars properly, but with some data feeds MC plots tick bars improperly. Screenshots with good and bad tick resolution plotting attached.

On one screenshot MC plots equal 100 tick bars – very good! On the other screenshot MC does not plot 100 ticks properly. Some of the bars are 200+, some are 400+ tick bars. This is far from very good!

One should not trade, apply studies, backtest and optimize strategies based on tick resolution unless one is sure that his data feed is processed properly by MC. Tick resolution is miscalculated and badly displayed. Tick resolution should be removed from MC or fixed.


[/u]
According to our tests all tick bars are correct. Moreover we don't split it so it should be correct since volume issue is not relevant here. Please reproduce wrong tick bars using volume indicator, not your Counter one. It is possible it doesn't work properly.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 04 Mar 2009

POINT RESOLUTION


With all of the supported data feeds MC calculates and plots Range bars incorrectly! Screenshot attached. Please see the differences in ranges.

One should not trade, apply studies, backtest and optimize strategies based on Point resolution. Point resolution is miscalculated and badly displayed. Point resolution should be removed from MC or fixed.
It works properly on our end and if you set right settings in QM It should work for you. Please see my screenshots. If you use wrong settings you bars may be spitted and after the fifth bar you will have a longer one.
Attachments
DAX_IB_100_Point.png
(22.25 KiB) Downloaded 5776 times
DAX_Setting.PNG
(11.47 KiB) Downloaded 5787 times

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 04 Mar 2009

You should understand that if I don't have the indicator code I never know how you calculate it. Please use the standard volume indicator or disclose the code. Otherwise the screenshots don’t allow us to be 100% that it is true.
Here you are: a screenshot with both Counter and the standard volume indicator put together.

Regards
Attachments
Volume&Counter.jpg
(116.27 KiB) Downloaded 5776 times

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 04 Mar 2009

POINT (ORIGINAL) RESOLUTION

With all of the supported data feeds MC plots Point (original) bars incorrectly! Screenshot attached. Please see the differences in ranges. Marina already admitted that this resolution is not calculated and displayed properly.

One should not trade, apply studies, backtest and optimize strategies based on Point (original) resolution. Point (original) resolution is miscalculated and badly displayed. Point (original) resolution should be removed from MC or fixed.
As Marina told you before it is a legacy and it is not right, but some people think that they are very good and begged us not to remove it!

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 04 Mar 2009

You should understand that if I don't have the indicator code I never know how you calculate it. Please use the standard volume indicator or disclose the code. Otherwise the screenshots don’t allow us to be 100% that it is true.
Here you are: a screenshot with both Counter and the standard volume indicator put together.

Regards
Please use the DLL attached and you will not encounter the spikes. What I want you to reproduce is under-complete bars.
I strongly suggest you to install the beta 4 and the DLL. Otherwise it is likely you waste your time.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 04 Mar 2009

P.S. Thank you for the help!

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 04 Mar 2009

But with all those candy-like new features, based on bad charting and bad data management (and bad math / algorithms that support charting and data management) MC looks like a giant on clay legs. I think you should add some strength and flexibility to these legs. Simply stop working on whatever bell and whistle you are working now and make basic features (charting and data management) rock-solid.

I will hopefully rate MC very positive on elitetrader.com is when you release Gold version. My rating will be very positive if you only fix charting and data management.

Regards
Tresor,
Could you specify "bad data management"? What exactly doesn't work?

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 04 Mar 2009

I've just checked range bars on DAX symbol and they are all identical. I can send screenshots to prove. I need to know your settings in QuoteManager for this symbol. If they are wrong the bars will be spitted improperly.
Here is the screenshot with DAX settings (the same as yours). As you can see the Range Bars are not plotted properly.

Andrew, I really wouldn't like to go deeper on my end (I am not a computer geek. Anything that is more complicated than ''copy and paste'' overwhelms my circuits).

I paid the programmer to code a tool which among other things helps in identifying problems with MC charting. What I am doing now is reporting these problems to you. This is my contribution to the product development. I spent many hours of doing tests. I am proving these tests with screenshots (no Photoshop tricks on my side).

Please get your engineers together and find the solution. Redesign the engine / check the quality of data feed providers / cross-check results / anything. But please for your own good, do it before Gold release.

Regards

P.S. And PLEASE avail EasyLanguage Time&Sales
Attachments
Range Bar DAX.jpg
(145.22 KiB) Downloaded 5796 times
Last edited by Tresor on 04 Mar 2009, edited 1 time in total.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 04 Mar 2009

POINT (ORIGINAL) RESOLUTION

With all of the supported data feeds MC plots Point (original) bars incorrectly! Screenshot attached. Please see the differences in ranges. Marina already admitted that this resolution is not calculated and displayed properly.

One should not trade, apply studies, backtest and optimize strategies based on Point (original) resolution. Point (original) resolution is miscalculated and badly displayed. Point (original) resolution should be removed from MC or fixed.
As Marina told you before it is a legacy and it is not right, but some people think that they are very good and begged us not to remove it!
Yeah but something that is not working may only confuse the new and unexperienced users and may make them frustrated.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 04 Mar 2009

Tresor,
Could you specify "bad data management"? What exactly doesn't work?
Please refer for this to the e-mail I sent to you and 3 brokers (and data feed providers) and the broker's reply. 30th Jan 2009.

Regards

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 04 Mar 2009

Please use the DLL attached and you will not encounter the spikes. What I want you to reproduce is under-complete bars.
I strongly suggest you to install the beta 4 and the DLL. Otherwise it is likely you waste your time.
Andrew, I am still using the beta 2. What I will do after a few days or a week (after I am back from my holiday for which I am taking off tomorrow morning), I will dowload the beta 5 which will hopefully have the promised revised processing of Transact data feed and I will run all the experiments once again.

Maybe MC's wrongly displaying tick bars with this data feed http://forum.tssupport.com/download.php?id=1983 and properly displaying tick bars with the other data feed http://forum.tssupport.com/download.php?id=1982 is data management-related, and not charting-related? Just guessing :roll:

After the experiments are done I will communicate the results to you via PM + I will send you Counter study for your eyes only (please remember the study is a beta that will not work perfectly with Time resolutions).

After a perfection in MC's charting is achieved on all supported data feeds (also on Change and Time resolutions not posted by me here today, although promised) we will both communicate this happily to the members of this forum.

You will be able to release Gold version of MC and I will finally be able to post my positive rating for MC on elitetrader.com

Would this plan be acceptable for you?

Regards

P.S. In case changes to the data management will be needed, it would be nice AT THE SAME TIME to adjust MC to official symbol settings given by exchanges, i.e. it would be sensical and logical to use 0.5 Min Movement for FDAX or 0.25 for ES as discussed in this thread http://forum.tssupport.com/viewtopic.ph ... t=settings

At the moment finding the right settings for symbols in MC is a neck-breaking procedure for both newbies and experienced users.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 05 Mar 2009


Maybe MC's wrongly displaying tick bars with this data feed http://forum.tssupport.com/download.php?id=1983 and properly displaying tick bars with the other data feed http://forum.tssupport.com/download.php?id=1982 is data management-related, and not charting-related? Just guessing
Tresor,
Tick charts work properly. I’m very confident. I don’t know how your counter calculates it, but they can’t work improperly due to their nature.

After the experiments are done I will communicate the results to you via PM + I will send you Counter study for your eyes only (please remember the study is a beta that will not work perfectly with Time resolutions).
Yes, I need to it see what is wrong.
After a perfection in MC's charting is achieved on all supported data feeds (also on Change and Time resolutions not posted by me here today, although promised) we will both communicate this happily to the members of this forum.
We can use the forum for all communications, because we don’t have thing that we want hide. If you prefer private communication it is ok for us too.

You will be able to release Gold version of MC and I will finally be able to post my positive rating for MC on elitetrader.com

Would this plan be acceptable for you?
My goal is to prove that our charting is correct (except volume bars that we already increased to 30 splits). If my arguments are not correct, we will fix the problems. So far we see that your range bars tests are incorrect.
Here is why:
You use 100 points bars. 100 points is about full day movement of DAX. You will have 1 or 2 bars a day. And the second bar will be incomplete due to the end of the session. Take a look at your screenshot. You have many small bars that are even not close to 100 points.
Explanation is quite simple. You used wrong default settings in QuoteManager. After I published my instructions you changed the settings to correct ones. However it will not take effect since the old wrong bars are cached in QM. You should Close charts, go to QM->select symbol->Clear cache. If you use the settings I provided before you chart will correct and you will have 1-2 bars a days.
Your range bars were different, because our 5 bar split mechanism closed the 5th bar.


Regards

P.S. In case changes to the data management will be needed, it would be nice AT THE SAME TIME to adjust MC to official symbol settings given by exchanges, i.e. it would be sensical and logical to use 0.5 Min Movement for FDAX or 0.25 for ES as discussed in this thread http://forum.tssupport.com/viewtopic.ph ... t=settings

At the moment finding the right settings for symbols in MC is a neck-breaking procedure for both newbies and experienced users.
Transact doesn’t give us correct settings for each symbol. IB Does. This is why we take by default exchange settings. However the exchanges settings can’t be true for all symbols trades on the exchange and this is why it will not work in all cases.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 05 Mar 2009

My goal is to prove that our charting is correct (except volume bars that we already increased to 30 splits). If my arguments are not correct, we will fix the problems. So far we see that your range bars tests are incorrect.
Here is why:
You use 100 points bars. 100 points is about full day movement of DAX. You will have 1 or 2 bars a day. And the second bar will be incomplete due to the end of the session. Take a look at your screenshot. You have many small bars that are even not close to 100 points.
Explanation is quite simple. You used wrong default settings in QuoteManager. After I published my instructions you changed the settings to correct ones. However it will not take effect since the old wrong bars are cached in QM. You should Close charts, go to QM->select symbol->Clear cache. If you use the settings I provided before you chart will correct and you will have 1-2 bars a days.
Your range bars were different, because our 5 bar split mechanism closed the 5th bar.
You are right on this point. I didn't realize that clearing cache is needed. My bad in this case. I have only 16 new bars so it is too little to draw conclusion at this stage.

Regards

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 05 Mar 2009


You are right on this point. I didn't realize that clearing cache is needed. My bad in this case. I have only 16 new bars so it is too little to draw conclusion at this stage.
Tresor,
Just set 5 point bar instead of 100 and you will have twenty times more! 5 point is quite significant movement for DAX so I think it makes sense.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 05 Mar 2009

Andrew,

Rest assured that if the incorrect results on resolution quality that I was given by MC result from my entering bad symbol settings in QM, I will remove all my posts (expect for volume) and I will remove the screenshots.

Regards

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 05 Mar 2009

I will be enormously grateful if a kind soul could tell me the proper symbol settings in QM:

Price Scale
Daily Limit
Min. Movement
Big Point Value

for the following symbols: 6E, EMD, ES, FESX, GC, NQ, TF and YM.

Thanks

SUPER
Posts: 646
Joined: 03 Mar 2007
Has thanked: 106 times
Been thanked: 84 times

Postby SUPER » 05 Mar 2009

I will be enormously grateful if a kind soul could tell me the proper symbol settings in QM:

Price Scale
Daily Limit
Min. Movement
Big Point Value

for the following symbols: 6E, EMD, ES, FESX, GC, NQ, TF and YM.

Thanks
Symbol: QM
Price Scale = 1/1000
Daily Limit = 1
Min. Movement = 25
Big Point Value = 500

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 05 Mar 2009

I will be enormously grateful if a kind soul could tell me the proper symbol settings in QM:

Price Scale
Daily Limit
Min. Movement
Big Point Value

for the following symbols: 6E, EMD, ES, FESX, GC, NQ, TF and YM.

Thanks
Symbol: QM
Price Scale = 1/1000
Daily Limit = 1
Min. Movement = 25
Big Point Value = 500
it is not correct settings.

6E
http://www.cme.com/trading/prd/fx/euro_OCS.html

EMD
http://www.cme.com/trading/prd/equity/s ... 0_FCS.html

ES
http://www.cme.com/trading/prd/equity/e ... 0_FCS.html
and so on. See CME.com
Last edited by Andrew Kirillov on 05 Mar 2009, edited 1 time in total.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 05 Mar 2009

THX

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 05 Mar 2009

Andrew,

Rest assured that if the incorrect results on resolution quality that I was given by MC result from my entering bad symbol settings in QM, I will remove all my posts (expect for volume) and I will remove the screenshots.

Regards
Fine.
The Volume bars should be good as well since we've increased the number of splits. Please test and let us know.

SUPER
Posts: 646
Joined: 03 Mar 2007
Has thanked: 106 times
Been thanked: 84 times

Postby SUPER » 05 Mar 2009

I will be enormously grateful if a kind soul could tell me the proper symbol settings in QM:

Price Scale
Daily Limit
Min. Movement
Big Point Value

for the following symbols: 6E, EMD, ES, FESX, GC, NQ, TF and YM.

Thanks
Symbol: QM
Price Scale = 1/1000
Daily Limit = 1
Min. Movement = 25
Big Point Value = 500
it is not correct settings.

6E
http://www.cme.com/trading/prd/fx/euro_OCS.html

EMD
http://www.cme.com/trading/prd/equity/s ... 0_FCS.html

ES
http://www.cme.com/trading/prd/equity/e ... 0_FCS.html
and so on. See CME.com
Andrew,

Please can you tell me what is wrong in QM Symbol setting, the values I quoted are from IB (see attachment).

http://www.interactivebrokers.co.uk/con ... 1236277790

Regards
Super
Attachments
QM-1.JPG
(146.81 KiB) Downloaded 5789 times
Last edited by SUPER on 05 Mar 2009, edited 1 time in total.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 05 Mar 2009

We need ''QM symbol settings for dummies'' thread or something similar.

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Postby TJ » 05 Mar 2009

Fine.
The Volume bars should be good as well since we've increased the number of splits. Please test and let us know.
you mean the split is dynamically updated ?
I don't need to download a new version of MC?

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 06 Mar 2009


Andrew,

Please can you tell me what is wrong in QM Symbol setting, the values I quoted are from IB (see attachment).
Super
Tresor meant QuoteManager when he said QM. You mean qm symbol.
He indicated the following symbols that have different settings:
6E, EMD, ES, FESX, GC, NQ, TF and YM.

You understand of “qm” symbol settings is correct.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 06 Mar 2009

Fine.
The Volume bars should be good as well since we've increased the number of splits. Please test and let us know.
you mean the split is dynamically updated ?
I don't need to download a new version of MC?
You will have to download Beta 5 to fix the issue. Beta 5 is under development now.

We confirm that range bars work improperly after the 5th split.

Tresor
Posts: 1104
Joined: 29 Mar 2008
Has thanked: 12 times
Been thanked: 53 times

Postby Tresor » 06 Mar 2009

TICK RESOLUTION

With some of the supported data feeds MC plots tick resolutions bars properly, but with some data feeds MC plots tick bars improperly. Screenshots with good and bad tick resolution plotting attached.

On one screenshot MC plots equal 100 tick bars – very good! On the other screenshot MC does not plot 100 ticks properly. Some of the bars are 200+, some are 400+ tick bars. This is far from very good!

One should not trade, apply studies, backtest and optimize strategies based on tick resolution unless one is sure that his data feed is processed properly by MC. Tick resolution is miscalculated and badly displayed. Tick resolution should be removed from MC or fixed.


[/u]
According to our tests all tick bars are correct. Moreover we don't split it so it should be correct since volume issue is not relevant here. Please reproduce wrong tick bars using volume indicator, not your Counter one. It is possible it doesn't work properly.
Andrew,

Yes, you were right on tick bars. I was wrong. I apologize for the mess I did with regard to tick bars which are calculated and displayed CORRECTLY.

Anybody can do the experiment by themselves, as per attached screenshot. If the volume bars are equal (build volume: tick count), then tick bars must be CORRECT.

Regards
Attachments
Tick resolution.jpg
(97.94 KiB) Downloaded 5790 times

Nick
Posts: 496
Joined: 04 Aug 2006
Has thanked: 4 times
Been thanked: 24 times

Postby Nick » 09 Mar 2009

Late to this thread and kind of skimmed it. One question

Why limit the number of splits? As long as the number is less than the maximum bars a chart can support whats the problem? I can't imagine why a would want to plot an instrument that trades in 1000's of lots with 1 lot volume charts but if they do, let them. Sure its going to take longer to process each print but if they accept that so what?

Why design in an arbitrary fixed limit? If 30 is good isn't 50 better?

Of course a tick filter might be prudent for those -1 volume prints.

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Postby TJ » 09 Mar 2009

Late to this thread and kind of skimmed it. One question

Why limit the number of splits? ....
because the computer will crash.

this is not to suggest MC has problem.
NT crashes under same condition as well.
everything has a limit.
this is just the limitation of computer/language/operating system.
you can design an error-check around it, but then I think MC has better things to do than to spend time to protect a rare minority situation.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 10 Mar 2009

Nick,
Initially we didn't have any restrictions and it caused us a lot of support. Newbie set very small volumes or used wrong price scale/min movement settings. As a result MC generates billions of artificial bars that eventually consumed all recourses and crashed PC.
This is why decided to limit it. I agree that 5 splits are not adequate.
30 should work.

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Postby TJ » 10 Mar 2009

Nick,
Initially we didn't have any restrictions and it caused us a lot of support. Newbie set very small volumes or used wrong price scale/min movement settings. As a result MC generates billions of artificial bars that eventually consumed all recourses and crashed PC.
This is why decided to limit it. I agree that 5 splits are not adequate.
30 should work.
MC should make a note on the Format Symbol page, that there is a split limit of 30.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 10 Mar 2009

it dounds reasonable. Maybe we need to make a popup message when you select count-based resolutions. However I think 30 splits will work for all for at least a year or two:)

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Postby TJ » 10 Mar 2009

it dounds reasonable. Maybe we need to make a popup message when you select count-based resolutions. However I think 30 splits will work for all for at least a year or two:)
I agree, I think 30 should do.

whoever chooses a contract volume that requires more than 10 splits must have a special reason for doing so. I don't see it useful in daytrading application because you will be getting lots of horizontal dashes.

The 30 split should cover most (99.9%) of the unusual high volume spike situations.

Thanks for making the change.

;-)

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 10 Mar 2009

Right.

Nick
Posts: 496
Joined: 04 Aug 2006
Has thanked: 4 times
Been thanked: 24 times

Postby Nick » 11 Mar 2009

Nick,
Initially we didn't have any restrictions and it caused us a lot of support. Newbie set very small volumes or used wrong price scale/min movement settings. As a result MC generates billions of artificial bars that eventually consumed all recourses and crashed PC.
This is why decided to limit it. I agree that 5 splits are not adequate.
30 should work.
Ahh OK maybe have a parameter that defaults to 30? Then a user would have to explicitly change it. Actually its not really that big of a deal for me.

Nick
Posts: 496
Joined: 04 Aug 2006
Has thanked: 4 times
Been thanked: 24 times

Postby Nick » 11 Mar 2009

Nick,
Initially we didn't have any restrictions and it caused us a lot of support. Newbie set very small volumes or used wrong price scale/min movement settings. As a result MC generates billions of artificial bars that eventually consumed all recourses and crashed PC.
This is why decided to limit it. I agree that 5 splits are not adequate.
30 should work.
Ahh OK maybe have a parameter that defaults to 30? Then a user would have to explicitly change it. Actually its not really that big of a deal for me.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 11 Mar 2009

Yes, but will make it as a registry key.


Return to “MultiCharts”