Chart Scaling Question...

[Alfred]

In the left screen shot below I stacked a 20 second chart over a 40 second chart...I expected to see that the 40 second chart would be 1/2 of the 20 second chart in horizontal size...I measured the size with an architects scale on the screen and show the actual dimensions....you can see graphically from the center line of the 20 second chart that the 40 second chart is quite a bit over 1/2 of the 20 second chart horizontally. Equidistant spacing was set to False on both charts. Bar width was set to 1 on both charts.

When equadistant spacing is set to True then the 40 second chart scales 1/2 of the 20 second chart horizontally OK...see right screen shot below.

You can see in the screen rulers that there are 1/2 of the bars in the 40 second chart vs the 20 second chart....All NT charts are set to max compression.

Out of curiosity I stacked a 1 minute and 2 minute chart in another program...you can see that graphically the 2 minute chart is 1/2 of the 1 minute chart in horizontal scale, as it should be. This is the center screen shot below.

Have been playing a lot with NT chart horizontal scales recently and got to wondering about this...any ideas ??

Thanks...

[Alfred]

7B6 Horizontal compression scaling of tick charts...

Here is a video on horizontal scaling of tick charts set to equidistant scaling. I am wondering if the various horizontal compression settings are working correctly...settings for the chart are included in the video.

http://www.screencast.com/t/OTExMDEzYjgt

Often have to redo the period on a chart several times to get the compression working in a seemingly correct scale.

Wonder if other users have noticed this sort of thing...

Thanks...

[NinjaTrader_JoshP]

Alfred,

For the first post, you should check the number of actual bars being shown within that 1/2 pt span on the top left chart and then compare that against the number of bars being shown in the same span on the bottom chart. Running a ruler across those spans should yield a bar count of the same number.

For the chart compression, I don't think you can just assume a 20% reduction. When you use non-equidistant charts the x-axis is independent of the bars and provides you equal chunks of time on the x-axis. How those bars fall within the inch of space on the x-axis depends on the time stamp for those ticks. You can see the x-axis time intervals changing as you change tick size to try and adjust for the changing chart.

[Alfred]

Josh....

The rulers on the first chart show the number of bars.

I TradeStation one can elect to show empty bar spaces or not...
there when you elect to show empty spaces the horizontal scale of the charts does not get totally messed up when one changes the chart period as in your system...and in TradeStation if there are empty bars an equal amount of time per chart period is inserted for each empty bar....couldn't there be something like that in NT ??....I am very interested in showing empty bars...for accurate graphic techniques....

The way NT has set this empty bar thing up, one never knows what a chart will look like until you try with trial and error to get something acceptable when you change the period...this thing is now often just all over the map...the way the empty bar idea is set up in NT now is very disapointing...is this the only way that could have been done ??

Thanks...

[Alfred]

A few more thoughts...

Josh...

Edited this earlier post...

Taking another look at a 44 tick bar chart....expanding and contracting to 55 ticks and 44 ticks...

Am now getting different results than in the first video...had to repair DB today and do new workspaces....am now getting (perhaps) more realistic expansion and contraction of the chart...but curiously the horizontal scale on the screen now remains the same dimension for all 3 periods...is this to be expected ??...or could I have an incorrect intallation of NT7 perhaps ??

Here is a new video on tick bars scaling shortly after the close...

http://www.screencast.com/t/YzYzNzU0

And here is a short video with seconds bars on the same chart pattern as earlier videos...here the horizontal scale changes OK as the period changes.

http://www.screencast.com/t/NGYwMGU1Nm

Any suggestions or ideas...??


Thanks for your patience ....

[Alfred]

Note...

An observation this evening...after hours it is next to impossible to scale an equidistant chart (set to false) horizontally properly to any period in either ticks or seconds...while if equidistant is set to true, it is then easy to accurately scale charts horizontally to any period in ticks or seconds...this was using a Globex hours custom session.

Then..the following am...changing the custom session to 7am to 3:15pm Central time M/F...find one can then easily set any period in charts that are set to equidistant true or false in both seconds and ticks.

Thanks....

[NinjaTrader_JoshP]

Alfred,

What I meant for the ruler is to measure the same distance. Your rulers were measuring different distances.

[Alfred]

Thanks Josh...

Per last below...now when using a more limited custom session to 7am to 3:15pm Central time M/F ....compression and setting periods seems to be working well...for some reason or other...

....

[Alfred]

Josh...

Here is a short video look at compression rates on 2 similar charts...
Would appreciate when you have the time to review this...

http://www.screencast.com/t/MGZmZjgwNjk

Could you also look at the problem I mentioned below that changing periods in tick & seconds charts when in the overnight session (Globex) does not seem to be working when equadistant is set to false...compression and chart appearance remains about the same no matter what you do...check say 6 hours after the close on this...

Look forward to your comments...

added note...charts in the video above were set to equadistant false.

Thanks....

[Alfred]

Additional thoughts...

In tick charts where equidistant is set to false...there is a regular time scale set across all of the data loaded. A change in tick size should change that regular time scale by a proportional amount, evenly across all of the data loaded.

For example, in the video linked just below in post #9...at 44 ticks, very close to 1 hour of time was displayed across the bottom of the chart, and then when changed to a 55 tick chart, 2 & 1/2 hours of time was displayed across the chart. Going from 44 to 55 ticks is an increase of 1.25...thus the time displayed across the bottom of the chart should have been around 1 hour & 15 minutes.....but instead double that was plotted (2 & 1/2 hours)... thus distorting the horizontal scale of all chart patterns far beyond our expectations with a tick increase of just 11 ticks per bar...going from 44 to 55 ticks one would expect any horizontal pattern to reduce by 20%...however we got a reduction in scale significantly greater than that...you can review the video below and see for yourself...this is what currently can often happen when changing tick sizes and perhaps second sizes in charts where equidistant is set to false....an earlier video in post #2 below also further demonstrates this problem where a small actual change in horizontal scale can yield significant horizontal distortion to the sort of results one is expecting.

The 2 & 1/2 hour time frame mentioned above (as shown in the video) is also at an extreme chart compression that cannot normally be achieved by hitting CTRL up arrow in NT....the chart was already at max NT compression at the 1 hour time frame. You will fairly often see this extreme compression upon redoing a chart period where equadistant is set to false.

And sometimes one will see a chart that is expanded several levels from max compression...and hitting CTRL up arrow will not compress...but changing the chart period 1 or more times so the time period can be redone will get things looking OK...

.....

[NinjaTrader_Ray]

Allfred,

Josh will follow up Monday.

[Alfred]

Here are some additional thoughts to chew on maybe...

The last 2 or 3 days have been using a custom Globex session rather than the 24/7 session used before...and with this custom session have not seen any problems with horizontal distortion in charts when changing periods (so far anyway)....changing tick or seconds times seem to work OK...accuratly all the time...did not see a single problem today, and did a lot of changing at the beginning....& don't think saw any problems yesterday either...

My custom globex session cuts out all the time the market does not trade, weekends or overnight etc..... My current theory is this...when you have large blank time gaps in data where no trading occurs...does the program know how to take that into account to set the average time increments accurately for the loaded data on charts set to equidistant false....when charts are set to use regular session you do not seem to see that extreme distortion horizontally occasional when changing periods....or also have the other problems with charts being expanded and not being able to compress...or wondering what is going to turn up when you change periods...often found when I was in say a 4 second chart that when changed it actually was not in the 4 second scale for example...& then also having to change periods several times to compress etc etc...

Perhaps having relatively continuous data ...even if thin overnight stuff...allows the program to size the time increments on equadistant charts for loaded data better ??

Your thoughts ...??...Thanks...

[Alfred]

Yesterday created a custom Globex session where holiday gap yesterday is eliminated (non trading period on weekend and between globex sessions also eliminated) on charts. (this custom session setup as in right screen shot below).

A 55 tick chart (equidistant false) using this custom session looks like the screen shot on left below.

A 55 tick chart using a typical 24/7 session template accessing the empty holiday period and empty weekend periods looks like the middle screen shot below. Compression much higher here.

Ideally, don't you want the compression on both of these tick charts to be identical ?? Looks like accessing holiday and weekend (and other non trading spaces between globex sessions) empty spaces can really effect the way tick chart compression rates and chart appearance looks.

Similar seconds charts on these two sesson templates accessing holiday period or not will have identical compression rates and identical appearance.

Thanks...

[NinjaTrader_JoshP]

Alfred,

This is what happens when you change the series when you are using non-equidistant in your video. When you change the tick to 55 it will try to preserve the same number of bars being displayed on the chart. On your smaller chart, it was able to repopulate the chart with 55T charts that did not go into the overnight session thus the compression remained the same. On your larger chart, to show the same # of bars required the display of some of the overnight bars. Because the overnight bars are so sporadic in spacing, that skewed your x-axis to 2.5hrs as you have outlined. The side effect of that was the higher compression on your regular hour 55t bars.

Hopefully that makes sense.

[Alfred]

Josh...

Here are 2 identical equidistant false seconds charts ...

1) custom globex session eliminates weekends, recent holiday period and other daily nontrading Globex periods.

2) traditional 24/7 session

These two seconds charts are identical in compression and appearance though accessing different data.

On equadistant false tick charts, as in the last video discussed below, with just a small incremetal change in tick size one can see a very radical change in compression rate and appearance...that is what I still am wondering about...how can showing 20 or 25% more data on the chart (even if thin overnight data) result in such a radical change in compression rate as shown in the video ??....both of these charts in that video accessed the same amount of data for calculations...5 days back I believe...yet the compression rate is radically different, aparently because of the horizontal size of the chart ??.

When you say "Because the overnight bars are so sporadic in spacing"...
but you had a time scale running acoss those thin areas that also included the empty spaces beteen the overnight actual bars..so why such a potentially radical change in compression and time dispalyed if including both spaces and bars in equadistant false calc's ??

Thanks...