Chart Scaling Question...

[NinjaTrader_JoshP]

Alfred,

Please consider this. Hopefully it explains it better. When you change from 44T to 55T, it will try to show the same absolute number of bars. If on the 44T you were showing 100 bars, when you switch to 55T it will show 100 bars. As you switch to 55T, lets say only 80 of those bars are regular session bars and the remaining 20 is overnight bars. Those overnight bars are spread out across an extended period of time span of say 2 hours. Now the chart needs to show an x-axis of those 2 hours + maybe 30 minutes for the regular session bars. So with 2.5hrs time axis evenly spaced out means your regular session bars will be compressed significantly more than your smaller chart which maybe only had 50 bars to show and is able to still show those 50 bars as regular session bars without adjusting the timeline.

[Alfred]

Josh...

I remain unconvinced by your argument...on say a 44 tick chart (equadistant set to false)...5 days of data loaded...the time scale increments set by Ninja remain the same dimension graphically over the whole 5 days...accessing all sorts of thin areas in 24/7 data...a relatively small change in tick size (up or down) should change that time scale over a 5 day period in a proportional incremental small amount...then one should see an even proportional incremental change in horizontal compression and general chart pattern apparance whatever the actual horizontal size of the chart...

....

[NinjaTrader_JoshP]

Alfred,

Your small chart, say is 5 inches long shows 50 visible bars.
Your large chart, say is 10 inches long shows 100 visible bars.

As you switch the small chart to 55T it will try to show 50 bars as well. It is able to show those 50 bars with regular session bars and not need to add any time axis adjustments.

As you switch the large chart to 55T it will try to show 100 bars as well. It is not able to show 100 regular session bars since only 80 exists. The remaining 20 bars that are needed to be shown is from thin overnight data. Since you are on non-equidistant, this means adding significant time axis adjustments to the same 10 inches resulting in an even more compressed 80 bar display on the right edge. This new compression level is used for the whole chart.

You are trying to correlate a 20% increase in tick bar size with some sort of change on the chart. This is not how it currently works. It uses the number of bars preexisting on the visible chart and tries to persist that number rolling into the new chart.

[Alfred]

Josh..

Where "It uses the number of bars preexisting on the visible chart and tries to persist that number rolling into the new chart."

Does a bar in your definition above include the empty spaces as well ?


Thanks...

[NinjaTrader_JoshP]

Not sure what you mean by empty space? A bar is literally one OHLC line or candlestick bar that has price information.

[Alfred]

Josh...an empty space would be the empty spaces between thinly traded bars showing price info, in say typical overnight areas...are these individual empty spaces also counted similarly in addition to the bars displaying price info when changing tick bar periods and deciding on how to plot a new chart ??

...appreciate your patience ! ...thanks...

[NinjaTrader_JoshP]

Empty space is not counted as a bar. As your chart dips into overnight data those empty spaces are needed to determine the x-axis scale, but is not used in determining anything about the number of visible bars to display.

[Alfred]

Count empty bars ...

Josh...

Below 2 screen shots with custom sessions...the one on left excludes yesterdays holiday period...the one on right includes the holiday period.

You can see how by not counting empty bars in chart construction it can dramatically skew the compression and time displayed in a chart.

If you counted empty bars you would get that smooth transition in compression and time displayed when changing periods or creating new charts that I discussed below.

The way you have it now...using absolute bars...things are just all over the map when one changes chart periods or creates a new chart...you often never know what is going to come up the way a revised chart will appear....You just have to wonder why it was done this way in Ninja ??

You can know the number of empty bars per time increment...

How about an option in Ninja properties to include the empty bar count in chart construction, when changing periods etc...for those who care about the way their charts appear....??


Thanks...

[NinjaTrader_JoshP]

Alfred,

Thank you for your comments. The reason it is done this way is because you do not know the amount of bars that would fall into that empty slot and you can't just assume it is 1 bar. It could be 5 bars or 10 bars depending on how volatile the market is when using chart types that are not time based.

[Alfred]

Josh...so with historical data you cannot count the empty bars between bars that have data in them ??...don't you have a fixed incremental time scale for the loaded data running across that data that shows bars and empty bars in a graphic horizontal scale on equidistant false charts...just curious how this works...thanks...

[NinjaTrader_JoshP]

Alfred,

There is no concept of empty bars between bars. There is only bar A and bar B. There is nothing inbetween A and B even if the time gap between A and B is say 1 hour. The time scale goes by absolute time and not by # of bars. If there is 100 bars that would fit within 1 inch of space then it will show 100 bars in there. If there would be 1000 bars in that 1 inch, then it will show 1000.

[Alfred]

Josh...

What is your view on the extreme time scale difference one can currently sometimes get when resizing a chart period ??

What about resizing on an approximate absolute time that would match as close as possible the approximate % increase or decrease in chart scale one expects on resizing seconds or tick charts ??...like pick the closest absolute bar to match the % change in horizontal visual scale expected as best as possible....would that be a possibility ??...

Thanks for the info....

[NinjaTrader_JoshP]

Alfred,

Thank you for the suggestion. It has been added to our list of future considerations.

[Alfred]

Here is another problem you can run into with the way horizontal chart scaling is set up now...below are 2 identically sized charts...one a 1 second chart..and below it a 2 second chart...they are both set to max compression, yet they look identical...this can sometimes happen perhaps after you let charts run awhile and re-size one of them I believe. You often just do not really know what you are looking at in terms of chart compression....screen shot below...

In equidistant false charts if you perhaps think of chart scaling in pixel slots rather that absolute bars...there will be empty pixel slots where no data occurs in a time period equal to a typical bar space slot that has data in it....why not count the empty slots along with the slots that have data in them...something like that...using absolute bars, as you have described it, to get the same number of bars in a chart when changing periods just does not seem at all accurate in general currently...


Thanks...

[NinjaTrader_JoshP]

Alfred,

Complications arise because a single pixel space can hold multiple bars depending on chart compression.