The Ruler...discussion & suggestions...

[Alfred]

Using the Ruler drawing tool...it seems to only count absolute bars in Equidistant False charts...what can one actually use that info for ??....you can move the same ruler set-up to another location and easily get a completely different bar count and Time span with the same graphically sized ruler.

It seems that one should ideally be able to measure time and actual bar count accurately with the ruler on all charts...be able get the number of bars (including "empty bars" in Equidistant False charts)...don't you want to be able to measure time accurately...say between pivot points or the length of any graphic object ?? Some graphic techniques (Gann fan, squaring price & time etc.) require the accurate measurement of time (and price also) to set up. An easy way to measure time in bars is critical.

What about the option of inserting a dummy close in any "empty" bar space equal to the last previous close if there is no actual data that falls within that time slot ??.... That way the ruler could measure time & bars correctly as all bars would then be measurable...and also.... when plotting into the future with code perhaps one could then reference just the number of bars desired from a known start point for any object. Also...one would not need to use absolute bars to set the horizontal scale of any chart if one could measure all bar spaces accurately...this could perhaps fix the current odd problems in scaling charts horizontally in Ninja....

Thanks...

[NinjaTrader_JoshP]

Alfred,

When you are using equidistant = false there is no way to hard set the number of bars for 1 inch of ruler. A ruler of 1 inch would measure a hard set timespan, but how many bars are within that 1 inch would be variable depending on how volatile the market was during that time span.

When equidistant is false, 1 physical inch on your screen would be the same amount of time anywhere on the chart. When equidistant = true, 1 physical inch on your screen could represent 5 minutes on one point and maybe 5 hours on another point depending on what times the bars close.

Problem with inserting dummy closes is that we really don't know how many to fill in. It is possible to have 1000 bars within 1 inch or 100 bars within that inch. There is no "norm" we can say as to how many bars occupies 1 inch of space.

[Alfred]

....

Josh...at max compression it appears that a bar width of 1 takes up 1 pixel horizontally...if you have a 5 second Equidistant False OHLC chart that 5 seconds should take up 1 pixel, whether it has data in it or is "empty" ...there is a regular time scale running across the bottom of the chart...don't the "empty" 5 second bars have to fit within that time scale as well...??...if your chart screen was 200 pixels wide it would hold 200 bars at max compression (both "empty" and regular bars)... or 200 X 5 = 1000 seconds of time on that particular chart....the data is time stamped in seconds...can the data be assigned to a particular second ??...

What is your view on being able to measure time accurately in both bars and time on equadistant false charts currently...?? What can the current way of "absolute" measuring be used for since it seems to be the luck of the draw where the ruler may be placed for the values displayed...?? I personally don't have a problem with the way equadistant true charts are currently displayed and measured...no fill in would be require there....only with the equadistant false charts...one would insert a dummy close there at the end of a bar period if no actual data came in during that period...and on possible subsequent "empty" bars intill real time data resumed in a bar...

....

[NinjaTrader_JoshP]

Alfred,

Max compressed charts has no rule that says 1 bar can take only 1 px. You can have bars overlap should the compression require so 1 px can actually hold 10 or 50 bars. You would not be able to tell. I am not sure what you mean by "empty" 5 seconds bars. An empty gap of 5 seconds in the time axis can represent the space for 5 bars or 50 bars. We have no way of knowing and predicting how many bars could/would fit into such a space.

I would not know of any possible way you can create a ruler on equidistant = false charts to show a ruler that has the same number of bars and time at the same time.

[Alfred]

Josh...measuring a chart window...where there are few if any empty bars the max compression at CTRL Up Arrow seems to be 1 pixel on Equidistant False OHLC seconds charts...what conditions would require a greater compression than that ??....

NT has gone to the trouble of creating Equidistant False seconds charts with a regular time scale but so far is not able to measure accurately what has been created...just thinking a bit on how that might be accomplished...

What would be wrong with inserting a dummy close in Equidistant False seconds (not tick ??) charts "empty bars" even if the compression were greater or less than than 1 pixel per bar ??...one could then measure time and bars accurately regardless of the compression...

Thanks...

[NinjaTrader_JoshP]

Alfred,

The issue is 1px does not mean 1 bar. 1px could mean 1 bar or 100 bars. There is no way a ruler can just say 1px=1 bar. When you are using equidistant=false the only thing the ruler can say for certain within 1 physical inch is the amount of time being represented within that 1 inch. Number of bars is irrelevant because on a liquid period in time it could be 1000 bars while on an illiquid time it could be 100 bars. Filling in "empty" bars means you are assuming that 1 inch can only hold x amount of bars when this is simply not true. Bars can and do overlap in the same space should their timestamps require it to so my point is because of that behavior we can't say that 1px=1bar.

[Alfred]

Josh...my idea in this area is something like this...say you have a 5 second Equidistant False chart running....the last bar on chart has just finished it's 5 second period and a close forms...then the next 5 second period starts....expecting some data to arrive perhaps...but no data comes in during that next 5 second period...so you then immediately create a dummy close equal to the previous bars close at the end of this latest particular 5 second period (this dummy close could be colored differently than a regular close perhaps)....this way you have equidistant false bars that can be measured accurately... a real bar count, (not just absolute bars)...even though data does not always come in for each bar period...bars per inch or the compression rate of the chart would be ignored....would something like this this be workable...??

[NinjaTrader_JoshP]

Alfred,

This would only be feasible assuming you used time based intervals on your chart. When you use tick charts, volume charts, etc. then this would not be possible. Having different ruler behavior depending on the chart's interval type may be confusing though. In either case, thank you for the suggestion, I understand you now. I have added it to our feedback list.

[Alfred]

Thanks Josh....am very interested in the development of this idea in NT...

Wanted to ask though...since Equidistant False (EF) tick charts are also time based, wouldn't this technique also work on those ??...not as if that is critical in any case...I have found the EF tick charts do not display anything that cannot be found in some variation of EF seconds charts (pivot points, chart patterns, density of bars at critical turns, chart patterns etc.)...the EF tick charts seem almost redundant to me...very seldom use them now....

This technique would also be really valuable when changing periods in EF seconds charts...one would not then depend upon absolute bars for horizontal scaling of charts....and it would not be necessary to sometimes get those haywire hyper compressed charts less than 1 pixel per bar & overlapping OHLC bars....currently in EF seconds charts, one never quite knows what will happen re horizontal scale when changing periods.

And a world of accuracy in trading techniques based in part on accurate time ...accurate bars....would open up...squaring price & time...measuring chart objects, Gann fans, length of swing projections, accurate cycles.... etc etc....Also...if one is interested in actual true fixed scaling...this would facilitate scaling in points or ticks per bar !! ....

The added dummy bar closes could also optionally be colored transparent if the user did not wish to see those on a chart....


Thanks...

[NinjaTrader_JoshP]

Alfred,

The difference between an EF tick and an EF seconds chart is as follows:
EF second = guaranteed bar every single second on a continuous stream of data

EF tick = no guaranteed bar for any given horizontal space. Could be 1 bar or 10 bars. No way to hypothesize what would fit into that space while on a EF second chart you could assume that a 1 second bar would "fit" there.

Further issues with synthetic bars though Alfred, is all indicator calculations will go off. You can't just skip over a bar in calculations. It breaks a lot of concepts in NinjaTrader.

[Alfred]

Concepts are made to be broken........

Josh...this idea could be an option in Properties or Data Series..

And....concepts are made to be broken....

....

[Alfred]

...

Josh...

Here is a video I did in another situation going over the importance of a correct bar count when setting up Gann fans for squaring of price & time on particular market events...maybe worth a look re the discussion in this thread on the desirability of a correct bar count...

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

This is just one of a number of techniques that would benefit from an accurate bar count...

Also...my general feeling is that existing NT studies may just as well benefit from the synthetic bars in EF seconds charts...things like Darvis and Donchian Channel and Regression Channel could certainly benifit...and the many typical oscillators could behave in perhaps even a more realistic way taking into account all bars...

Thanks...

[NinjaTrader_JoshP]

Alfred,

What I suggest is you use equidistant=true. Thank you for the video, but synthetic bars is something that will require a 100% overhaul of Bars logic and we will not be doing this at this stage of the beta. What I have done is added it to our list for future considerations and reviews.

[Alfred]

This is a minor consideration on the ruler...in the screen shot below the sequence of the anchor points as the ruler is plotted in stages are shown in green...1, 2, 3...

But in the ruler properties, anchor point 3 (magenta color) is the middle point, not the end point...might be a bit better to have the anchor points labeled in the sequence in which they are plotted....??

Thanks...

[NinjaTrader_JoshP]

Alfred,

I am not exactly sure how you achieved the anchor point 3 to not be attached to the ruler textbox since the 3rd point is where the textbox is added onto. Can you please clarify?