Range bar vertical gaps ?

[RJay]

Here is how I see it. I have to agree with supertrader. QT has it right.

On a 4 range chart, once a bar completes a 4 range spread, a new bar should begin where the last bar left off.

Lost tick values on the ES mini "trading range charts" ( NT vs QT )

8 trades a day = $ (200.00)
5 days a week = $ (1,000.00)
50 weeks a year = $ (50,000.00)

It shouldn't be this way.

Could you please have support look in to this.

Thanks,

RJay

[MJT]

RJay,
I am sure every range bar trader has their own view, but mine is that the current implementation is best. I want a new bar only when the Range is exceeded, not when it is reached.
To address your concern about lost ticks, could you not enter/exit a trade when the bar reaches it's full range, without waiting for the new bar to form?

[RJay]

RJay,
I am sure every range bar trader has their own view, but mine is that the current implementation is best. I want a new bar only when the Range is exceeded, not when it is reached.
To address your concern about lost ticks, could you not enter/exit a trade when the bar reaches it's full range, without waiting for the new bar to form?

MJT,

That would be nice except how do I break the cycle. Every bar opens a tick late and begins its 4 count from there (Range 4) . I would have to create a completely new range chart to trade with if things stay as they are right know.

It would also be nice to have both. then everyone would be happy.

I also believe the current AltRangeBars Chart Just covers up the gaps once they have been created. ie Close[1] does not equal Open[0].

I am curious, not looking for specifics but how are you able to trade around the tick penalty on the range bar charts?

RJay

[MJT]

RJay,
I agree, the chart will be different, and yes it would be nice to have both options available. The RangeAlt bar type was provided to appease those who disliked the fact that Range bars always fill in any price gaps (apart from the 1 tick between bars of course). A new custom bar type could be made to perform the way you want, but that is outside of the boundaries of NinjaTrader's normal support.

I don't see the 1 tick gap as a penalty - it is just the way the chart is constructed. My system as back tested uses this, and the event that triggers a trade is the open of a new bar. I don't want to enter at the close of the prior bar, as I don't consider the trigger to occur until price goes one tick past that.

I hope that helps, MJT

[VTtrader]

Why would there ever be a new bar, if the opening tick is the same as the closing/previous tick? By definition it has to be a tick lower/higher to start a new bar. Otherwise, it would be within the range of the previous bar, and no new bar would form. As for a 1 tick penalty, that seems more of a question in backtesting or on historical bars and not a real world event.

Just my opinion,
VT

[RJay]

Why would there ever be a new bar, if the opening tick is the same as the closing/previous tick? By definition it has to be a tick lower/higher to start a new bar. Otherwise, it would be within the range of the previous bar, and no new bar would form. As for a 1 tick penalty, that seems more of a question in backtesting or on historical bars and not a real world event.

Just my opinion,
VT

VT,


My thoughts are that this would be true if the markets stayed inside the same 4 tick range all day(Range=4). It would also be true for the current Range charts that you are now using.

Thanks for the feedback.

RJay

-----------------------------------------------------------------------


NT,

I've had some time to think about this. The indicators I'm using also update at bar close. This means all my indicators are also one tick behind the market as well.

Also, since most indicators are built off the OHLC bar values, How is the tick value of the gap between each bar suppose to be calculated into each indicator?

Am I losing my mind!!!

Can you please explain to me why I am wrong about the current Range Chart configuration.

Thanks,

RJay

[NinjaTrader_Bertrand]

Hi RJay, your indicators using the 'Close' do not differ from the behavior seen on other chart types...when the bar closes the OnBarUpdate() is called which updates your indicator values (with CalculateOnBarClose set to true). You might be able to get an 'edge' by setting this to false and entering early on the developing bar when the set range is nearly fullfilled...but of course this can create some whipsaws along the way.

[VTtrader]

RJay,
What I was saying is this. A 4 tick range bar has 4 ticks high to low. Let's say ticks for the bar are (low-high) 867, 867.25, 867.50, 867.75, the bar is now fully formed.
If it stays within that range it will never start a new bar, and Close[0] will show the last traded price, but the actual close will not be "set" since the bar hasn't closed. In order for a new bar to print that range will have have to be exceeded by 1 tick (866.75 or 868).
As soon as it trades at either one of those prices a new bar is triggered and the open price will be that first tick at either 866.75 or 868, and the close of the previous bar will be the low or high of the previous bar, either 867.75 or 867, 1 tick away.

The new bar may or may not trade up to the previous close, but the open price will be 1 tick away. To do it any other way would require arbitrarily cutting off ticks from the bar1 and say they belong to bar2, or by creating a tick that doesn't exist. IOW, if you were waiting for the close of bar1 to trigger a trade at the open of bar2, and you made a range bar that used the closing price of bar1 as the open of bar2, that open price tick doesn't exist in the real world, and you won't get executed at that price unless bar2 retraces by that amount...sometimes it will, but often it wont. As opposed to NT having a 1 tick penalty, doing it the other way would give overly optimistic results.

BTW, the picture early in this thread from QT actually look more like renko bars rather than true range bars.

Hope this clarifies what I was saying.
VT

[RJay]

VT,

Thank you for your latest response.

First, I would like to say that I like Range bars becouse they plot price action in both axis. I think this is a plus.

If someone can prove to me that I'm wrong about the level of profitibility between the two types of Range bars in question, I would most definfitely want to hear about it.

I believe everyone is entiltled to their viewpoints on any subject. I always try to listen. Its how I learn.

I also tend to think a little outside the box in respect to "conventional wisdom". This means I'm wrong alot until I find the "Truth".

I also seem to have difficulty clearly expressing my thoughts.

I have an image in my mind of what I want to achieve here and the current range bars are not it.

I don't agree with some of your other observations in your latest post but that will make me do more research.

Please do not take any of my remarks the wrong way.

I'm just a little fustrated. The level of programming needed to create what I want is beyond my meager programming skills.

RJay

[NinjaTrader_Bertrand]

RJay, we appreciate the input and critical thinking from our users. You can always post a suggestion here and our development team will consider those for futures releases - http://www.ninjatrader-support2.com/...splay.php?f=11

[VTtrader]

Rjay,
Don't worry, I'm not taking any of your remarks the wrong way, and I hope you're not taking any of mine the wrong way either. I was just trying to help with your frustration.

That said, let me approach this from a different angle. First of all, any chart is simply a graphical representation of the stream of market data being fed to it. As a result, any "bar" whether it's minute, weekly, range, tick, volume etc. is an arbitrary break in a contiguous stream of data. The only exception to this is a daily bar, due to the fact that there is a defined break in the data at close, and a defined open the following day.

In order to break this never-ending stream of data into bars, there needs to be an event that "triggers" a new bar. There are time based triggers (minute/second charts), and data based triggers (tick/range/volume charts). Of the second group, in tick and volume charts, the triggers are "close based". IOW, the required number of ticks or volume has occurred, so the bar closes and the software knows the next tick will be a new bar. However, for range bars the trigger is "open based", the software doesn't know in advance whether the next tick is a trigger for a new bar or not until it arrives and is evaluated. As a result, it doesn't know that bar1 has closed until bar2 has already started trading.

So what does all this mean? It's important to realize/remember a couple of things. First, your trades become part of that contiguous data stream, and "bars" are just a convenience to aid in our decision making process. Second, there has to be at least 1 trade/tick at any given price to have even a remote chance of getting filled. Also, backtesting strategies and analyzing charts etc., as valuable as that can be, is at best, an approximation of how your trades will be handled in the real-world data stream.

Here's an example:
each tick is 1 point
using 4 tick range bars.
Bar1 is blue
Bar2 is red
Bar3 is black

the data comes in like this:
11 11 11 12 11 12 12 12 11 12 13 13 12 13 12 12 11 11 12 11 10 10 10 11
bar1 has now reached it's full range of 4 ticks but is still open....
11 10 11 10 10 10 9 8 8 8 8 8 7 7 7 8 8 7 7 6 7 6 6 5 5 6 6 7 6 6 5 5 4 4

You can see from this data that once the 4 tick range was exceeded it triggered a new bar and the open/first tick was 9, and there was never a trade in bar2 that was the same as bar1. Further more, the software never knew to close bar1 until the first tick at 9 so it just uses the low of the range as the close. To create an "open" at the same price as the close, would be a distortion of the data.

I've seen a different implementation of range bars that triggers a new bar when the full range has been reached. It then closes the bar and opens the new one using the same price. That implementation has the same flaws, it creates a tick that doesn't exist in the real world. Say we used this implementation on the data above and a 5 tick range bar. It would trigger a new bar when it traded at 9 and set the close of bar1 and the open of bar2 both at 9, but with only 1 trade at 9 is that the best representation of the data? What would happen if it never traded at 9 at all and went straight to 8?

Having the new bar triggered on an open based event, makes range bars particularly prone to data distortion since the previous bar is closed only after the market has already traded at the next bars open. Tick and volume charts aren't subject to the same data distortion problem, since the close of the bar is known before the first trade of the new bar.

Hope this helps.
VT

[NinjaTrader_JoshP]

NinjaTrader does not know closing of a bar till the open of the next bar. This has always been the case and this is because NT is bar driven.

Your different approach for building a range bar has been noted and been added to our list of future considerations already.

[VTtrader]

Josh,
I'm curious, I'm not sure if you were saying that to me. I'm not advocating for a different approach to building range bars (certainly not to replace the current one), I think the current way is the best/most accurate method.

I do wonder though, if you were saying that tick and volume bars were also "closed" at the open of the next bar, as opposed to on the last tick/contract of the current bar?
It's not that big of a deal, since it seems that it would be just the point at which the last tick was processed, but it's good to know if that's the case.
Kind of makes that "CalculateOnBarClose" a bit of a misnomer though

VT

[NinjaTrader_JoshP]

I will have to confirm with development.

[RJay]

NinjaTrader_Josh,

I'm the one Who would like to see an additional version of Range bars added to the platform.

(No gaps between bars Close[1] = Open[0])

Who should communicate with about this request?

RJay