Drawing lines when using logarithmic scale

[NinjaTrader_JoshP]

terryt,

Can you provide me with step by step instructions with exactly how you achieve it? Steps from the very very beginning would be useful. All the way back from File>New Chart. I have been trying to chase this for awhile now and cannot get it to reproduce on my end. Thank you.

Attached is ES weekly chart and the lines look parallel to me.

[terryt]

Josh:

I hope you can reproduce it from this.

- NT7.0.0.8
- new chart AAPL, weekly, load 1825 days from eSignal,
use instrument settings, no template,
instrument uses US Equities RTH
(the problem is probably not sensitive
to some of the data series settings above)
- right click over the price scale, choose properties
- change to logarithmic, click OK.
- [snap mode on my system is on "Bar". Is that the default? )
- F3 (to draw a ray)
- 1st pt (click on weekly high during Feb 09)
- 2nd pt (click on weekly high during Oct 09)
[btw. all time high during jan 10 is not snapable]
- select ray with a left click at the beginning of august 09
- ctrl-c, then ctrl-v

The resulting ray is not pasted directly on top of the source ray.
Instead it is pasted below the source ray.
When you drag the new ray on top of the source ray,
they do not overlap perfectly. The slopes are different.


I did some playing after setting snap mode to "Disabled".
Sometimes I saw a problem sometimes I did not.
But I did get consistent results.

I drew 5 rays to use as source rays.
Then I individually copied and pasted them.
The slope mismatch varied with the angle of source ray.

slope ~0/1: result appears have same slope
slope ~1/2: result has larger slope
slope ~1/1: result has larger slope (most pronounced visually)
slope ~2/1: result has slightly larger slope (least pronounced)
slope ~1/0: tough to select vertical line. appears to have same slope.

Hope this helps.
-Terry-

[NinjaTrader_JoshP]

Terry,

Thank you very much for the detailed instructions. I was able to reproduce.

[Tartan]

Tartan,

Your video just shows the end results, but I still do not know how you got to those results. When I tried to reproduce my lines copy as parallel lines and I have no problem with the Andrew's Pitchfork. Please clarify. Thank you.

I am not clear on what you want me to provide. Please clarify.

Also, please use a data set that include lots of data over the course of about 12 months. The data set should have doubled over that time. I think the results will be clearer if you select such a data set. You could use SPX over the past 12 months.

[Tartan]

Josh:

Are you able to replicate the fibonacci problem on log charts that I pointed out?

[NinjaTrader_JoshP]

Tartan,

Yes I was able to, but I am not sure why you feel the numbers should change? The Fibs are drawn at those exact price levels regardless of log/linear scale. If a line is at 100, it stays at 100 regardless of scale.

[Tartan]

Tartan,

Yes I was able to, but I am not sure why you feel the numbers should change? The Fibs are drawn at those exact price levels regardless of log/linear scale. If a line is at 100, it stays at 100 regardless of scale.

Josh: On a log chart, NT should be using the log of the prices to create the fib ratios. On an linear chart, NT should be using the actual prices to create the fib ratios.

[NinjaTrader_JoshP]

Tartan,

I am not sure if that would be the case. Fib numbers are calculated the same regardless of the visualization scale.

[Tartan]

Josh: I am sure that you have access to Tradestation, eSignal and other competitive products. I don't.

Can you please verify that competitive products treat Fibs on both log and linear scales in the way that you say. I believe that they behave in the manner that I mentioned:
- On a log chart, the log of the prices to create the fib ratios.
- On an linear chart, the actual prices to create the fib ratios.

Thanks!

[NinjaTrader_JoshP]

Tartan,

Had a colleague check, TradeStation has it behave the same way we have it currently.

[Tartan]

Thanks Josh. That's not the way I thought it worked. I appreciate your time.