Has anyone wondered why and how much the trend building "walks"? - page 2

 

TheXpert:

But just reproduced a 20-point shift on the clock when manually plotting.

How would I be able to reproduce that?
 

It is likely to be visual (or inaccurate in plotting the line on the graph), the actual values are unchanged.

Files:
 
poruchik:

The TF gap is caused by the fact that a candle of a high TF consists of several candles of a low TF.

In case you haven't noticed, builds and scrolls are performed on the same timeframe. There is no TF switching.
sanyooooook:

Holes in the history of 1-2 minutes or as many as there are in a day (when the history is swapped, which happens during a TF change, there may be a shift)

No TF change, no holes, no history swap.

ZFS: the reasons are several and cannot recollect all, it is strange that only noticed)

Only two -- TF switching and history holes. Neither one fits...

ZZZY: in this case most likely inaccuracy of construction when changing the vertical scale

20 pips? That's hilarious.

sergeev:
how would I be able to reproduce it?

I don't know. I'll think of something.
 

Just two -- TF switching and holes in the history. Neither one fits...

At 20 pips? That's hilarious.



This is only visually, an inaccurate display, in fact the coordinates of each line point are unchanged when you change the scale (vertical), which is what happens when you scroll through the graph.

ZS: this often happens when the coordinates of the starting points are outside the chart boundaries

ZZZY: the scale is 1:1 ) and there are no shifts ))

 

inaccuracy of display + - 20 pips, on H4.

SZY: we draw a trend line called L, we draw a vertical line called T, when we change the vertical scale, in principle also the horizontal one, the visual intersection and the actual one differ by +- 20 pips, if the coordinates of the points of construction are outside the chart.

ZZZY: If the trend has been built a long time ago, in fact we may get trapped by 20 pips.

UZZYZY: Spreads were invented and here we are also screwed by 20 pips

Files:
 
sanyooooook:

This is only visually, an inaccurate display,

And when you do the plots you don't look at them visually afterwards?

In fact, the coordinates of each line point are unchanged when you change the scale (vertical), which is what happens when you scroll through the graph.

And how do you explain the shift of the crossover point by several bars if the coordinates are unchanged?
 
TheXpert:
And when you make constructions, do you not look at them visually afterwards? And how do you explain shifting the intersection point by several bars if the coordinates are unchanged?
The graph is made up of several layers, apparently there is a mismatch somewhere at this level.
 

Here's an example of the plotting.

Let's assume that the accuracy of the reference points is 1 pip. Then at the intersection point the difference can be, at the most, 3-4 pips (taking into account the ratio of the lengths of the segments). But the difference is 11.

IBM clock tool.

here is an approximation one

here's two

I don't know if it's important or not, the anchor points are both outside the window.

Preliminary diagnosis -- bug in edge point calculation to display trend lines if anchor points are outside the screen.

 
TheXpert:
And how do you explain shifting the intersection point by several bars if the coordinates are unchanged?

the layers are not correctly scaled relative to each other.

 
Hee-hee, remove the OHLC and everything seems to be falling into place