Does any of the TA work on tick charts? - page 6

 
ULAD:
I agree. Although I don't use 33 at all in my TS.
By ZigZag I did not mean an indicator, but a way of forming BPs for analysis.
 
Mathemat:

There is no continuity there, Sergei. Three years ago I posted one micro-study of tics. What were the conclusions?

1. If tics followed each other at equal time intervals, everything would be tip-top: the distribution of "amplitudes" of tics itself is almost stationary.

2. All non-stationarity of bar (candlestick) returns comes from a highly unequal distribution of lags between ticks.

3. I haven't tried to build tickframe indictors, but it's very likely that many of them would look very different than on barframes. But the tickframe plots lack the most important information - the time between arrivals of neighbouring ticks.


1. I am well aware that ticks do not go with the accuracy of a quartz oscillator, which is what breaks many mats, but does not override them. The ticks also need to be prepared filtered, for later use.

2. a tick has a time. the distance on the time axis between them is known. Bars hide this information. They distort the tick flow. It is they (bars) that arrive at equal intervals, which the market doesn't. It only seems to us that the time between bars is constant (time interval).

3. When building an indicator by Clowes, we censor the sample. We censor every 100 ticks (right or wrong, this is another matter). We do not manage the censoring. Once again, when selecting a slope to build, we select a tick that arrived at the end of a minute (hour). But its exact position on the time axis is unknown to us - the bar hides this information from us. In DSP theory it is called the sampling noise.

4. Sampling noise can be dealt with, but it's the worst noise. They try to avoid it by all means from quartz stabilization and cooling to attomic clocks. There's no such thing on the market here. We work with what we've got. If you need equal time intervals for mathematical calculations, having two points on the time axis, you can always put a third point between them, where you need it (for example Simpson's method).

 
hrenfx:

...

P.S. I'm sure almost no one will agree with me.

Why, I agree. And absolutely. You need a stream, and you can slice a lot out of it, you can use zigzags. you can use Renko charts, equivolume charts, etc. you can even use a multi-currency bar (like when the dollar index changes).
 
Mathemat, Prival thanks for the links. while I take time out to study primary sources....
 
Prival:

2. a tick has a time. the distance on the time axis between them is known. Bars hide this information. distort the tick flow. It is them (bars) that appear at equal intervals, which is not present in the market. It only seems to us that the time between bars is constant (time interval).

The tick lifetime is very short, and the information in one tick is too small for analysis. Of course, if we examine a certain period of ticks, we will get much more information, like on any other timeframe.

By the way. Equal periods of time can also be calculated from a zero bar, I often use such a solution in my indicators.

 
hrenfx:

The time between ticks makes no difference.

Whether the market came in 100 ticks in a second or a minute makes no difference.

Can volatility be measured without time interval?
 
Andrei01:
Can volatility be measured without regard to time horizon?
Yes, we can. It has been discussed here.
 
hrenfx:
You can. It has been discussed here.

Well, what's the point? Doesn't it matter what the total amplitude of currency fluctuations per day or per year is?

the same analogy can be applied to individual ticks.

 
Andrei01:

Well, what is the point? Doesn't it matter what the total amplitude of currency fluctuations in a day or a year is?

Taken from the same conversation:

The idea is simple. Look at the results. For example, the file Amounts_1.00%.txt:

Interval: 2009.10.20 - 2010.09.03
Price MinChange = 1.00%

01 SILVER amount = 664
02 NZDJPY amount = 389
03 AUDJPY amount = 369
04 USDPLN amount = 343
05 USDHUF amount = 335
06 CADJPY amount = 289
07 GBPJPY amount = 265
08 GBPZAR amount = 261
09 USDZAR amount = 257
10 GOLD amount = 241

The name of the file tells us that it shows the number (Amount) of ZigZag knees with a min knee condition of 1%.

Yes, this is not the amount of knees, just the number of knees. But it was possible to assess this way in this case, as the results are presented at once on different knee sizes.

So you see that SILVER changed at least 1% 664 times in the studied interval (2009.10.20 - 2010.09.03). So much for volatility.

Say that the number of knees is not an indicator, as NZDJPY could have changed by 2%, not 1%. And you will be right, but not in this context. As the results are given for other knee sizes as well.

It is absolutely correct to count volatility as I wrote above - sum of knees.

 
hrenfx:
So you see that SILVER changed at least 1% 664 times during the interval studied (2009.10.20 - 2010.09.03). So much for volatility.
so is the time interval present after all?