You need an indicator that reflects the price in operating time. - page 3
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I agree that if it were NZR, then maybe +3 sko would cover the required value with probability 0.97. If non-normal, it would simply be less. I think there is a probability of 0.7 for any distribution law that has mode and variance. If you need, I can give you some calculations, a consequence of Chebyshev's theorem.
Is it worth using ticks at all? Sadly enough, Renat is right, using ticks is a dead end. There is too much unnecessary noise caused by errors in the Internet, as well as by different filters in the chain of obtaining quotes, etc. Although in my opinion, bars are not suitable for autotrading, especially for older periods, because they were invented long before computers and are intended for human perception. In my opinion, the most suitable period (of those available) is a minute one. On ticks your "scope" will work like this:
On ticks, your "scope" will work like this:
Prival, here are the preliminary results of the history calculations for bars with constant volume: the catastrophes have not disappeared, but the statistics, roughly speaking, have remained the same. Fat tails remain, so does the high peak. However price will be plotted differently, it is certain: some equivolume bars corresponding to H4 will hold just over half an hour (usually it is a strong movement but not necessarily) and some others will hold over 16 hours (a calm market). Can you smell what this smells like?
I haven't looked at the Hi-Lo ("high minus low" i.e. swing) stats yet, it could be interesting and different from the baseline for normal bars.
The illusions have become one less...
P.S. I'll post the code in a few days, when I check the calculations and learn how to draw candlesticks.
Prival, here are the preliminary results of the history calculations for bars with constant volume: the catastrophes have not disappeared, but the statistics, roughly speaking, have remained the same. Fat tails remain, so does the high peak. However price will be plotted differently, this is certain: some equivolume bars corresponding to H4 will hold just over half an hour (these are usually strong movements but not necessary) and some others will hold over 16 hours (a calm market). Can you smell what this smells like?
I haven't looked at the Hi-Lo ("high minus low" i.e. swing) stats yet, it could be interesting and different from the baseline for normal bars.
The illusions have become one less...
P.S. I'll post the code in a few days, when I check the calculations and learn how to draw candlesticks.
I've been smelling for a while now, don't make bars (see my sight picture) +-3sko (the tail should get thinner)
The thing is, a scope is not enough, there has to be a clever noise filter:) The number of ticks is not always the data... The cyclic pass filter helps in this regard, but again it all depends on the settings. My experience tells me that no type of bar is not data, as it can contain as much as noise, but a lot of other things that are not at all easy to pass through this kind of filter, in the case of bars, the filter will only add unknowns. You could say that bars are a primitive filter. However, if you watch ticks, again you will easily notice everything I am talking about.
However, primitive does not mean bad, if you understand what bars are, you will hardly understand a filter not made by you, on the contrary, it will play against you.
Cycle filter
A little more detail, what is it ?
For example, you will display a cycle of repetitions on a tick chart, up and down, etc., how to filter out this cycle, for example by enclosing it in a formed range, with coordinates of the upper and lower price, in this case you will get the average price or the price where the filtering started. If the next tick breaks your amplitude terms, then this tick is drawn as a unit.
This method can be applied to a strict sequence as well as an approximate one. It looks a bit like a scope, but that's not all...
The range of amplitude can be in one unit of price or many more.
What is noise, it is when the price jumps and falls out of the amplitude, here we can focus on the constancy of the jumps and it is also to some extent the amplitude created by the market conditions, but there is what I call funnels, when some development of amplitude movements can be recognized, funnel formation is like the weather conditions of storm formation, where many such events are laid down and the most difficult is not how to filter out the noise but how to categorize them.
P.S.: The nature is very similar to the market and I am struggling with controlling the nature of the market, learning to recognize storms, their direction and use them:) Again, many people mistake this for trading on noise, the final data on many storms and their success in predicting them, gives me confidence in controlling the market, and therefore confidence in my actions. Unfortunately the primitiveness of thinking does not allow some to look beyond their noses, which is very sad, it is harder to use this for good as we get a lot of sticks in the wheel, which however we also learn to use.
The reality on the market is that there are noises that no one will let you diagnose right away, because they are created specifically for processing and methods are definitely improving, at data provider level and I prefer to keep the efficiency of my methods a secret and I advise you, it's like a game of poker here :) One can and must, but for everyone there is a different way. All know what abuse leads to, like trading on noise, plays against you again, because the broker does not need you, and use, what I'm talking about for the greater good few are going to and for this reason, we stand in place and people prove me wrong, although to convince me of this is not possible, because I've already received some benefits from it, no going back, as they say... But I also can't agree that it doesn't find its application, because my own comfort and convenience in that direction depends on it.
P.S.: As pointed out more than once Renat, you'll understand everything in your own experience, but how far you'll go, no one knows but you, so go for it :)
I agree that if it were NZR, then maybe +3 sko would cover the required value with probability 0.97. If non-normal, it will simply be less. Pomoyama with a probability 0.7 for any distribution law that has maybe and variance. If you need, I can give you some calculations, a consequence of Chebyshev's theorem
I attach the proof, only my error is not 0.7 but 8/9=0.88(9). Double-check, I did it myself once, the proof is the simplest one.
I found it on the web, so there's no mistake.http://cito-web.yspu.yar.ru/link1/metod/theory/node21.html
To get a valid result, we need to work on a lot of data, for example, in multicurrency analysis it becomes immediately in perspective and as a consequence in the relativity of approaches more easy to catch noise and therefore sift them out, but do I need to sift them out, I consider them but so their speciation that it becomes simple, that is instead of a lot of data I process their bases, thus many ticks are not lost in the filters but converted into indicators, in order to get tick indicators I need to apply many filters
P.S.: I hope you get some useful information from this:)