Discussion of article "Combinatorics and probability for trading (Part IV): Bernoulli Logic"
Insider, will give volatility of oil, before the news about oil reserves from API. That is, the reasons and probabilities of price movement occurred before the price itself was formed.
That's kind of both a criticism and a question.
Insider, will give volatility of oil, before the news about oil reserves from API. That is, the reasons and probabilities of price movement occurred before the price itself was formed.
That's kind of both a criticism and a question.
Well, I agree, but it depends on how you approach trading. If we talk about automated trading, an insider can hardly explain anything to a robot. Either you analyse the root cause or its effect. Both approaches have the right to exist. An Expert Advisor will need to operate with some simpler data. And take your API for example, what if it dies? And in addition, not everyone knows what and how. Here the idea is a little different - to take the minimum of data and get the maximum out of it.
Good night, Eugene!
Glad to see another article of yours.
I need to clarify the question about the plurality of states, namely, on the example of flipping the same coin:
1. Flip with the stamped side;
2. Falling out with the "tails" side;
3. Falling out on the edge.
For the first case - this is buying.
For the second - selling.
For the third - smoking and not trading.
In any case we have only buying or selling.
We keep silent about the trend, because we use the probability of falling out of the state, and then we can take it as it comes, i.e. "Taki-profit" or stumbled upon "Moose".
Question:
- What else can there be here, in terms of the multiplicity of states?
We keep silent about the trend, because we use the probability of falling out of the state and then we can take it as it comes, i.e. "Taki-profit" or stumbled upon "Moose".
Question:
- What else can there be here, in terms of the multiplicity of states?
Hi Alexander, there could be many possibilities here. You have three events, from them you can make sets, for example ( Herb, Tails ) ( Herb, rib ) ( Herb, Herb ) ( Tails , Tails ) ( Tails , rib ) ( Tails , Herb ) ( rib , Tails ) ( rib , rib ) ( rib, rib ) ( rib , rib ) , rib )( rib , Herb ) ( rib , Herb )
That is, instead of considering a single event, you can consider chains made up of these events, and you can get completely unexpected samples from them.. The number of such sets will be counted as follows:
- Pow(n,N)
- N is the length of the chain (the number of tosses that follow each other).
- n - number of possible states after tossing the coin ( their probabilities form a complete group).
In other words, the number of outcomes of tossing is multiplied by the length of the chain. On the example of the double chain I got 9 sets, let's check!
- 3-number of tossups
- 2-the length of the chain
- 3^2 = 9 - it all adds up.
And you can consider chains of 3 ... 4 .... + infinity of states, depending on how long your sample is, the more experiments you make with tossing, the more you will get the squeeze for analysis. Sets can also be combined if you are interested in probabilities of some group of these sets, as all these sets form a complete group of events.
Hi Alexander, there can be many variants here. You have three events, from them you can make sets, for example ( Herb, Tails ) ( Herb, rib ) ( Herb, Herb ) ( Tails , Tails ) ( Tails , rib ) ( Tails , Herb ) ( rib , Tails ) ( rib, Tails ) ( rib, Tails ) ( rib , Herb ). , rib )( rib , Herb ) ( rib , Herb )
That is, instead of considering a single event, you can consider chains made up of these events, and you can get completely unexpected samples from them.. The number of such sets will be counted as follows:
- Pow(n,N)
- N is the length of the chain (the number of tosses that follow each other).
- n - number of possible states after tossing the coin ( their probabilities form a complete group )
In other words, the number of outcomes of tossing is multiplied by the length of the chain. On the example of the double chain I got 9 sets, let's check!
- 3-number of tossups
- 2-the length of the chain
- 3^2 = 9 - it all adds up.
And you can consider chains of 3 ... 4 .... + infinity of states, depending on how long your sample is, the more experiments you make with tossing, the more you will get the squeeze for analysis. Sets can also be combined if you are interested in probabilities of some group of these sets, as all these sets form a complete group of events.
Thank you very much, Eugene, for your answer!
Good...
But I, for example, have only 200 tosses per year in my trader's time line, i.e. one toss before the beginning of a trading day.
Accordingly, there is only one event in the set.
By experience, I would like to point out that constantly changing the direction of trades has a very bad effect on statistics and financial results.
I think that solving the problem with many sets is very redundant and is a theoretical extension without practical application.
Let's get closer to practice!
OK?
Thank you very much, Eugene, for your answer!
Good...
But I, for example, have only 200 tosses per year in the trader's time line, i.e. one toss before the beginning of the trading day.
Accordingly, there is only one event in the set.
By experience, I would like to point out that constantly changing the direction of trades has a very bad effect on statistics and financial results.
I think that solving the problem with many sets is very redundant and is a theoretical extension without practical application.
Let's get closer to practice!
OK?
Let's get practical. "200" tosses for example. If we analyse this whole sequence of trials, we can distinguish not single tosses, but for example different chains with different sets of states. In trading, if we do not analyse chains of trades but the price, they are called patterns. Any pattern can be represented with sufficient accuracy by a chain of states. It is interesting that when we consider a single state or just a step, we will get chaos most likely, but as soon as these states are combined into a chain, a pattern is formed and this pattern can speak about both buying and selling, all you need to do is to analyse what happens after the pattern and make statistics. Backtest or trading history is also a curve and patterns can be searched not only at the price level but also at the virtual trading level. I will describe this later in another article, there is just a lot of material and it should appear in due time.
And so in general it's good that you are trying to dig further, it's good to see).
We keep silent about the trend, because we use the probability of falling out of the state and then we can take it as it comes, i.e. "Taki-profit" or stumbled upon "Moose".
Question:
- What else can there be here, in terms of the multiplicity of states?
To be precise, the taki edge fallout implies falling on the edge connecting the plane of the eagle and the plane of the tails. So, there appears another variant - a real falling on the edge, when the coin stands slightly tilted.
To be more precise, the falling out of the edge is the falling out on the edge connecting the plane of the eagle and the plane of the tails. So, there is another variant - a real falling on the edge, when the coin stands slightly tilted.
In continuation of this "logic" one should definitely consider hanging in the air.
And in general, if the connection of the edge and the plane of the eagle/tails is also a plane contributing to the stable position, then anything can be considered (let's not forget about hanging).
Let's get to practice. "200" tosses, for example. If we analyse this entire sequence of trials, we can identify not single tosses, but, for example, different chains with different sets of states. In trading, if we do not analyse chains of trades but the price, they are called patterns. Any pattern can be represented with sufficient accuracy by a chain of states. It is interesting that when we consider a single state or just a step, we will get chaos most likely, but as soon as these states are combined into a chain, a pattern is formed and this pattern can speak about both buying and selling, all you need to do is to analyse what happens after the pattern and make statistics. Backtest or trading history is also a curve and patterns can be searched not only at the price level but also at the virtual trading level. I will describe this later in another article, there is just a lot of material and it should appear in due time.
And so in general it's good that you are trying to dig further, it's good to see).
"Interesting that when considering a single state or just a step, then we get chaos most likely ..."
- this is where we need to stop.
Chaos or turbulence in the market occurs very rarely once 5-7 years and it is expressed in a sharp flight or influx,
which affects the rapid growth, which then sharply deflates, or a panic fall in the value of a financial instrument.
Therefore, you can consider even simply and without price patterns, which are a great number, and which do not always give the direction that is expected of them.
Is not it true, Eugene?
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
New article Combinatorics and probability for trading (Part IV): Bernoulli Logic has been published:
In this article, I decided to highlight the well-known Bernoulli scheme and to show how it can be used to describe trading-related data arrays. All this will then be used to create a self-adapting trading system. We will also look for a more generic algorithm, a special case of which is the Bernoulli formula, and will find an application for it.
If we consider the analysis of the possibilities of describing trading history and backtests in the language of mathematics, first we need to understand the purpose and possible outcome of such analysis. Is there any added value in such an analysis? In fact, it is impossible to give a clear answer right away. But there is an answer, which can gradually lead to simple and working solutions. However, we should delve into more details first. Given the experience of previous articles, I was interested in the following questions:
The answers to all these questions are as follows. It is possible to reduce some strategies to fractal description. I have developed this algorithm and I will describe it further. It is suitable for other purposes as well, as it is a universal fractal. Now, let's think and try to answer the following question: What is the trading history in the language of random numbers and probability theory? The answer is simple: it is a set of isolated entities or vectors, the occurrence of which in a certain period of time has a certain probability and the time utilization factor. The main characteristic of each such entity is the probability of its occurrence. The time utilization factor is an auxiliary value that helps determine how much of the available time is being used for trading. The following figure may assist in understanding the idea:
Author: Evgeniy Ilin