Classical thechanalysis doesn't work any more. What works, maybe quantum? - page 19

 

If Papa Carlo carved a mechanical Pinocchio with a knife from a log (and his friend Giuseppe started doing it with an axe), and Pinocchio missed out on five gold pieces, it does not mean that the Kolomna wooden palace cannot be built with an axe and a knife (and even 300 years before that):


 
timbo >> :

For any series the characteristics will be unreliable, the only question is the degree of unreliability.

Not for any one. For a stationary series everything is very well estimated.

timbo >> :

Markowitz's strength is not in his formulas, which simply cannot be true due to his simplistic approach, but in the ideas behind them.

No doubt, it's a good idea, but I don't have the right data.

timbo >> :

One of the ideas is that TA is impossible.

There is no such idea there.

FOXXXi >> :

Unsteady random time series can be described by a formula but that does not mean it can be predicted. To say that there is nothing to substitute in reality is complete nonsense.

You apparently didn't really get into the gist of the previous discussion.

 
HideYourRichess >> :

Not for any one. For a stationary series, everything is very well estimated.

Even for a stationary series the parameters are only estimated with some precision, i.e. there is always a certain degree of unreliability.

By the way, I do not see any problem in estimation of parameters of non-stationary series, another question is that knowledge of these parameters will not help to get profit. Example:

x(t+1) =x(t) +e(t), where e(t) ~N(0,1). I know the parameters of this process with 100% certainty, I set them myself, and there is no money to be made on it.

The idea that TA is impossible comes from the theory of efficient markets, which is behind CAPM. If you work in that paradigm, you acknowledge it a priori.

You think, as I understand it, that prices are non-stationary series, so you're in the same boat - no indicator can predict non-stationary series, so all TA (for you) is tambourine jiggling.

 
timbo >> :

Even for a stationary series the parameters are only estimated with some precision, i.e. a certain degree of uncertainty is always present.


Naturally. What is called "accuracy" is the cornerstone of terrain, after all. But "credibility" is different. Roughly speaking, it is when accuracy cannot be estimated. Under steady-state conditions, a range of values converges to its true limit as the number of measurements increases. This is also the reason why accuracy can always be estimated. But with non-stationary series nothing converges anywhere, and generally it is not known what happens there, hence the unreliability.


timbo >> :


By the way, I don't see any problem in estimating parameters of non-stationary series, another question is that knowledge of these parameters will not help to get profit. Example:

x(t+1) =x(t) +e(t), where e(t) ~N(0,1). I know the parameters of this process with 100% certainty, I set them myself, and there is no money to be made on it.

The idea that TA is impossible comes from the theory of efficient markets, which is behind CAPM. If you work in that paradigm, you acknowledge it a priori.

You think, as I understand it, that prices are non-stationary series, so you're in the same boat - no indicator can predict non-stationary series, so all TA (for you) is tambourine jiggling.

I only objected to the list of laureates. see p.12

 
Svinozavr >> :

And the fact that you are posting to a person who uses this "inoperative" TA and has had no other source of income since 2004 than to use this "inoperative" TA in the stock market - does that not occur to you?

Read Taleb's "Fooled by chance", we think - there are many similar examples.

 
timbo >> :

We read Taleb's "Fooled by Randomness", we think - there are a lot of similar examples.

We read Erasmus of Rotterdam's "A Praise for Foolishness", and we think it's all about that.

However, do not get upset - I'm not able to do a lot of things too. Although I do not claim that no one is able to.

Get rid of complexes - there will be less aggression and more fun from life.

 
timbo пис ал(а) >>

Even for a stationary series the parameters are only estimated with some precision, i.e. a certain degree of unreliability is always present.

By the way, I do not see any problem in estimating the parameters of a non-stationary series, another matter is that knowledge of these parameters will not help to make a profit. Example:

x(t+1) =x(t) +e(t), where e(t) ~N(0,1). I know the parameters of this process with 100% certainty, I set them myself, and there is no money to be made on it.

The idea that TA is impossible comes from the theory of efficient markets, which is behind CAPM. If you work in that paradigm, you acknowledge it a priori.

You believe, as I understand it, that prices are non-stationary series, so you're in the same boat - no indicator can predict a non-stationary series, and therefore all TA (for you) is tambourine-playing.

Predicting is not necessary if a credible model is available. You can get by just fine with the real one.
 
begemot61 >> :
You don't have to predict at all if you have a credible model. You can make do with real one.

Completely agree about the present.

Here's a valid model of some process: x(t+1) =x(t) + e(t), where e(t) ~N(0,1). How to make money on it?

 
timbo писал(а) >>

- No indicator can predict a non-stationary series, which means that all TA (for you) is a tambourine-playing game.

Not recognizing BP as non-stationary will not change and it will remain as a dynamical non-stationary process. The strength of TA is that it doesn't go into statistical characteristics of BP, but follows another way - the patterns: there is a crossover of two MAs - there is a signal. On the basis of patterns one can build a profitable trading system and there are a lot of examples (see Champion). But the forum is full of specialists (like Prutkov's flux) who know the mathematics of stationary processes, they convert nonstationary process to stationary one and build TS for stationary model, forgetting that this system should be applied to non-stationary process, not to its stationary model.

 
Svinozavr >> :

We read Erasmus Rotterdam's Praise of Folly and think it's all about that.

However, don't get upset - I can't do many things either. Although I don't claim to be able to do it.

That's exactly what you're doing: if you have a profit for 5 years, it means TA works. If the period was 55 years, it would be possible to talk about the importance of this fact, but it's an accident, on the basis of which you're making global conclusions.