A question about making money in the FOREX market - page 11
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
Right. Particularly the normality requirements. How will you rely on the same error if it is only the average temperature across the floor?
In my opinion, in the first place we should refuse to optimise EAs at all.
Secondly, we should be skeptical of strategy tester (as mentioned earlier).
Third, trading in DC is just training or strategy testing in conditions close to reality, but that is not real.
Fourthly, it is better to deal with brokers, which have access to the exchange (here I did not reveal America).
No problem. Test and basta. No ACF, no forecast. If there is ACF, there may be an SB with drift, we can fiddle around. And if it's just ACF, then the market is predictable and if it fails, then we just don't know how.
What I like here is that we have a model whose properties are known to us only theoretically and an unknown market, a small fraction of whose properties this model is supposed to exploit. Whether these properties exist or not is not known. The nature of market prices is also not entirely clear. There are too many uncertainties. Let us replace the SB market - now all its characteristics are known to us in advance. Let us test the model on it - yes, the model does not behave the way it is required by the SB - and there is only one explanation, it is the model itself, the inaccuracy or error can only be there and nowhere else. We correct it. We see the result. It corresponds to the requirements of the SB - good, the model corresponds at least to itself. Run the model on real rows. Compare results: there is a difference that must be corrected. Analyze the difference, see how you can earn on it, improve the model in the direction of increasing the difference between the SB and the real rows.
I am not denying TA, and I am not engaging in chit-chat. I was only talking about the applicability of specific TA and even more broadly, any
And keep in mind that the very concept of "prediction" has different interpretations ;)
I do not deny TA, and I do not engage in idle chatter. I stated only the applicability of particular TA and even in a wider range, any model that attempts to predict the market. If a model makes forecasts where they are impossible in principle, why should we believe the model where forecasts are possible but not guaranteed?
that's exactly chatter. Stream of consciousness:
1. where and why is the prediction fundamentally impossible? What is the basis for that conclusion?
2. In all stochastic systems, prediction is possible, but not guaranteed. They are not deterministic systems!
Don't play with the minutiae and don't go into the general science - it's not yours. Concentrate on what's important - so some TA methods are applicable and some are useless? Which ones?
Any predictive models in the market are useless? Then what should be done if not forecasting?
A specific question - what should a model do in the market to make a profit if it does not forecast?
The normality requirement is your requirement, and there is no requirement at all.
You are mistaken. Ask yourself the question: How does it follow that a prediction is "fundamentally impossible"? The answer is that it only follows from prejudice. "It can't be, because it can never be" ;)
And keep in mind that the very notion of "prediction" has different interpretations ;)
It follows from the fact that we purposely prepare data where a prediction is fundamentally impossible. We go to a casino, record 1,000 ball rolls - we've prepared the data. If zeros or anything else is falling out more often than it could be, then there's only one reason - the roulette wheel. Either it's crooked, or it's a fraud by the founders. We look for our ideal wheel - we find it, we try to make a prediction on it - it doesn't work. We set our system on this wheel, if it sits down to play, it means that it sees patterns where there are none, which means that what it sees is not unique and characteristic not only for some non-random process, but for everything that can be represented as a graph, and therefore it cannot be something that we can identify and use and therefore make money.
This follows from the fact that we purposely prepare data where prediction is fundamentally impossible. We go to the casino, record 1,000 ball rolls - we've prepared the data. If zero or anything else is falling out more often than it could be, then there is only one reason for it - the roulette wheel. Either it's crooked, or it's a fraud by the founders. We look for our ideal wheel - we find it, we try to make a prediction on it - it doesn't work. We set our system on this wheel, if it sits down to play, it means that it sees a pattern where there is none, which means that what it sees is not unique and characteristic not only for a non-random process, but for everything that can be represented as a graph, and therefore it cannot be something that we can identify and use and therefore make money.
What does a casino have to do with...? Such a comparison is wrong.
I'm not comparing markets with casinos. I'm not saying that markets = casinos, in fact I know they don't. But if it is not true, then we need methods capable of proving it, identifying the difference between the two and building a trading model that generates profit on the basis of that difference. If the supposed profit making method cannot even distinguish casinos from non-casinos, how is it going to make a profit? Where is the guarantee that he won't mistakenly sit down at the roulette table instead of at the money printing press?
By the way, you promised to help me get Hearst up to speed. It's been a while. I'm just afraid I will have to go to some econometricians, but I don't want to - I find it hard to communicate with them and very few people are interested in the subject.