Dependency statistics in quotes (information theory, correlation and other feature selection methods) - page 32
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
You are aware of the influence of volatility clustering on the lower timeframes, and yet you draw any conclusions that supposedly the higher timeframes are more noisy. The only basis for this is your belief. If you really want to compare different time scales then compare them not directly, but their residuals from the effects of volatility, otherwise it all resembles a belief of your own.
The volatility clustering effect mentioned is there on all timeframes, and on higher ones too. I don't know which TFs are more noisy...
And the residuals from volatility effects are also small on all timeframes and it is problematic to choose the best TF in this regard. But there is one "But". On daily charts there is no total autocorrelation of neighboring lags by mutual information, and we can benefit from it by removing mutually informative predictors from the model (if we want to predict volatility). On hourly bars, for example, this is almost impossible to do, because there neighbouring bars correlate more strongly with each other than with the zero bar.
faa1947:
It is econometrics that justifies a forecast one step ahead and shows the reasons for fatal deterioration of the forecast several steps ahead.
And what does econometrics say about forecasts of similar length?
The prediction is fulfilled (the moment of prediction is marked with "x") 17 steps ahead. What does econometrics even say about the forecasting methods adopted in Adverse Tactics?
Has it not investigated the mathematical apparatus and geometric primitives of TA? Or is predicting more than one step ahead impossible because it is impossible?
NASDAQ Daily
Another example is the 41 step forward forecast:
Nikkei 225 Quarterly
Forecast a few hundred steps ahead:
USDCHF Monthly
Long-term predictions on such a horizon are the lot of geniuses, charlatans and idiots.
If you are a genius, where is your PAMM or the article about you in Forbes?
Long-term predictions on such a horizon are the lot of geniuses, charlatans and idiots.
Short-term, please:
GAZP 5 min
Please short-term:
GAZP 5 min
That's not how it's done!
You make today's forecast for tomorrow with a specific trading signal (sell or buy). Publish it today. Check it tomorrow. And so for ten days. At the end of it, we sum up the trades.
We have even created a special branch for you.
Dots, are you still alive or what? Let's cut the crap about patterns giving them mysticism, get to their basics. Returns of increment sizes and their combinations .
In particular, how did you come up with these exact forms of patterns.
That's not how it's done!
Make today's forecast for tomorrow with a specific trading signal (sell or buy). Publish it today. Tomorrow you check it. And so it goes on for ten days. At the end of it, you get a summary of trades.
If you have not yet understood, I am not here to convince you of anything.
I'm here to talk about current scientific thinking in the field of forecasting. About trends, perspectives...
As for you, this is my last answer. If it's interesting, sort it out, not so;)
In case you haven't realised by now, I'm not here to convince you of anything.
I am here to talk about current scientific thinking in the field of prognosis. About trends, perspectives...
As for you, this is my last response. Interesting, deal with it, not so;)
1. Why did you draw pictures then?
2. Econometrics does not prohibit making forecasts for any number of steps forward - the quality of forecasts deteriorates.
3. Econometrics is a scientific discipline within the framework of which one can make forecasts and mathematically evaluate them. Tactics adverse - something that allows you to make a forecast, but has no apparatus for mathematically assessing its quality.
4. What is there in current scientific thinking in the field of forecasting? I only know correlation and regression analysis, NS, expert methods. What else is there?