Machine learning in trading: theory, models, practice and algo-trading - page 2821
![MQL5 - Language of trade strategies built-in the MetaTrader 5 client terminal](https://c.mql5.com/i/registerlandings/logo-2.png)
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
that's the probability of a cluster, geometrically interpreted.
you've been told you're delusional and don't know what you're talking about. No one has changed since then.
Well.... and you equate this geometric probability of proximity, which is in clusters, to the normal probability, which is in hmm, and say that they work the same.
Curious article.
Translation of abstract
This paper compares the forecasting accuracy of neural networks and conditional heteroscedastic models such as ARCH, GARCH, GARCH-M, TGARCH, EGARCH and
IGARCH, for forecasting a range of exchange rates.Multilayer perseptron networks (MLP) and the
radial basis function (RBF) networks with different architectures and conditional
heteroscedastic models are used to forecast five time series of exchange rate. The results show
that both neural network and conditional heteroscedastic models can be effectively used for forecasting
for forecasting. RBF networks perform significantly better than MLP networks in the neural
network case study. IGARCH and TGARCH perform better than other conditional heteroscedastic
models. The performance of neural networks
in predicting the exchange rate is better than conditional heteroscedasticity models. It is shown that the neural network can be effectively
used for estimation of conditional volatility of exchange rate series and implied volatility of options N
volatility of NIFTY options. It is found that the neural network outperforms conditional heteroscedastic
models in out-of-sample forecasting.
The advantage of arche-like is the minimum number of parameters, probably in relation to the number of weights neurons have. RBF has fewer weights than mlp too. Though that's how you count it.
Archie models non-stationarity, and in quite a bit of detail.
MO models, probably neuroncs as well, exploit the idea of "history repeats itself" by looking for patterns.
Does the article imply that the pattern-finding path is more promising than modelling non-stationarity?
archi modelling nonstationarity, and in quite a lot of detail.
MO models, probably neuronics as well, exploit the idea of "history repeats itself" by looking for patterns.
Does the article imply that the pattern-finding pathway is more promising than modelling non-stationarity?
Modelling non-stationarity implies modelling volatility, as I understand it. Without directional trades. In this respect, patterns or shifting average increments are more promising for directional trading. Haven't looked at the article yet.
I agree.
In our terminals trades sign. What is volatility is not clear at all.
But if one is forecasting the absolute value of an asset, that's another matter. Volatility is risk, which is crucial in predicting the value of an asset.
Probably something like that.
So will forget about the garchas.
Modelling non-stationarity implies modelling volatility, as I understand it. Without directional trades. In this respect, patterns or shifting average increments are more promising for directional trading. Haven't looked at the article yet.
I don't think that now is the time to determine what is better, to determine the absence of a signal in chaos by comparing the environment with chaos, or the presence of a signal by distinguishing it from the chaos. Neither is it better to predict or determine state. It's a time of experimentation.