Discussion of article "Econometric Approach to Analysis of Charts"
Isn't the GARCH Model suppose to describe volatility clusters in the time-series? If so, there's not much of an approach here to model time-series returns in USDJPY but rather volatility. If best, you can include how to base a strategy off of these econometric modelling.
It was a decent article. I enjoyed it a lot. I want to predict whether a trend is going to start or no on a currency pair in a specific time frame, say H1. For this purpose, I first get the returns within a time frame of length, say, the past N "H1 candles", and then use the Q test. If it passes the Q test, then I fit the parameters of a GARCH(1,1) model on the obtained returns from the chosen time window, and then calculate the expected value of the predicted variance for the next H1 candle. If it is above a specific threshold, then we can expect that a trend is coming.
But based on your experience, do you think such a method has a good accuracy in practice?
But based on your experience, do you think such a method has a good accuracy in practice?
Hadi Hadizadeh:
It was a decent article. I enjoyed it a lot. I want to predict whether a trend is going to start or no on a currency pair in a specific time frame, say H1. For this purpose, I first get the returns within a time frame of length, say, the past N "H1 candles", and then use the Q test. If it passes the Q test, then I fit the parameters of a GARCH(1,1) model on the obtained returns from the chosen time window, and then calculate the expected value of the predicted variance for the next H1 candle. If it is above a specific threshold, then we can expect that a trend is coming.
But based on your experience, do you think such a method has a good accuracy in practice?
It was a decent article. I enjoyed it a lot. I want to predict whether a trend is going to start or no on a currency pair in a specific time frame, say H1. For this purpose, I first get the returns within a time frame of length, say, the past N "H1 candles", and then use the Q test. If it passes the Q test, then I fit the parameters of a GARCH(1,1) model on the obtained returns from the chosen time window, and then calculate the expected value of the predicted variance for the next H1 candle. If it is above a specific threshold, then we can expect that a trend is coming.
But based on your experience, do you think such a method has a good accuracy in practice?
Thanks for your opinion. The model does not predict the trend or flat inception. It rather allows to define the bounds for future returns. And the 2nd opportunity - to simulate future returns (prices) within the validated bounds.
Thank you Denis for your valuable response. So, the model predicts the "bounds" for the future returns not the "sign" of the future returns. Is it somehow possible to predict the sign of the future returns by means of other complementary statistical models?
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
New article Econometric Approach to Analysis of Charts is published:
This article describes the econometric methods of analysis, the autocorrelation analysis and the analysis of conditional variance in particular. What is the benefit of the approach described here? Use of the non-linear GARCH models allows representing the analyzed series formally from the mathematical point of view and creating a forecast for a specified number of steps.
Author: Dennis Kirichenko