Indicators: GARCH

 

GARCH:

This a little contribution for the wonderful community. I inspired my indicator by Edgar Peters work. It is a fractal volatily indicator based on Bollerslev Model. ARMA and ARCH models could help to understand forecast.

The indicator grows when the volatility of the market is high. The changes or variation can be predicted as well. Almost you have greater value of indicator when up or down tendency. All critics are accepted to improve the indicator.

Fractal volatily indicator based on Bollerslev Model

Author: andres.barale

 

Hi,

Thank you for sharing this indicator.

Can you build custom MTF indicator using ArrayCopy()?

 

Hi,

I should be write a new indicator with the same book and fractal behaviuor like FRASMA_V2. I try to will work on it. Thanks for the comment.

Andres Barale 

 
The beta new parameter can be changed to 0.5, the result is more clear and without unperfomance efficient. Enjoy. Thanks for downloading
 

Can you please explain the parameters and add example for BUY/SELL signals?


Thanks

 

I will write a most clear a possible explanation of de this indicator. Thanks for asking.

I in office right now, when I will come to my home I submit the explanations.

 

Hey nice indicator, just two quick questions.

 1) The GARCH Model isn't being estimated in the time-series but only forecasted with custom input of beta and alpha correct?

 2) Whats the motivation for the residual being e = this price - last period price.  Are you assuming that the best forecast is the last price?  If so, is this a supported practice that practitioners use?

 
kpei:

Hey nice indicator, just two quick questions.

 1) The GARCH Model isn't being estimated in the time-series but only forecasted with custom input of beta and alpha correct?

 2) Whats the motivation for the residual being e = this price - last period price.  Are you assuming that the best forecast is the last price?  If so, is this a supported practice that practitioners use?

 

1) alpha and beta parameters according Edgar Peters must be apha 0.10 and beta 0.80. In a lost of testing I doing I saw other alpha and beta parameters give more correct prediction.

2) No, the last price are not the best forecast. Only I calculating resiual, but with the second terms caculated have you a correct prediction beta multply be long time memory series.

I expect I'm clarifing yourself. Thanks for comment 

 

 
Indicator have looking forward error. Author not correctly ask prev. bar in this line "residual = inputData[posData] - inputData[posData - 1];" - inputData[posData - 1] is next bar and not prev. Please, fix it. Thank you! ;)
 
elab74:
Indicator have looking forward error. Author not correctly ask prev. bar in this line "residual = inputData[posData] - inputData[posData - 1];" - inputData[posData - 1] is next bar and not prev. Please, fix it. Thank you! ;)

Hi elab74

Excuse me but you wrong

the initialize os teh variable is int posData = ArraySize(inputData)-1;

and final line of de loop posData--; Then is convertion in reverse loop and - inputData[posData - 1] is the previus value

Thanks for commment.

If you use the indicator ups when before the "jump" of rpice or volatily  will happend.

See you 

 
elab74:
Indicator have looking forward error. Author not correctly ask prev. bar in this line "residual = inputData[posData] - inputData[posData - 1];" - inputData[posData - 1] is next bar and not prev. Please, fix it. Thank you! ;)

elab74: I transalate with google transalator your comment in Russian. If you that functionality; all was thinked. Put the parameter reverse in 1 insted of 0 as defulat a the code works alone exactly as you wanted. But I prefer to predict working like as it works with this parameter in 0.

See you 

Reason: