On Saturday, May 21, 2011 4:24:32 PM UTC-4, Alexana wrote:
>
> The whole concept of LaViola's paper is estimation of local linear 
> regression parameters. Of course the time series is not linear, so linear 
> regression can only be used within some local region, where some linearity 
> can be assumed. The smoothing constant alpha defines the size of the "local" 
> region. Larger alpha assumes rapidly changing function and, therefore, 
> very small "local" region. Equation 1 defines the fast EMA and Equation 2 
> the slow one. Equation 3 computes the difference (derivative). Equation 5 is 
> classic linear regression forecasting.
>
>  
Indeed, LaViola's equation (3) is calculating the difference between a 
"fast" and "slow" smoothers. That's what JBT does in a number of its 
indicators, including Tension. There is a difference, however:

LaViola: derivative = EMA(N) - (EMA(N) of (EMA(N)), i.e. the difference 
between the EMA and the double-EMA
JBT: derivative = EMA(N) - EMA(M), where N < M

I'm going to try LaViola's approach to see if it improves my backtesting 
results. Will report here.


 

-- 
You received this message because you are subscribed to the Google Groups 
"JBookTrader" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/jbooktrader?hl=en.

Reply via email to