A model developed by JP Morgan?

[Pyng]

Hi guys,

I am quite confused with this equation:

σ²_t = βσ²_t-1 + (1-β)ε²_t

If we have the residual error (Actual Value - Predicted Value) at time t, that means we already have known the actual variance at time t. Then why do we still need to forecast the volatility at time t anyway?

[thinkwiseandact]

I haven’t seen a regression equation with a)error term squared, b)a coefficient. May be this is not an error term, it is an independent variable? Maybe is should have been (1-β)ε²_t-1?

Is this model public?

[Pyng]

This equation is shown on page 27 Reading 16 CFA Book 3 2019.

[thinkwiseandact]

After reading the page you mentioned, I think it is not a residual. It is an independent variable.

[Pyng]

Isn’t it stated as the noise term at time t?

[rgk]

This isn’t a regression at all, it’s just a weighted average. Don’t let the notation trip you up. All they’re doing here is smoothing out volatility with an exponential noise factor ε.

[Pyng]

@rgk

This isn’t a regression at all, it’s just a weighted average. Don’t let the notation trip you up. All they’re doing here is smoothing out volatility with an exponential noise factor ε.

How does the noise factor derive from?