-
Notifications
You must be signed in to change notification settings - Fork 1
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
1 changed file
with
205 additions
and
10 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
091cbeb
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for these additional details.
I think I haven't been clear. You do need to replicate Figure 1. of SS - this is a basic test that you are able to correctly replicate the stylized properties of their model.
Repeating the exercise for the FP implementation at least confirms that when the FP reduces to the SS model you are able to replicate the SS properties as represented in that figure.
For the two simulated sample paths:
091cbeb
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Some facts need to be clarify for the Schwartz and Smith model and the polynomial diffusion model:
$$x_t = c + E x_{t-1} + w_t. $$
$$\log{(T_{t,T})} = d_t + F_t x_t + v_t, $$
$$F_{t,T} = H(x_t)^\top e^{(T-t)G} \vec{p} + e_t. $$ $F_{t,T}$ is the futures price at time $t$ with maturity $T$ , and $x_t$ is the hidden state vector. $c, E, d_t, F_t, G, \vec{p}$ are some matrices only depend on the parameters. As you can see, one models the logarithm of futures price, but the other one models the futures price directly. Therefore, they are actually different models.
For both models, the state equation is exactly the same,
However, for the Schwartz Smith model, the measurement equation is
while for the polynomial diffusion model the measurement equation is
Here
Now as for you questions about the simulated sample paths:
Additionally, I noticed that users may also be misled by this plot. Therefore, I just deleted it.
091cbeb
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It would help matters to have equation numbers, and expressions for the futures prices.
Do I understand correctly:
If the above is accurate then you need to think of a way to prove the implementation of the FP model is correct.
If you can't get a general reduction of the SS expressions from the FP expressions, you may have to settle for parameter values for which the the two models coincide at a point(s) - think of a line intersecting with a circle.
If that is not even possible then I've run out of shortcuts to help you and you'll need to write the unit and integration test suites that are also referred to in #6.
Here I assume that you replicate Fig 1 from SS.
Has anyone else implemented the FP model? Can you validate your implementation against theirs.