- Joined
- 11/20/14
- Messages
- 3
- Points
- 11
I know that futures are linear derivatives, thus the VaR of futures contract can be easily derived from the spot.
However, for commodities, it has a a strong implication: the stability of the convenience yield.
I was wondering if it was possible to compute a parametric VaR directly from the payoff equation from Helyette Geman (2004):
\(V_{p}(t)=e^{-r(T-t)}(F^{T}(t)-F^{T}(0))\)
F is the price of the future, \(V_{p}\) is the market value (corresponding to the P&L), T is the maturity and t a period.
I was thinking to compute the expected value for the period corresponding to the time to maturity.
Thus, I was thinking to compute an expected value corresponding to the time-to-maturity (T-t) at the initial period t as following:
\(E(V_{p}(i))=\frac{1}{T-t}\sum^{T-t}_{i=t}e^{-r(T-i)}(E(F^{T}(i))-F^{T}(t))\)
Then, calculating the variance-covariance matrix between maturities.
After, I could compute my parametric VaR with normal or gamma distributed returns.
However, I was considering that it could require a lot of data and thus I was making things harder from nothing.
What is your point of view?
However, for commodities, it has a a strong implication: the stability of the convenience yield.
I was wondering if it was possible to compute a parametric VaR directly from the payoff equation from Helyette Geman (2004):
\(V_{p}(t)=e^{-r(T-t)}(F^{T}(t)-F^{T}(0))\)
F is the price of the future, \(V_{p}\) is the market value (corresponding to the P&L), T is the maturity and t a period.
I was thinking to compute the expected value for the period corresponding to the time to maturity.
Thus, I was thinking to compute an expected value corresponding to the time-to-maturity (T-t) at the initial period t as following:
\(E(V_{p}(i))=\frac{1}{T-t}\sum^{T-t}_{i=t}e^{-r(T-i)}(E(F^{T}(i))-F^{T}(t))\)
Then, calculating the variance-covariance matrix between maturities.
After, I could compute my parametric VaR with normal or gamma distributed returns.
However, I was considering that it could require a lot of data and thus I was making things harder from nothing.
What is your point of view?
