1
$\begingroup$

If we suppose that r(t) follows a Vasicek model, which is: $$dr(t) = (\mu - \kappa r(t))dt + \sqrt\sigma dW(t)$$ How to derive an expression for Eurodollar future rate?

$\endgroup$
1
  • $\begingroup$ If $\sigma$ is a constant, then the use of $\sqrt{\sigma}$ does not appear sensible. $\endgroup$
    – Gordon
    Commented Dec 14, 2018 at 16:47

1 Answer 1

2
$\begingroup$

there are many ways to solve Vasicek system, for me personally I markov short rate approach. Without going into the details of proofs:

Note that eurodollar future is calculated under risk neutral Q measure of libor rate at each settlement $t_{fix}$ (on three months interval each)

libor rate $l(t_{fix}) = \frac{1}{tenor} e^{A_{diff} - B_{diff} * r(t_{fix})} $ where $A_{diff} = A(t_0 - t_{fix}) - A(t_1 - t_{fix}) $ and similarly to $B_{diff}$. $t_0$ and $t_fix$ is starting time of current eurodollar settlement and $t_1$ is its time of maturity (eg, $t_0 = 0.25, t_{fix} = 0.25, t_1 = 0.5$)

A and B are calculated using the following system:

$\frac{dB}{dt} = kB - 1$

$\frac{dA}{dt} = \mu B - \frac{1}{2} \sigma B^2 $

denote $ A_{diff} - B_{diff} * r(t_{fix})$ = $\psi$ (just for typing purpose)

Thus it became clear that to calculate eurodollar rate = $E^Q(l(t_{fix}) = \frac{1}{tenor} e^{E(\psi) + \frac{1}{2}\sigma(\psi)^2}$, you already have A and B and just need mean and variance of short rate process, which are very straightforward under defined vasicek model.

mean: $r_0e^{-kt_{fix} + \mu (\frac{1-e^-t_{fix}}{k})} $

variance: $\frac{\sigma}{2k} (1-e^{-2k t_{fix}})$

also don't forget to -1 inside the expectation after rate

$\endgroup$
3
  • $\begingroup$ Hi, when I do the parameter fit for the dataset on a cross sectional basis, which is a particular day, I find that the fit and the actual data is very close. Is this collinearity and how to improve that in this case? $\endgroup$
    – Qing
    Commented Dec 14, 2018 at 20:57
  • $\begingroup$ Hi Qing, I assumed your actual data is different eurodollar settlements, so there you go the fitted curve is supposed to be close to the actual data. $\endgroup$
    – numerairX
    Commented Dec 14, 2018 at 21:15
  • $\begingroup$ May I ask how to do the fit process for Affine model to fit the Eurodollar future rate curve? $\endgroup$
    – Qing
    Commented Dec 16, 2018 at 2:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.