Differential of traders wealth function equals $0$. Correct?

32 Views Asked by At

Calculating the differential of the trader's wealth function I get $dV_t \equiv 0$, which not only surprises me but also stops me from going forward with the current financial model I am looking at (I modified Kyle's model with information asymmetries).

Because I am already stuck over this for a while, I wanted to ask if I maybe did any calculation mistake. Simplified (without the information asymmetries which would only make it longer but not change the content) my financial model is the following:

Market makers use the signalling process $$dY_t = a(Y_t)dX_t + b(Y_t)dt,$$ where $X_t = W_t + \theta_t$ is the cumulative demand of the noise traders ($W_t$ is a standard Brownian motion) and the individual trader (see below), and $a$ and $b$ are weight and drift functions respectively, in combination with $h$, a twice differentiable, strictly increasing function, to price a risky asset via $h(Y_t)$. In equilibrium, the price process is supposed to be a martingale. The risk neutral trader has a cumulative demand of $$\theta_t = \int_0^t \alpha_s ds$$ where $\alpha$ is adapted to the traders filtration. Based on that, the trader's wealth at time $t$ can be described via $$V_t = \int_0^t \theta_s dh(Y_s),$$ given that $V_0=0$ holds. If I now look at the differential $dV_t$ via Itô's lemma and use the definition of $\theta$, I get the following: \begin{align} dV_t &= \theta_t dh(Y_t) \\ &=\theta_t (h^\prime dY_t + \frac{1}{2}h^{\prime\prime}d\langle Y \rangle_t ) \\ &= \theta_t(h^\prime(adW_t + ad\theta_t + bdt) + \frac{1}{2}h^{\prime \prime} a^2 dt) \\ &= \alpha_t dt (h^\prime(adW_t + a \alpha_t dt + bdt) + \frac{1}{2}h^{\prime \prime} a^2 dt) \\ &= \alpha_t(h^\prime(adtdW_t + a \alpha_t dtdt + bdtdt) + \frac{1}{2}h^{\prime \prime} a^2 dtdt) \\ &= 0 \end{align} where the last step follows from $dtdt = dtdW_t = 0$. Did I miss anything in my calculation? Thinking about it from a financial point of view, I interpret this result as no growth in the traders wealth which seems odd to me.

Already big thanks for any help and let me know if you need more information!

1

There are 1 best solutions below

0
On BEST ANSWER

The mistake you make is replacing $\theta_t$ by $\alpha_t dt$ on line four and then trying to treat resultant terms like $dt dW_t$ as being the same as $d \langle \operatorname{Id},W \rangle_t$. To see this clearly, note that the same reasoning, if correct, would say that for a Brownian motion $W$, $$ \int_0^t W_s ds = 0$$ since $W_s ds = 1 dW_s ds$. This is obviously false.

To see what's happened, it will help to write things out in integral notation. A term like $\theta_t h^\prime(Y_t) a(Y_t) dW_t$ really means $$\int_0^t \theta_s h^\prime(Y_s) a(Y_s) dW_s = \int_0^t \int_0^s \alpha_u h'(Y_s) a(Y_s)du dW_s.$$ Unfortunately, this integral does not cleanly split in to the product of two integrals and so we can not just cleanly take the covariation and apply the fact that $\langle \operatorname{Id},W \rangle = 0$. To treat e.g. $dtdW_t$ like $d \langle \operatorname{Id}, W \rangle_t$ you need $dtdW_t$ to mean a product of the two integrals and not an iterated integral like it does here.