Variational approach to nonlinear pulse evolution in stock derivative markets

Abstract

The Ivancevic option pricing model is studied via variational approach. Both the $Gaussiananstz$ and the $sech$ ansatz are used, and each has a unique results from one another. But in terms of existance of soliton solutions they both agree that hot market temperatures support the existance of soliton solutions.

I Introduction

Since the discovery of the Nonlinear Schrodinger equation (NLS) like option pricing model in 2010, it has drawn much interest theoretically in various soliton phenomena in the financial world. Variuos techniques such as the generalized Kudryashov method (GKM) demiray2022new , adomian decomposition method (ADM) and spectral boundry approximation (SBA) method adama2022exact , improved physics-informed neural network (IPINN) bai2022application , the method of Lie symmetries obaidullah2023analytical , the Elsaki-Adomian decomposition method gonz2015solving , projected differential transform method (PDTM) edeki2017analytical , He’s frecuency amplitude formulation gonzalez2017solving , $q-homotopyanalysistransformmethod$ ($q$-HATM) veeresha2022analysis and modulational instability (MI) analysis  chen2022soliton  gaafele2024modulational among others raheel2023research  wroblewski2023quantum .

Variational approach (VA) has also been a versatile method for nonlinear like models for different branches of physics including optics anderson1983variational  malomed1997pulse  malomed2002variational and in Bose-Einstein condesates sabari2013variational  wamba2008variational  wamba2014dynamical . In most studies this analysis has the same results as MI and other method that are used for the NLS like models, in terms of soliton dynamics and stability. Unlike MI VA is suitable in discribing soliton pulse propagation. VA can either use $Gaussian$ ansatz nicolin2008nonlinear or $sech$ ansatz malomed2002variational among others since they are localize and assumes the shape of a soliton. $Gaussian$ ansatz with its mathematical simplicity its advantageous because of its analytical solvability and also gives a unique propagation characteristics. They tend to spread more uniformly. $sech$ ansatz are unique in VA since they give a soliton solution that is stable and can propagate for longer distances and gives better approximation to the soliton solution. $sech$ pulses naturally account for strong nonlinear effects such as self-phase modulation, self-steepening, and the formation of dispersive waves, which are crucial in nonlinear optics and pulse shaping. $Sech$ pulses are effective in managing dispersion, maintaining their shape over longer distances even in the presence of significant dispersion rusin2019variational , malomed2002variational , curilef2013tsallis , rossi2015non , lott2009numerical . In this paper both the $sech$ and the Gaussian ansatz will be used in VA in the Ivancevic option pricing model (IOPM) due to the uniqueness of the results they produce.

The IOPM is the NLS like model used in asset derivative markets alternative approach to the traditional Black-Scholes (BS) model for pricing options. Developed by Vladimir Ivancevic, this model incorporates elements of both classical mechanics and financial mathematics ivancevic2010adaptive . It uses a Hamiltonian formalism to describe the dynamics of the underlying asset price and employs a different stochastic process than the standard geometric Brownian motion used in the Black-Scholes model. The model incorporates nonlinear dynamics, which can better capture the real-world behavior of financial markets, including features such as volatility clustering and heavy tails in the distribution of returns. Ivancevic’s approach sometimes draws analogies from quantum mechanics, using path integrals to represent the probabilities of different price paths kleinert2009path . This can provide a more flexible and accurate representation of price movements over time. The Black-Scholes equation is a linear partial differential equation, while the Ivancevic model often results in nonlinear equations due to its more complex stochastic processes. The IOPM can easily incorporate quantum neural computations ivancevic2009quantum .

The remaining section are as follows: Section II will focus on the Ivancevic option pricing model. Section III will focus on the variational calculation of the model using both the $sech$ and the $Gaussian$ ansatz. Section V shows direct simulation of the ansatz for different conditions. Section IV gives the discussion of the results and the paper is concluded by Section VI.

II The Ivancevic Option Pricing Model

The IOPM is a quantum probability equation given by

$$i\frac{\partial\psi}{\partial t}+\frac{1}{2}\sigma^{2}\frac{\partial^{2}\psi}{\partial s^{2}}+\beta|\psi|^{2}\psi=0,$$ (1)

where $\psi(s,t)$ is the probability distribution function (PDF) for the option price at time $t$ for the stock/asset price $s$, $\sigma$ is the volatility, $\beta$ is the Landau coefficient which is the adaptive market heat potential. Both $\sigma$ and $\beta$ are assumed to be constants. The adaptive market heat potential can be related to the market temperature and it obeys the Boltzmann distribution. It depends on the set of adjustable parameters $\{w_{i}\}$ which are refereed to as synaptic weight vector components ivancevic2010adaptive .

III A variational formulation of the IOPM

III.1 Gaussian Ansatz

It can be shown that from the IOPM Eq.(1) that the Lagrangian density is given by

$$\mathcal{L}=\frac{i}{2}\left(\psi\frac{\partial\psi^{*}}{\partial t}-\psi^{*}\frac{\partial\psi}{\partial t}\right)+\frac{1}{2}\sigma^{2}\left|\frac{\partial\psi}{\partial s}\right|^{2}-\frac{1}{2}\beta|\psi|^{4}.$$ (2)

We assume the Gaussian ansatz

$$\psi(s,t)=A(t)\exp\left(-\frac{s^{2}}{2w^{2}(t)}+ib(t)s^{2}\right).$$ (3)

By substituting Eq.(3) into Eq.(2) and using $L=\int\mathcal{L}ds$ we get the effective lagrangian $L_{eff}$ from the dispersion and interaction terms

$$L_{eff}=\frac{1}{4}\sigma^{2}A^{2}w\sqrt{\pi}+2\sigma^{2}b^{2}A^{2}w^{3}\sqrt{\pi}-\frac{\beta}{4\sqrt{2}}A^{4}w\sqrt{\pi}.$$ (4)

which will be used to derive the equation of motions for the variational parameters.

III.2 Variational equations

The equations of motions are derived using the Euler-Lagrange equation of motion $\frac{d}{dt}\frac{\partial L}{\partial\dot{q}_{i}}-\frac{\partial L}{\partial q_{i}}$.

For $q=w$ we get

$$\frac{\partial L_{eff}}{\partial w}=\frac{\sigma^{2}A^{2}\sqrt{\pi}}{4}+6\sigma^{2}b^{2}A^{2}w^{2}\sqrt{\pi}-\frac{\beta}{4\sqrt{2}}A^{4}\sqrt{\pi},$$ (5)

using Newton’s second law of motion, $\frac{\partial^{2}w}{\partial t^{2}}=-\frac{\partial v_{eff}}{\partial w}$we have

$$\frac{\partial^{2}w}{\partial t^{2}}=\frac{\sigma^{2}A^{2}}{2}w+6\sigma^{2}b^{2}A^{2}w^{2}-\frac{\beta}{2\sqrt{2}}A^{4}w,$$ (6)

integrating we get

$$v_{eff}=-\frac{\sigma^{2}A^{2}}{4}w^{2}-2\sigma^{2}b^{2}A^{2}w^{3}+\frac{\beta A^{4}}{4\sqrt{2}}w^{2},$$ (7)

we can plug the efficient potential in the energy equation to get.

$$E=-\frac{1}{2}\left(\frac{dw}{dt}\right)^{2}-\frac{\sigma^{2}A^{2}}{4}w^{2}-2\sigma^{2}b^{2}A^{2}w^{3}+\frac{\beta A^{4}}{4\sqrt{2}}w^{2},$$ (8)

which is anologous to that of a particle moving in a potential well.

For $q=b$ we get

$$\frac{dw}{dt}=16\sigma^{2}bw,$$ (9)

Which is substituted to the whole lagrangian $L$ with kinetic terms $\frac{\sqrt{\pi}}{2}w^{3}A^{2}\frac{\partial b}{\partial t}$. Which gives

$$b=-\frac{32\sigma^{3}}{A^{2}w}\pm\frac{\sqrt{128\sigma^{2}+\frac{A^{4}w}{4}\left(1-\frac{\beta A^{2}}{\sigma^{2}\sqrt{2}}\right)}}{\sigma wA^{2}},$$ (10)

the frequency chirp equation, which can be determined once the width is known. For positive width, the frequency chirp can be complex given that $\beta>\frac{\sigma^{2}4\sqrt{2}}{A^{6}w}\left(128\sigma^{2}+\frac{A^{4}w}{4}\right)>0$. That is for this shape to be stable $b$ has to be complex and it is complex for hot market temperatures. Otherwise it disintegrates into plane waves. Which has the same results as that from MIdauxois2006physics and other methods of solving the NLS equation. Note that for $0<\beta<\frac{\sigma^{2}\sqrt{2}}{A^{2}}$ the soliton solution is unstable even though the market is at hot temperatures.

III.3 Sech Ansatz

In this section we assume the sech ansatz.

$$\psi(s,t)=A(t)sech\left(\frac{s}{w(t)}\right)exp\left(ib(t)s^{2}\right),$$ (11)

By substituting Eq.(11) into Eq.(2) we get

$$L_{eff}=\frac{\sigma^{2}A^{2}}{6w}+\frac{2}{3}\sigma^{2}b^{2}A^{2}w^{3}-\frac{\beta A^{4}}{3w},$$ (12)

by using the same approach as in sectionIII.1.

III.4 Variational equations

The equations of motions are derived using the Euler-Lagrange equation of motion $\frac{d}{dt}\frac{\partial L}{\partial\dot{q}_{i}}-\frac{\partial L}{\partial q_{i}}$.

For $q=w$ we get

$$\frac{\partial L}{\partial w}=-\frac{\sigma^{2}A^{2}}{6w^{2}}+\frac{8}{3}\sigma^{2}b^{2}A^{2}w^{2}+\frac{\beta}{3w}A^{4},$$ (13)

Using Newton’s second law of motion $\frac{\partial^{2}w}{\partial t^{2}}=-\frac{\partial v_{eff}}{\partial w}$ we get

$$\frac{d^{2}w}{dt^{2}}=\frac{\sigma^{2}A^{2}}{6w^{2}}-\frac{8}{3}\sigma^{2}b^{2}A^{2}w^{2}-\frac{\beta}{3w^{2}}A^{4},$$ (14)

intergrating we get

$$v_{eff}=\frac{\sigma^{2}A^{2}}{6w}+\frac{2}{3}\sigma^{2}b^{2}A^{2}w^{3}+\frac{\beta}{3w}A^{4},$$ (15)

the energy of the system

$$E=+\frac{1}{2}\left(\frac{dw}{dt}\right)^{2}+\frac{\sigma^{2}A^{2}}{6w}+\frac{2}{3}\sigma^{2}b^{2}A^{2}w^{3}+\frac{\beta}{3w}A^{4},$$ (16)

Which is similar to the equation describing a particle moving in a potential well.

For $q=b$ we get

$$\frac{dw}{dt}=\frac{8}{\pi}\sigma^{2}bw^{3},$$ (17)

Which is substituted to the whole lagrangian $L$ with kinetic term $\frac{\pi}{6}wA^{2}\frac{\partial b}{\partial t}$. Which gives

$$b=\pm\sqrt{-\frac{1+2A^{2}\beta}{6w^{4}\left(\frac{32}{\pi}\sigma^{2}w^{3}+\frac{2}{3}\right)}},$$ (18)

the frequency chirp equation, which can be determined once the width is known. For positive width, the frequency chirp can be complex given that $\beta>\frac{1}{2A^{2}}$. Which is similar to the results above. In order for the shape given by the $sech$ function to be stable $b$ has to be complex, and it is complex for hot market temperatures. Which has the same results as that from MIdauxois2006physics and other methods of solving the NLS equation. It is important to note that for $\frac{1}{2A^{2}}>\beta>0$ the soliton solution of this type is unstable even though the market is at a hot temperature.

IV Discussion

In this section, the $Gaussian$ and $sech$ ansatz are compared and discussed. Figure(1) shows the plots of the imaginary part of the frequency chirp $Im(b)_{s,g}$. One can observe that, the plots are nonzero for $\beta>0$, which is the case for hot market temperatures. These graphs shows that the stability of the $Gaussian$ ansatz is higher than that of the $sech$. It is important to note that the imaginary conditions of Eq.(18) does not make the $sech$ ansatz stable only but the $Gaussian$ part of the ansatz also to be stable, also the $Gaussian$ part of the $sech$ ansatz is not divided by the width $w$ which makes only the $sech$ part affected by the width as shown by figure(2). The effect of the width for both the $Gaussian$ and the $sech$ is inversely proportional to the chirp $b$ and the gain of the chirp $Im(b)$ as shown by figure(2). The same relationship is observed for the volatility $\sigma$ and $Im(b)$ for both the $Gaussian$ and $sech$ as shown by figure(3).

Figure 1: The panels shows the stability of the $Gaussian$ and the $sech$ ansatz with respect to the adaptive market heat potential $\beta$, for $A=10$, $w=1$, $\sigma=0.5$

Figure 2: The panels shows the stability of the $sech$ ansatz (right) and the $Gaussian$ ansatz (left) with respect to the adaptive market heat potential $\beta$, for $A=10$, $\sigma=0.2$ with varying width $w$

Figure 3: The panels shows the stability of the $sech$ ansatz (right) and the $Gaussian$ ansatz (left) with respect to the adaptive market heat potential $\beta$, for $A=10$, $w=2$ with varying volatility $\sigma$

V Numerical Simulation

In this section, direct numerical simulations for the IOPM are discussed. The forth-order Runger-Kutta method is used. The simulation employs a discrete scheme of time with discrete time steps of $\Delta t=1/4000$.

Figure 4: The panels shows the evolution of the $Gaussian$ ansatz for $w=2$, $\sigma=0.2$ and $\beta=10$ with the ansatz at initial time $t=0$ and at time $t=9$.

Figure 5: The panels shows the evolution of the $Gaussian$ ansatz for $w=2$, $\sigma=0.8$ and $\beta=10$ with the ansatz at initial time $t=0$ and at time $t=9$.

Figure 6: The panels shows the evolution of the $Gaussian$ ansatz for $w=8$, $\sigma=0.2$ and $\beta=10$ with the ansatz at initial time $t=0$ and at time $t=9$.

Figure 7: The panels shows the evolution of the $Gaussian$ ansatz for $w=8$, $\sigma=0.8$ and $\beta=10$ with the ansatz at initial time $t=0$ and at time $t=9$.

Figure 8: The panels shows the evolution of the $sech$ ansatz for $w=2$, $\sigma=0.2$ and $\beta=10$ with the ansatz at initial time $t=0$ and at time $t=9$.

Figure 9: The panels shows the evolution of the $sech$ ansatz for $w=2$, $\sigma=0.8$ and $\beta=10$ with the ansatz at initial time $t=0$ and at time $t=9$.

Figure 10: The panels shows the evolution of the $sech$ ansatz for $w=8$, $\sigma=0.2$ and $\beta=10$ with the ansatz at initial time $t=0$ and at time $t=9$.

Figure 11: The panels shows the evolution of the $sech$ ansatz for $w=8$, $\sigma=0.8$ and $\beta=10$ with the ansatz at initial time $t=0$ and at time $t=9$.

The direct numerical simulation as shown by Figure(4) to Figure(11), shows that for both the $Gaussian$ and $sech$ ansats the ansatz keep their relative shape for hot market temperatures, even thought the shape does not keep a smooth shape at a later time.

VI Conclusion

I was able to reproduce the conditions for the stability of the solitons solutions using both the $sech$ and the $Gaussian$ ansatz and they both agree with the literature. That is soliton solutions only exists for hot market temperatures, but for cold market temperatures solitons disintegrates into plane waves.