Forums > Pricing & Modelling > Introducing New Technologies in Mathematical Finance Discipline. A Totally New Paradigm.

 Page 1 of 1
Display using: amin Total Posts: 284 Joined: Aug 2005  amin Total Posts: 284 Joined: Aug 2005
 Posted: 2019-07-13 09:44 As the friends would notice that the above program is quite accurate for reasonable volatilities but for very high volatilities, the results start to deteriorate after a year and more. On the other hand, well written PDEs give precise results even for extremely high volatilities. Therefore I decided to add PDEs to my existing research repertoire and substitute numerical solution of Fokker Planck where it is easily available and continue with the analytics. Once a precisely calculated density on a grid is available, it does not matter for further analytics if we obtained the grid from Ito-Taylor density generation method or from numerical solution of a PDE. I will continue to try to make the results of my Ito-Taylor density method sharply precise for extreme volatilities but in the mean time would also would do analytics with numerical solution of forward Fokker-Planck PDEs. Here is a post I made on Linkedin that elaborates on this idea.https://www.linkedin.com/pulse/introducing-new-technologies-mathematical-finance-discipline-amin/Introducing New Technologies in Mathematical Finance Discipline. A Totally New Paradigm.I want to mention some interesting research achievements at Infiniti Derivatives Technologies in the area of stochastics leading to huge improvements in mathematical methods of derivatives valuation, hedging, trading and risk management.1. Valuation of Path Dependent Stochastic Integrals on a PDE Grid.It has been common historical wisdom that path dependent stochastic integrals cannot be calculated on a PDE grid. In our new research, we debunk these past notions and show that there are methods of introducing path dependence on a PDE grid. We have introduced new technologies that can calculate any stochastic path integral on a one or more dimensional PDE grid in fraction of a second after the PDE has been solved. Of course calculating path integrals on higher dimensional PDEs would require larger time but the time required to solve stochastic path integrals would strictly be far lesser than required to solve the PDE itself. Some of these stochastic integrals could be for example conditional value of path integral of squared volatility on the PDE grid for valuation of derivatives with stochastic volatility. Once the PDE has been solved conditional backward stochastic integrals like exponential of path integral of interest rates representing conditional bond prices or discount factor could also be calculated on the same PDE grid. From the PDE of interest rates, you could calculate the backward evolution of difficult stochastic path integrals depending on interest rates of whose PDE cannot be written easily. For Asian options, you could find the path integral of underlying as a forward stochastic integral on the Fokker planck equation solved by a one dimensional PDE or on a two dimensional PDE with stochastic volatility. The conditional values of these stochastic path integrals are available anywhere on the PDE grid along the backward or forward evolution of the PDE. PDE methods can only calculate backward evolution of payoffs whose explicit PDE has been written with arbitrary final data but we can calculate backward dynamic programming valuation of far more complex stochastic integrals dependent on PDE underlying of whose explicit PDE is not available but conditional of the value of the underlying as calculated in the forward PDE. We do not use any simulation or random numbers and only rely on fast analytics to achieve the above mentioned goals.2. Calculation of Optimal Quantizers for Monte Carlo Simulations.Once a Fokker planck PDE has been solved, we can calculate optimal quantizers in one or two dimensions in fraction of a second. You could use these optimal quantizer schemes for forward evolution of stochastic integrals or backward dynamic programming in a monte carlo setting. As opposed to existing optimal quantizer methods that are based on monte carlo discretization schemes coupled with optimization techniques, we take a Fokker planck PDE solution as a starting point which is more cheaply available and simply use very fast proprietary analytics to generate optimal quantizer grid that can later be used for extremely fast monte carlo simulations and backward dynamic programming.If you find any of this interesting and would like to implement this solution at your institution, you can email me at anan2999 yahoo com for further information and discussions. I want to mention again that none of these new technologies are in public domain.  amin Total Posts: 284 Joined: Aug 2005  amin Total Posts: 284 Joined: Aug 2005 