Posts

RiskMinds 2019

Flying Monday to Amsterdam with my colleagues from Superfly Analytics of Danske Bank, including Brian Huge and Ove Scavenius, to attend the RiskMinds 2019 risk management conference and the award ceremony, where our group is nominated for ‘Excellence in risk management and modeling’.

EDIT: Superfly Analytics now won the award:

Excellence in Risk Management and Modelling, winner: Superfly Analytics at Danske Bank

We will be presenting our vision of modern risk management systems and the ‘One Analytics’ platform: full front to back consistency with scripting of cash-flows , model hierarchies and AAD. Further, we will present ‘Deep Analytics’: leveraging risk management systems with AI to learn revaluation and risk analytics on the fly. For those unable to attend, we posted our slides online here:

www.deep-analytics.org

click on the picture to see the presentation

Article with code, Wilmott Nov19

My paper  “Computation graphs for AAD and Machine Learning part I: Introduction to computation graphs and automatic differentiation” was just published in Wilmott, the first in a series of articles with code dedicated to Computational Finance. It covers computation graphs, backpropagation, AAD and implementation in finance, taking inspiration in the recent achievements of the Superfly Analytics group of Danske Bank.

A. Savine, “Computation graphs for aad and machine learning part i: introduction to computation graphs and automatic differentiation,” Wilmott, vol. 2019, iss. 104, p. 36–61, 2019.

Deep learning derivatives pricing

I made two simplistic TensorFlow (1.x) notebooks for the benefit of my students at Copenhagen University, to demonstrate how vanilla neural nets (deeply) learn pricing of European calls and high dimensional basket options, together with a comparison with conventional polynomial regression models (a la LSM) and a quick, simple introduction to the implementation of deep learning models in TensorFlow.

Notebook 1: European call in Black and Scholes

Notebook 2: Basket option in Bachelier

The notebooks run directly on Google colab, in the cloud and on GPU, no installation required.

The lecture slides and material are all found here: github.com/asavine/CompFinLecture

Some (much) more advanced considerations for efficiently learning prices of trading books, including “twin” neural nets who learn values and risks, and the super efficient differential regularization, are found here: slideshare.net/AntoineSavine/deep-analytics