Lim, Ye-Sheen;
(2022)
Deep Learning of the Order Flow for Modelling Price Formation.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
Lim_thesis.pdf Download (2MB) | Preview |
Abstract
The objective of this thesis is to apply deep learning to order flow data in novel ways, in order to improve price prediction models, and thus improve on current deep price formation models. A survey of previous work in the deep modelling of price formation revealed the importance of utilising the order flow for the deep learning of price formation had previously been over looked. Previous work in the statistical modelling of the price formation process in contrast has always focused on order flow data. To demonstrate the advantage of utilising order flow data for learning deep price formation models, the thesis first benchmarks order flow trained Recurrent Neural Networks (RNNs), against methods used in previous work for predicting directional mid-price movements. To further improve the price modelling capability of the RNN, a novel deep mixture model extension to the model architecture is then proposed. This extension provides a more realistically uncertain prediction of the mid-price, and also jointly models the direction and size of the mid-price movements. Experiments conducted showed that this novel architecture resulted in an improved model compared to common benchmarks. Lastly, a novel application of Generative Adversarial Networks (GANs) was introduced for generative modelling of the order flow sequences that induce the mid-price movements. Experiments are presented that show the GAN model is able to generate more realistic sequences than a well-known benchmark model. Also, the mid-price time-series resulting from the SeqGAN generated order flow is able to better reproduce the statistical behaviour of the real mid-price time-series.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | Deep Learning of the Order Flow for Modelling Price Formation |
| Event: | UCL (University College London) |
| Open access status: | An open access version is available from UCL Discovery |
| Language: | English |
| Additional information: | Copyright © The Author 2021. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| UCL classification: | UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Computer Science UCL > Provost and Vice Provost Offices > UCL BEAMS UCL |
| URI: | https://discovery.ucl.ac.uk/id/eprint/10142659 |
Loading...
Loading...Loading...Loading...Archive Staff Only
 |
View Item |
