M7_WQU_MLF_Module_7_Compiled_Content.pdf - MScFE xxx[Course...

This preview shows page 1 - 6 out of 39 pages.

MScFE xxx [Course Name] - Module X: Collaborative Review Task 1 Revised: 09/07/2019
MScFE 650 Machine Learning in Finance Table of Contents © 2019 - WorldQuant University All rights reserved. 2 Module 7: Machine Learning for Finance ....................................................... 3 Unit 1: Introduction ................................................................................................................. 4 Unit 2: Workflows and Solution Documents .................................................................. 5 Unit 3: Financial Data Structures ....................................................................................... 7 Unit 4: Labeling Techniques ............................................................................................... 13 Bibliography ............................................................................................................................. 15
MScFE 650 Machine Learning in Finance - Module 7: Summary © 2019 - WorldQuant University All rights reserved. 3 Module 7: Machine Learning for Finance In this module students are exposed to the top reasons most machine learning funds fail. We introduce new financial data structures that have their roots in high-frequency trading, and we end the module by investigating various labeling techniques − including meta -labeling, which is used to boost the performance metrics of a primary model.
MScFE 650 Machine Learning in Finance - Module 7: Unit 1 © 2019 - WorldQuant University All rights reserved. 4 Unit 1: Introduction Welcome to the final module of this course, where we will be discussing some of the latest advances in financial machine learning. In this module, we will lightly cover some of the material in Dr Marcos Lopéz de Prado’s book, Advances in Financial Machine Learning. The video lectures are made of Dr Lopéz de Prado’ s keynote lecture, from the 2018 Bloomberg Quant seminar series, on why most machine learning funds fail. To complement the guest lectures, the three sets of notes in this module also cover, on a highly practical level, some of the latest advances in financial machine learning. We begin by introducing a data science workflow and solutions document; then we move on to cover new sampling techniques for creating financial data structures that have their roots in high-frequency trading (HFT). Finally, we address labeling techniques, as well as how to boost performance metrics using meta-labeling.
MScFE 650 Machine Learning in Finance - Module 7: Unit 2 © 2019 - WorldQuant University All rights reserved. 5 Unit 2: Workflows and Solution Documents The first set of notes consist of two outlines, both inspired by interviews from top Kagglers: the first is a description of a workflow for data science projects; the second, of another very valuable tool namely, a solutions document. What a typical workflow looks like Following much of the agile philosophy of project management, we recommend the following approach: fail fast, iterate, and pivot. This allows you to adapt your techniques to a given situatio n in real time. Click here for a lecture by Atlassian which introduces agile and some of its tools. Now, outside of the typical structure of agile project management which includes notions such as sprint planning, daily stand ups, and sprint reviews the following is the typical structure we recommend you follow when tackling a new project: 1 Set up and update the solutions document.

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture