Algorithmic Trading Challenge implemented as part of the term project for Foundations of Machine Learning at NYU Courant in Fall 2016 (http://cs.nyu.edu/courses/fall16/CSCI-GA.2566-001/index.html/)
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
data
evaluations
models
notebooks
predictions
random_scripts
run_models
.gitattributes
.gitignore
README.md
evaluate_prediction.py

README.md

Algorithmic Trading Challenge

Foundations of Machine Learning

CSCI-GA.2566-001

Term Project

Report

Introduction

The aim of the Kaggle challenge Algorithmic Trading Challenge is to architect emprical models to predict market response to large trades. The challenge is already closed and has been picked up for learning purposes and as the term project for the course Foundations of Machine Learning at NYU Courant in Fall 2016.

Abstract

The aim of the project is to develop empirical predictive models for market response to large trades or more intuitively known as liquidity shock. The large dataset provided by the challenge consists of Training and Test datasets separately containing trade and quote data from London Stock Exchange(LSE) bfore/after a liquidity shock. This project analyses and evaluates different models, namely, boosting techniques like AdaBoost and Random Forests Regressors[1] and Linear Regression with different Heuristics for Feature Engineering.

Most of the programming and visualizations are done in python (with Jupyter Notebook for book-keeping and visualizations).

References

  1. Algorithmic Trading Challenge
  2. Liaw, Andy, and Matthew Wiener. "Classification and regression by randomForest." R news 2.3 (2002): 18-22.
  3. de Abril, Ildefons Magrans, and Masashi Sugiyama. "Winning the kaggle algorithmic trading challenge with the composition of many models and feature engineering" IEICE TRANSACTIONS on Information and Systems 96.3 (2013): 742-745.