John A. Paulson School of Engineering and Applied Sciences

Slides:



Advertisements
Similar presentations
An Example of Quant’s Task in Croatian Banking Industry
Advertisements

Ind – Develop a foundational knowledge of pricing to understand its role in marketing. (Part II) Entrepreneurship I.
Quantitative Stock Selection Portable Alpha Gambo Audu Preston Brown Xiaoxi Li Vivek Sugavanam Wee Tang Yee.
OPTIONS TRAINING PROGRAM
Saeed Ebrahimijam Spring 2013 Faculty of Business and Economics Department of Banking and Finance Doğu Akdeniz Üniversitesi FINA417.
CHAPTER 9 Technical Analysis. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc., All Rights Reserved. Technical Analysis Technical analysis attempts.
1 Chapter 7: Principles of Asset Valuation Copyright © Prentice Hall Inc Author: Nick Bagley, bdellaSoft, Inc. Objective Explain the principles.
Model In order to automatically identify technical indicators our model: quantizes real-time market trades in 15 second intervals scans ~40,000 data points.
Market Efficiency Chapter 10.
Rajesh Shekhar Data Mining Prof. Chris Volinsky. ◦ Use Data Mining techniques to build a portfolio with superior return/risk characteristics using technical.
Strategic Synergy Investments, LLC – “SSI” An Explanation of Core Strategies COMMODITY TRADING INVOLVES SUBSTANTIAL RISK OF LOSS AND IS NOT SUITABLE FOR.
Copyright © 2011 Pearson Prentice Hall. All rights reserved. Chapter 10 Capital Markets and the Pricing of Risk.
Practical Investment Management
Mr. Perminous KAHOME, University of Nairobi, Nairobi, Kenya. Dr. Elisha T.O. OPIYO, SCI, University of Nairobi, Nairobi, Kenya. Prof. William OKELLO-ODONGO,
A DVERTISING B UDGET YWCA, Lecture 8. O RGANIZATION & O BJECTIVES ORGANIZATION’S GOALS & OBJECTIVES OPERATIONS OBJECTIVE FINANCE OBJECTIVE MARKETING OBJECTIVE.
Is it Possible to “Beat The Market”?  Are markets efficient?  What should we expect of professional money managers?  Observed calendar based effects.
NEURAL NETWORKS FOR TECHNICAL ANALYSIS: A STUDY ON KLCI 授課教師:楊婉秀 報告人:李宗霖.
THE STOCK MARKET CRASH AND THE GREAT DEPRESSION EVENTS THAT HELPED CAUSE THE STOCK MARKET CRASH: 1. OVERSPECULATION: Stock prices had risen far above the.
Saeed Ebrahimijam SPRING Faculty of Business and Economics Department of Banking and Finance Doğu Akdeniz Üniversitesi FINA417.
FOUNDATION BUSINESS SIMULATION
A Neural Network Approach to Predicting Stock Performance John Piefer ECE/CS 539 Project Presentation.
CHAPTER NINE MARKET EFFICIENCY © 2001 South-Western College Publishing.
Presentation by: Bryan Durand Josh Amoss Suri Thummala Steve Beuchaw Matthew Malouin Global Asset Allocation February 28, 2005.
John Locke Locke In Your Success, LLC
Efficient Market Hypothesis by Indrani Pramanick (44)
ECONOMICS 211 Chapter 7 – Clicker Question Set #3.
Efficient Market Hypothesis EMH Presented by Inderpal Singh.
1 International Securities Exchange. 2 Stock Repair Strategy Alex Jacobson ISE Education.
Options on Futures Program Acuvest Trading in futures and options involves a high degree of risk and is not suitable for everyone.
8 8 C h a p t e r Stock Price Behavior And Market Efficiency second edition Fundamentals of Investments Valuation & Management Charles J. Corrado Bradford.
Chapter 16 Jones, Investments: Analysis and Management
CHAPTER EIGHTEEN Technical Analysis CHAPTER EIGHTEEN Technical Analysis Cleary / Jones Investments: Analysis and Management.
CHAPTER NINE MARKET EFFICIENCY Practical Investment Management Robert A. Strong.
Dr. Scott Brown Stock Options. Stocks vs Options Options Are sensitive to: The direction of the underlying stock. The time remaining before expiration.
2.02 Supply and Demand Understand Economics and Economic Systems Interpret supply and demand graphs.
1 MBF 2263 Portfolio Management & Security Analysis Lecture 7 Efficient Market Hypothesis.
Finance 663 – International Finance Passive / Active Strategy on the Euro Stoxx 50 Index.
Research and Evaluation 4.1 INVESTMENT PRINCIPLES.
 Portfolio rebalancing is the process of bringing the different asset classes back into proper relationship following a significant change in one or.
Chapter 10 Market Efficiency.
Stock market forecasting using LASSO Linear Regression model
Research Via Financial Services. About us Research via is a leading financial services provider with presence in Indian and other global capital markets.
Planning Stock Trades: Entry and Exit. Options involve risk and are not suitable for all investors. For more information, please read the Characteristics.
The highly intelligent virtual agents for modeling financial markets G. Yang 1, Y. Chen 2 and J. P. Huang 1 1 Department of Physics, Fudan University.
Ratio Analysis…. Types of ratios…  Performance Ratios: Return on capital employed. (Income Statement and Balance Sheet) Gross profit margin (Income Statement)
What is a stock? What does it mean when we say there is risk and reward in the stock market? Why do people invest their money in the stock market?
 Analysis of statistics generated by market activity such as past price and volume to come up with reasonable outcome in future using charts as a primary.
Managing Technology and Innovation
Figure 5: Change in Blackjack Posterior Distributions over Time.
MARKET EFFICIENCY The concept of Market Efficiency:
TYPES OF MARKETING in the Service Industry
Simplifying how derivative products operate in the stock market – By Prof. Simply Simple TM Hopefully the lesson on “Futures” and “Options” helped you.
An Investigation of Market Dynamics and Wealth Distributions
Capital Asset Pricing Model
Behavioral Finance Unit II.
The Opening Bell Deviation Theory
The Opening Bell Deviation Theory
Chapter 10: Adjustments to the list of quoted prices
Technical Analysis Chapter 16
Start Small and Retire Early With Weekly Options
Student Investment Management Financial Sector
Mean Reverting Asset Trading
Ratio Analysis A2 Accounting.
Ind – Develop a foundational knowledge of pricing to understand its role in marketing. (Part II) Entrepreneurship I.
Market-based Dynamic Task Allocation in Mobile Surveillance Systems
Momentum Momentum is used to indicate the speed with which prices are changing. When a change in direction occurs in a short-term trend, technicians say.
Why do we need a controlled experimental stock market(CESM)?
Beating the market -- forecasting the S&P 500 Index
AS-AD curves: how natural is the natural rate of unemployment?
A Deep Reinforcement Learning Approach to Traffic Management
Presentation transcript:

John A. Paulson School of Engineering and Applied Sciences Stock Trading Using Reinforcement Learning By Jonah Varon and Anthony Soroka CS 182: Artificial Intelligence Fall 2016 Introduction Abstract: We applied Q-Learning to stock trading. Specifically we attempted to place our agent in states to find profitable strategies through technical analysis and mean reversion. Our results were mixed, leading us to question the applicability of reinforcement learning in the stock market – a space that differs in significant ways from those where reinforcement learning has proven suitable. However, our multi-agent technical analysis approach highlighted specific stocks where Q-Learning proved effective for certain time periods. Considering the interest, there is surprisingly limited available research on reinforcement learning applied to stock trading. Hoping to either further public research or find a profitable strategy, we applied Q-Learning to stock trading. We used Quantopian to test our strategies. Quantopian is a crowd-sourced investment firm which provides market data and a python-based development platform for algorithmic investors. Results Approach Figure 1. YELP Tech-Analysis Agent Figure 2. ATML Tech-Analysis Agent Results Summarized: Most of our strategies did not outperform reasonable benchmarks (S&P 500 / specific stock). However, using a multi-agent approach on the technical analysis agent, we were able to scan the market for specific stocks and specific periods where the agent was profitable. Example successful agents are illustrated above. State Representation Q-Learning is defined by the following update function: The state space for the stock market is vast. Hence, limiting the space to allow the agent to learn is crucial. Given such, we represented states in hopes of positioning our agent to find known trading strategies: Technical Analysis: last 3 minutes of price and volume data {-1: Decreased, 0: Unchanged, 1: Increased}. # of States = 3^6 = 729 # of Actions per Year = 98,280 Mean Reversion: Previous week’s stock performance {-1: Stock was in the bottom X Percentile, 1: Stock was in top X percentile} For our project, we typically applied the below: a = {Buy, Sell} r = Stock Performance The stock market however has a few interesting characteristics that slightly alter Q-Learning: a) Future State: Due to the ability to rebalance, max Q(s’,a’) is the same regardless of Q(s,a) (presuming you are a small enough investor with minimal market impact). b) Fully Observable: Limiting the agent to actions of Buy and Sell, the reward from Q(s,a) is fully observable regardless of the actual action taken. We tried both approaches to updating Q(s,a): 1) Only for action taken 2) For both taken and not taken actions Conclusions Our attempts to find alpha using Q-Learning illustrate how competitive and efficient markets are. Additionally it highlights that reinforcement learning is perhaps not the best suited for stock trading given: Stochasticity of the market Limited amount of train data Agent can observe rewards without taking actions Agent has little impact on the environment That being said the possible methods to define a state space are limitless and should still be investigated. Moreover, our multi-agent approach proved useful in highlighting specific stocks where Q-Learning might be applicable. Lastly, we believe Deep Learning techniques such as LSTM might be better suited for this problem. Data Quantopian provides a wealth of technical, fundamental, and event information from thousands of stocks. Code Snippet. Updating Technical Analysis Agent’s State John A. Paulson School of Engineering and Applied Sciences