1 What does genetic programming teach us about the foreign exchange market ? Chris Neely* Paul Weller † Rob Dittmar** December 1-2, 1998 *Economist, Federal.

Slides:



Advertisements
Similar presentations
Efficient Market Hypothesis (EMH). Premises of An Efficient Market -A large number of competing profit-maximizing participants analyze and value securities,
Advertisements

Draft lecture – FIN 352 Professor Dow CSU-Northridge March 2012.
Pretests for genetic-programming evolved trading programs : “zero-intelligence” strategies and lottery trading Nicolas NAVET INRIA - AIECON NCCU
1 Futures Futures Markets Futures and Forward Trading Mechanism Speculation versus Hedging Futures Pricing Foreign Exchange, stock index, and Interest.
Futures markets. Forward - an agreement calling for a future delivery of an asset at an agreed-upon price Futures - similar to forward but feature formalized.
Models and methods to estimate the appropriate r
By Paul Cottrell, BSc, MBA, ABD. Author Complexity Science, Behavioral Finance, Dynamic Hedging, Financial Statistics, Chaos Theory Proprietary Trader.
FIN352 Vicentiu Covrig 1 Asset Pricing Models (chapter 9)
USING A GENETIC PROGRAM TO PREDICT EXCHANGE RATE VOLATILITY Christopher J. Neely Paul A. Weller.
SOME LESSONS FROM CAPITAL MARKET HISTORY Chapter 12 1.
Chapter 8 Exchange Rate Forecasting, Technical Analysis and Trading Rules.
Trading Rules and Market Efficiency Fin250f: Lecture 4.3 Fall 2005 Reading: Taylor, chapter 7.
FINANCE 9. Optimal Portfolio Choice Professor André Farber Solvay Business School Université Libre de Bruxelles Fall 2007.
Corporate Finance Introduction to risk Prof. André Farber SOLVAY BUSINESS SCHOOL UNIVERSITÉ LIBRE DE BRUXELLES.
International Fixed Income Topic IVC: International Fixed Income Pricing - The Predictability of Returns.
Lectures , & : International Asset Portfolios Galina A Schwartz Department of Finance University of Michigan Business School.
1 1 Ch22&23 – MBA 567 Futures Futures Markets Futures and Forward Trading Mechanism Speculation versus Hedging Futures Pricing Foreign Exchange, stock.
McGraw-Hill/Irwin © 2008 The McGraw-Hill Companies, Inc., All Rights Reserved. Capital Asset Pricing and Arbitrage Pricing Theory CHAPTER 7.
FINANCIAL ECONOMETRICS FALL 2000 Rob Engle. OUTLINE DATA MOMENTS FORECASTING RETURNS EFFICIENT MARKET HYPOTHESIS FOR THE ECONOMETRICIAN TRADING RULES.
Investment Analysis and Portfolio Management
EXCHANGE RATE DETERMINEATION National Balance of Payments; International Monetary Systems; Methods of determining exchange rates:
McGraw-Hill/Irwin © 2008 The McGraw-Hill Companies, Inc., All Rights Reserved. Capital Asset Pricing and Arbitrage Pricing Theory CHAPTER 7.
Principles of Microeconomics & Principles of Macroeconomics: Ch. 2 Second Canadian Edition Chapter 2 Thinking Like an Economist © 2002 by Nelson, a division.
Slide 1–1. Part I Introduction Chapter One Why Study Financial Markets and Institutions?
Article 2 The theory of stock market efficiency Dr. Yang April 15, 2015 Group 2 Greg Werthman Kapil Jain Aaron Cyr Richard Oluoha Jen-Chiang La.
Chapter 13 The Foreign Exchange Market. Copyright © 2007 Pearson Addison-Wesley. All rights reserved Topics to be Covered Foreign Exchange Market.
13 CHAPTER Money, the Price Level and Inflation © Pearson Education 2012 After studying this chapter you will be able to:  Define money and describe.
Exchange Rates Dr. Antony Mueller The Continental Economics Institute
FINANCIAL MANAGEMENT Financial Management.
Computational Finance Lecture 2 Option Pricing: Binomial Tree Model.
Foreign Exchange Risk Chapter 14 © 2008 The McGraw-Hill Companies, Inc., All Rights Reserved. McGraw-Hill/Irwin.
Ch. 22 International Business Finance  2002, Prentice Hall, Inc.
A comparison of MA and RSI returns with exchange rate intervention Group Members: Zhang Duo A Tang Wai Hoh A Fan Li A
Capital Asset Pricing Model CAPM Security Market Line CAPM and Market Efficiency Alpha (  ) vs. Beta (  )
Finance - Pedro Barroso
Lecture #3 All Rights Reserved1 Managing Portfolios: Theory Chapter 3 Modern Portfolio Theory Capital Asset Pricing Model Arbitrage Pricing Theory.
Investment and portfolio management MGT 531. Investment and portfolio management  MGT 531.
FIN 352 – Professor Dow.  Fama: Test the efficient market hypothesis using different information sets.  Three categories:  Weak  Semi-Strong  Strong.
Dr. Tucker Balch Associate Professor School of Interactive Computing CS 7646: Machine Learning for Trading Company Value Find out how modern electronic.
Money and Capital Markets 25 C h a p t e r Eighth Edition Financial Institutions and Instruments in a Global Marketplace Peter S. Rose McGraw Hill / IrwinSlides.
Financial Forces McGraw-Hill/Irwin International Business, 11/e Copyright © 2008 The McGraw-Hill Companies, Inc. All rights reserved. chapter eleven.
Principles of Microeconomics & Principles of Macroeconomics: Ch. 2 First Canadian Edition The Economic Way of Thinking u The Scientific Method uses abstract.
McGraw-Hill/Irwin Copyright © 2008 The McGraw-Hill Companies, Inc., All Rights Reserved. Capital Asset Pricing and Arbitrage Pricing Theory CHAPTER 7.
“Differential Information and Performance Measurement Using a Security Market Line” by Philip H. Dybvig and Stephen A. Ross Presented by Jane Zhao.
NS3040 Fall Term 2014 Keynesian/Monetarist Debates.
Chapter 2 Thinking Like an Economist Ratna K. Shrestha.
Real Exchange Rate Fluctuations: Reflections on the Uruguayan Experience Umberto Della Mea * Economic Policy Division Central Bank of Uruguay Outline I.
INVESTMENTS | BODIE, KANE, MARCUS Copyright © 2011 by The McGraw-Hill Companies, Inc. All rights reserved. McGraw-Hill/Irwin CHAPTER 4 Risk and Portfolio.
Chapter 12 The Foreign- Exchange Market. ©2013 Pearson Education, Inc. All rights reserved Topics to be Covered Spot Rates Forward Rates Arbitrage.
25-1 Economics: Theory Through Applications This work is licensed under the Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported.
Slide 9-1 Market Efficiency 1. Performance of portfolio managers 2. Anomalies 3. Behavioral Finance as a challenge to the EMH 1/7/
IT Applications for Decision Making. Operations Research Initiated in England during the world war II Make scientifically based decisions regarding the.
Soft Computing methods for High frequency tradin.
Chapter 1 Why Study Money, Banking, and Financial Markets?
TECHNICAL ANALYSIS.  Technical analysis attempts to exploit recurring and predictable patterns in stock prices to generate high investment returns.
Chapter 3 Foreign Exchange Determination and Forecasting.
PowerPoint Presentation by Charlie Cook Copyright © 2004 South-Western. All rights reserved. Chapter 11 Rational Expectations, New Classical Macroeconomics,
7.0 Project Cycle Preamble Whether it is an investment or servicing project, every project is expected to take off with initiation of ideas and creation.
Momentum and Reversal.
Capital Market Theory: An Overview
Basic Finance Securities Markets
Empirical Financial Economics
Basic Finance Securities Markets
Capital Asset Pricing and Arbitrage Pricing Theory
The CAPM is a simple linear model expressed in terms of expected returns and expected risk.
Momentum Effect (JT 1993).
The Foreign Exchange Market
NS3040 Fall Term 2018 Keynesian/Monetarist Debates
Capital Asset Pricing Model
Presentation transcript:

1 What does genetic programming teach us about the foreign exchange market ? Chris Neely* Paul Weller † Rob Dittmar** December 1-2, 1998 *Economist, Federal Reserve Bank of St. Louis †Professor, Department of Finance, University of Iowa **Scientific Programmer, Federal Reserve Bank of St. Louis

2 Disclaimer The views expressed are my own and do not necessarily reflect official positions of the Federal Reserve Bank of St. Louis, or the Federal Reserve System.

3 What does genetic programming teach us about foreign exchange markets? I) Broad overview of an ongoing project II) Foreign exchange market efficiency and technical analysis III) What is genetic programming? IV) Results from dollar exchange rates V) Results from the European Monetary System VI) Results using Federal Reserve Intervention VII) Work in Progress

4 II)Foreign exchange market efficiency and technical analysis A) Foreign exchange market efficiency 1)Exchange rates reflect information to the point where the potential excess returns do not exceed the transactions costs of acting (trading) on that information (Jensen, 1978). 2)Borrowing in one currency to lend in another should not profit you, except to the extent that this is risky strategy.

5

6 B)The puzzle of technical analysis 1)Technical analysis is the use of past prices to guide trading decisions. 2)For about 15 years, people have been finding that technical trading rules make excess returns in the foreign exchange market. (Sweeney, 1986; 1988) (a)Moving average rules and filter rules. 3)The success of technical trading rules seems to contradict the efficient markets hypothesis.

7 C)Explanations for the success of technical analysis 1)Data mining 2)The returns to technical analysis are compensation for bearing risk (a)Measures of risk: Sharpe ratios, CAPM betas 3)Central bank intervention

8 III) What is genetic programming? A) GP is a variation of genetic algorithms 1) GA are due to Holland (1975) 2) Genetic algorithms are computer search procedures based on the principles of natural selection as originally expounded in Darwin's theory of evolution. B) GP is a similar search algorithm for spaces that consist of decision trees. 1) GP was developed by Koza (1992). 2) We can think of trading rules as decision trees.

9 IV)“Is Technical Analysis in the Foreign Exchange Market Profitable? A Genetic Programming Approach” Neely, Weller and Dittmar (1997) A) The data: daily exchange rates and interest rates. $/DM, $/¥, $/£, $/SF, DM/¥ and £/SF 1)Normalized by a 250-day moving average B)The fitness criterion is the excess return over borrowing in one currency and investing in the other, each day.

10 C) Sample periods: training period: ; selection period: : validation period: :10:11.

11

12 E) Structure of the rules 1) Rules were usually too complex to analyze by hand but those that were understandable had extrapolative features. 2) The 55th best $/DM rule over the selection period, whose excess return was 7.34, number of trades was 37 and correlation with the median rule was , prescribed: "Take a long position if the four-day minimum of the normalized exchange rate is greater than one." F) Rules trained on $/DM data proved profitable on other exchange rates, out of sample.

13

14 G) Transactions costs can aid in avoiding overfitting the data 1) Transactions costs were set at per round trip in the training/selection period, in the validation period. H) Risk measures 1) Sharpe ratios were between 0.1 and 0.5. The S&P500 Sharpe ratio was about 0.3 over the same period 2) CAPM beta

15

16 V) "Technical Trading Rules in the European Monetary System” -- Neely and Weller (1998) A) The Data and sample periods 1) Four ERM exchange rates: DEM/FRF, DEM/ITL, DEM/NLG and DEM/GBP 2) Training period, 3/13/79 to 1/2/83; selection period, 1/3/83 to 1/1/86; validation period, 1/2/86 to 6/21/96.

17 B) Results from portfolio rules on ERM data

18 C) Structure of the rules: The interest differential and not the past exchange rate series is the most important informational input to the trading rule. 1) Example: “Go long if the interest differential (British minus German) is greater than 4.42 per cent.” a) This simple rule was the 28th best out-of-sample for the DEM/GBP, having an excess return of 3.32 per cent per year and a correlation of 0.95 with the median rule. 2) Moving average and filter rules did not do well. 3) High interest rate and mean reversion rules did not do well.

19 D) Risk measures: 1) CAPM betas were close to zero. 2) X* statistics were close to the unadjusted returns.

20 VI) “Technical Analysis and Central Bank Intervention” (Neely and Weller, 1998) A) An explanation for the success of TA B) There is previous work linking technical analysis and central bank intervention. C) Can CBI information improve an ex ante trading rule? D) Method: Supply intervention information as 1, 2, or 3 to the rule generating program.

21

22 F) How do we explain the failure of intervention information to improve returns? 1) A structural break in the CBI data generation process?

23 VII) What do we learn about market efficiency from these exercises? A) The success of TA presents a puzzle to the EMH; the success of GP deepens this puzzle because GP provides a true, ex ante test of technical trading rules. B) There has been work on institutional constraints that may explain some lack of risk arbitrage.

24 C) There has been additional work on behavioral finance as a result of the success of TA. D) The success of TA underscores our need for a better understanding of risk.

25 VIII) Work in Progress A) High frequency trading rules B) Options pricing 1) HLP showed how neural networks could price and delta hedge options. We are exploring similar issues with GP.

26 The End