A comparison of MA and RSI returns with exchange rate intervention Group Members: Zhang Duo A0075433 Tang Wai Hoh A0075413 Fan Li A0075376.

A comparison of MA and RSI returns

with exchange rate intervention

Group Members: Zhang Duo A0075433Tang Wai Hoh A0075413Fan Li A0075376

outline

• Introduction of the paper• Data & Trading Strategy• Methodology & Empirical Results• Work in Progress & Future

Exploration

Introduction of the Paper

Introduction of the Paper

• TOPIC: A comparison of MA and RSI returns with

exchange rate intervention

• Authors: Thomas C. Shik and Terence Tai-Leung Chong

• Structure: Two Trading Strategies: MA & RSI Six Currencies: AUD/USD, CHF/USD, DEM/USD,

JPY/USD, BP/USD, EUR/USD Three Panels: Panel A: All observations Panel B: Remove domestic & foreign interests Panel C: Remove factor of intervention

Report StructureTrading Rules

MA(N=10,20,50,150)

RSI(N=10,20,50,150)

Empirical Results

Panel A All observations

Test Model:ObsMeant-statp-valueSharpe(1)Sharpe(2)

Test Model:ObsMeant-statp-valueSharpe(1)Sharpe(2)

(1)RSI&MA: positive risk adjusted returns on JPY/USD & DEM/USD

(2)RSI>MA for DEM/USD;

MA>RSI for JPY/USD

Panel BRemove domestic & foreign interest

Test Model:ObsMeant-statp-valueSharpe(1)Sharpe(2)

Test Model:ObsMeant-statp-valueSharpe(1)Sharpe(2)

(3) Little impact on trading rules performance

Panel CRemove factor of intervention

Test Model:ObsMeant-statp-valueSharpe(1)Sharpe(2)

Test Model:ObsMeant-statp-valueSharpe(1)Sharpe(2)

(4) Profitability of trading rules positively related to interventions

Data

DATA of the PaperFX Rate Interest Rate Intervention

Data

AUD/USD Noon buying rates in New York time

Unofficial market 11 a.m. call rates series

The Reserve Bank of Australia (1983-2002)

CHF/USD Swiss Call money rate Swiss National Bank(1986-1995)

DEM/USD Daily overnight interest rate data at 0900GMT from BIS

Deutsche Bundesbank

JPY/USD Ministry of Finance Japan(since 1991)

BP/USD

EUR/USD

Risk-free Rate Daily interest rate of 30-year US Treasury Bonds

DATA of Our Research• Foreign Exchange Rates Resource 1 : H.10 Federal Reserve Statistical Release (since

1971) http://www.federalreserve.gov/releases/h10/hist/default1989.htm Resource 2: Yahoo Finance (since 27 Dec 2007) http://sg.finance.yahoo.com/q/hp?s=GBB

• Interest Rates(Some) Resource 3: Bloomberg (since 2001)

• Intervention Data(Only One) Resource 4: JPY/USD Intervention Data(Apr 1991-Mar 2001) http://www.mof.go.jp/english/feio/e034_133.htm

• Risk-free Rate(Done) Resource 5: H.15 Federal Reserve Statistical Release (since

1977) http://www.federalreserve.gov/releases/h15/data/Business_day/H15_TCMNOM_Y30.txt

Trading Strategy

ACF s for Exchange Rate Series• Serially Correlated possible to make profits by investigating its history

1ln lnt t ty e e

t k t k ty y AUD/USD

CHF/USD

DEM/USD

JPY/USD (A)

JPY/USD (B)

BP/USD

EUR/USD

ACF(1)s.e.

0.0376640.016639

0.0338780.019953

0.0429370.016017

0.0702680.022661

0.0482420.022619

0.0747090.015991

0.0014570.025090

ACF(2)s.e.

ACF(3)s.e.

-0.0450370.016616

0.0627980.022715

-0.0546010.022610

ACF(4)s.e.

ACF(5)s.e.

Serially Correlated!!

Define: Regression:

Trading Strategies

• Definition

• Long the USD if

• Short the USD if

1

0

1( )

N

t t ii

MA N PN

( )t tP MA N

( )t tP MA N

1

1 101

10

( ) { }( ) 100

| |

N

t i t i t i t iit N

t i t ii

P P I P PRSI N

P P

( ) 50tRSI N

( ) 50tRSI N

• Definition

• Long the USD if

• Short the USD if

MAMA

Note: Pt : Exchange rate at time t, N: Number of days.

RSIRSI

Trading Strategies

RSI in the paper-SMA

RSI in the paper-SMA

100100 100

1 1( )t

RSRSI N

RS RS

SMA( , )

SMA( , )

U nRS

D n

RSI in General FormRSI in General Form

( , )

( , )

Average U tRS

Average D t

1

0

1( )

N

t t ii

SMA N PN

----Average of N days up prices----Average of N days down prices

RSI in algoquant-EMA

RSI in algoquant-EMA

EMA( , )

EMA( , )

U nRS

D n

1(1 ()( ) )t ttE P EMA N MA N

Methodology & Empirical Results

Methodology & Empirical Result

• Mean Annual Returns & Standard Deviation

• Hypothesis Testing t-statistic

• Sharpe Ratios

Mean Annual Returns

• Window widths for the study are 10-, 20-, 50- and 150- days.

• Daily returns:

• Removing interest rate differentials:

signalsellfor

signalbuyfor

rrPP

t

tttttt

1

1

1ln1lnlnln *1

tttt PP lnln 1

Note: rt* = foreign interest rate, rt is the domestic interest rate

Total Returns

• Sum of daily returns:

• Average return from time 0 to T:

• Mean Annual Returns is computed by multiplying with the number of trading days.

1

0

T

it

TT

T

i t

1

0

Hypothesis Testing• Let μ and σ be the mean and standard deviation

of the daily returns respectively.

• Sample mean:

• Hypothesis H0 : μ = 0 vs H1: μ ≠ 0 tested using.

- Efficient market with no arbitrage has a mean zero.

• S = sample standard deviation

TN 2,~

TSt

/

Sharpe Ratios

• be the mean annual return and be standard deviation.

• Rf is the risk-free rate.

S

S

Sharpe

1

S

RSharpe f

2

measure of the excess return (or risk premium) per unit of risk in an investment asset or a trading strategy

Statistic terms in the Tables

• Obs. : number of observations.

• Mean : mean annual return in percentage.

• Std (daily returns) : standard deviation of the daily returns.

• t-stat : t-statistic value.

• P-value : tail probability generated by the observed test statistic under the null hypothesis.

Report Structure

Trading Rules for JPY/USD & DEM/USD

MA(N=10,20,50,150)

JPY/USD

RSI(N=10,20,50,150)

DEM/USD

Empirical Results

Panel A (all obs)

Test Model: MA10Obs: 1932Mean: 12.27%Std: 0.63%t-stat: 3.40P-value: 0.00Sharpe(1): 1.23Sharpe(2): 0.58

Test Model: RSI 10Obs: 3882Mean: 7.50%Std: 0.69%t-stat: 2.70P-value: 0.01Sharpe(1): 0.69Sharpe(2): 0.17

(1)RSI&MA: positive risk adjusted returns on JPY/USD & DEM/USD

(2)RSI>MA for DEM/USD;

MA>RSI for JPY/USD

Panel B(remove demostic &

foreign interest)

Test Model: MA10Obs: 1932 Mean: 12.15%Std: 0.63%t-stat: 3.36P-value: 0.00Sharpe(1): 1.21Sharpe(2): 0.57

Test Model: RSI 10Obs: 3882Mean: 7.41%Std: 0.69%t-stat: 2.66P-value: 0.01Sharpe(1): 0.68Sharpe(2): 0.17

(3) Little impact on trading rules performance

Panel C(remove factor of intervention)

Test Model: n.aObs: n.aMean: n.aStd: n.a t-stat: n.aP-value: n.aSharpe(1): n.a Sharpe(2): n.a

Test Model: RSI 10Obs: 3534Mean: 5.45% Std: 0.67%t-stat: 2.12P-value: 0.03Sharpe(1): 0.52Sharpe(2): -0.01

(4) Profitability of trading rules positively related to interventions

Work in Progress & Future Exploration

Work in Progress

EUR/USD – 50-days window width

Work in Progress

EUR/USD – 50-days window width

Work in Progress

JYN/USD – 50-days window width

Work in Progress

JYN/USD – 50-days window width

Work in Progress

GBP/USD – 10-days window width

Work in Progress

GBP/USD – 10-days window width

Current Challenges

• Quotation data other than Yahoo Finance

• How to store the data/dates at the crossover

points and then to do the return analysis

(mean, std, t-stat, P-value, Sharpe ratios)

• Obstacles in getting information regarding

government interventions and interest rate

Further Exploration

• Simulate results under Panel A & C (Domestic/Foreign interest rates & GOV

interventions)

• Introducing other MA model (i.e. EMA)

• Include transaction cost

• Include two more bounds for RSI a) >70 – overbought - SELLb) <30 – oversold - BUY

RSI with 30, 50, 70 bounds

Hit 30 from above:Buy

Hit 30 from above:Buy

Hit 70 from below:Sell

Hit 70 from below:Sell

Hit 50 from above:Sell

Hit 50 from above:Sell

Hit 50 from below:Buy

Hit 50 from below:Buy

Q & AThank You!