Description
Autocorelation coefficient tests
EMPIRICAL EVIDENCE ON WEAK FORM EFFICIENT MARKET HYPOTHESIS
Weak-Form Efficiency
? Prices reflect all past price and volume data. ? Technical analysis, which relies on the past history of
prices, is of little or no value in assessing future changes in price.
? Market adjusts or incorporates this information quickly
and fully.
Weak-Form Evidence
? Test for independence (randomness) of stock
price changes
? ?
If independent, trends in price changes do not exist Overreaction hypothesis and evidence
? Test for profitability of trading rules after
brokerage costs
?
Simple buy-and-hold better
Autocorelation coefficients test
One of the most direct and intuitive test of random walk is to check for serial corelation. If stock prices exibit a random walk, the returns of ) cov(rt , rt ? k stocks are uncorrelated at all cov(rt , rt ? k ) ? k ) lags. ? leads(and? var[rt ] var[rt ] * var[rt ? k )
The join hypothesis states that all the serial correlation coefficients, are simultaneosly equal to zero.
Weak-Form EMH
? Runs tests
? looking for patterns in signs of returns ? i.e. + + - + - + ? Filter rules ? sell after falls a certain % or buy after rises a
certain % ? Technical trading rules have not consistently outperformed the market on average
Two Apparent Contradictions to the Weak-Form EMH
1.
?
Momentum or persistence in stock returns
Tendency of stocks that have done well over the past 6 to 12 months to continue to do well over the next 6 to 12 months.
2.
?
“Contrarian” Strategies
Stocks that have done well over the past 3-5 year period, will do poorly over the subsequent 3-5 year period.
Thank You
doc_751805406.ppt
Autocorelation coefficient tests
EMPIRICAL EVIDENCE ON WEAK FORM EFFICIENT MARKET HYPOTHESIS
Weak-Form Efficiency
? Prices reflect all past price and volume data. ? Technical analysis, which relies on the past history of
prices, is of little or no value in assessing future changes in price.
? Market adjusts or incorporates this information quickly
and fully.
Weak-Form Evidence
? Test for independence (randomness) of stock
price changes
? ?
If independent, trends in price changes do not exist Overreaction hypothesis and evidence
? Test for profitability of trading rules after
brokerage costs
?
Simple buy-and-hold better
Autocorelation coefficients test
One of the most direct and intuitive test of random walk is to check for serial corelation. If stock prices exibit a random walk, the returns of ) cov(rt , rt ? k stocks are uncorrelated at all cov(rt , rt ? k ) ? k ) lags. ? leads(and? var[rt ] var[rt ] * var[rt ? k )
The join hypothesis states that all the serial correlation coefficients, are simultaneosly equal to zero.
Weak-Form EMH
? Runs tests
? looking for patterns in signs of returns ? i.e. + + - + - + ? Filter rules ? sell after falls a certain % or buy after rises a
certain % ? Technical trading rules have not consistently outperformed the market on average
Two Apparent Contradictions to the Weak-Form EMH
1.
?
Momentum or persistence in stock returns
Tendency of stocks that have done well over the past 6 to 12 months to continue to do well over the next 6 to 12 months.
2.
?
“Contrarian” Strategies
Stocks that have done well over the past 3-5 year period, will do poorly over the subsequent 3-5 year period.
Thank You
doc_751805406.ppt