; Speeches
Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out
Your Federal Quarterly Tax Payments are due April 15th Get Help Now >>

Speeches

VIEWS: 32 PAGES: 7

  • pg 1
									Proposal Presentation

Introduction (30 seconds)

Hypothesis & Strategy (Matt) (1 minute)
Hypothesis: Assuming extreme price movements are driven by investor overreaction and
information asymmetry, we should make a profit investing in stocks that are likely to
exhibit a short term correction in price.

Strategy: Stocks with extreme past price movements and decreasing trading volume
should exhibit extreme price movements in the opposite direction.


Strategy Overview (3 minutes)
Step 1 - Rank 300 largest NYSE/AMEX stocks.

At the beginning of each year, select the 300 largest stocks by market capitalization. Only
large stocks will be included in the contrarian portfolio. In a few minutes will discuss the
rationale of excluding small stocks.

Step 2 – Filter stocks based on previous week lagged returns.

We will then divide the large pool of 300 stocks into smaller groups based on their past
week returns. Each filter group will have two breakpoints. If a stock’s return falls
between the two breakpoints it will be included in that group. The breakpoints will
determined by the overall sample distribution of the entire group of stocks. We will
replicate Cooper who, in a similar study, uses filter groups that target approximately the
1, 2.5, 5, 10, 25, 50, 75, 90, 95, 97.5, 99 percentile points.

We are only interested in securities with the largest past week price movement, as we
believe that stocks with the greatest past changes will have the greatest reversals. Several
filter groups will define big winners to big losers based on the magnitude of their lagged
returns. By only including assets in the loser or winner filter if it’s lagged weekly return
moved up or down by a specified amount we can pick out stocks with “large” past price
movements that could be investor overreaction. We can eliminate securities that
experience smaller lagged returns which are likely just noise.

Step 3 – Filter stocks based on % change in trading volume.
After separating all securities based on their returns, we will further divide them based on
volume. We will be looking at the percentage change in their trading volume. It is our
hypothesis that stocks with the largest decrease in trading volume will exhibit the greatest
reversals. Jared will explain the motivation behind this in a few minutes. Since
decreasing volume is related to only interested with stocks with the greatest decrease in
trading volume. To construct volume filters we analyze individual security weekly
percentage changes in volume, adjusted for the number of shares outstanding.

Step 4 – Form strategy portfolio.
At this point we have put all the stocks in NYSE and AMEX with 3 filters. First we have
eliminated small stocks by selecting the 300 largest capitalization stocks. Secondly, out
of the 300 stocks, we have selected out the ones with the largest past week price changes.
We will have a group of winners and a group of losers. Finally we put the stocks through
a volume filter. We are only interested in stocks with decreasing volume. Thus we will
have two groups of stocks…One with terrible lagged past week returns and decreasing
trading volume, “the losers” and one with incredible past week returns and decreasing
trading volume, “the winners”. We will buy the winners and short the losers.

Step 5 – Liquidate portfolio at the end of the week and repeat the formation
procedure.
Finally at the end of each week we liquidate our portfolio and repeat the formation
procedure.

Motivations
Overreaction
-Lehmann (1 minute)
-Allen B. Atkins and Edward A. Dyl (1 minute)
-Zarowin (1 minute)
- DeBondt and Thaler (1 minute)

300 Largest
- Conrad, Gultekin, and Kaul (1 minute)
- Forester and Keim (1 minute)
- Keim and Madhavan (1 minute)

Volume
-Wang (1 minute)

Motivations – Volume (1 minute)

According to Wang, the behaviour of trading volume is closely linked to the underlying
differences between investors. Investors differ in their information and private investment
opportunities. The informed investors have private information about the stock’s future
dividends but the uniformed investors must extract information from realized dividends,
prices, and other public signals. These two groups trade competitively in the stock
market. The informed investors trade when they receive private information about the
stock’s future cash flows. Since the uninformed investors cannot perfectly identify the
informed investors' motives, they face the risk of trading against the informed investors'
private information.
As the true state of the economy is revealed, the uninformed investors realize the
mistakes in their previous trading and trade to revise their positions. A high realized
return reveals that the uninformed investors underestimated the value of the stock and
underinvested in the stock. They will buy more shares and the stock price will continue to
rise. Thus, according to Wang a high return accompanied by high volume implies high
future returns. Therefore, reversals are more likely to occur when trading volume is
decreasing. When this is the case it is more likely that large price movements were the
result of a common overreaction in the market.

Data (1 minute)

To determine the effects of our filters on contrarian profits, we will examine a sample
consisting of Wednesday-close to Wednesday close weekly returns and weekly volume
for the 300 largest market capitalization N Y S E and AM EX individual securities in C R
S P between January 1, 1994 and December 31, 2006.

Data Analysis (3 minutes)

Matt mentioned our filters earlier but I think they could use some further detail could be
helpful.

We are forming portfolios this week by including stocks that meet the appropriate lagged
filter constraints last week. Although we have not yet defined our filter breakpoints and
ranges, I will illustrate our plan using the filter rules suggested by Cooper in his study.
According to him stocks with more than a 10% change in price and a 75% decrease in
volume exhibit the greatest reversals. Therefore, anticipating we will use similar filter
rules we would buy any stock with a loss of 10% in the last week and who’s trading
volume has decreased by 75%. We will short any stock who experiences a return of 10%
last week and who’s trading volume has decreased by 75%.

Each stock that meets the filter break points will be given an equal weight in the
portfolio. If there are no successful stocks, our strategy invests in the T-Bills for that
week. We will hold securities for one week and then liquidate the positions.


To determine the strength of our hypothesis, we will look at the mean returns for weeks
in which the portfolio holds equity positions. If the portfolio’s mean returns are
significantly different from zero it will be taken as evidence in favour of return
predictability. The null hypothesis of no predictability is that the mean return of the
portfolio equals zero. We will calculate mean-equal-to zero t-statistics to determine the
significance of our findings.

In addition, we will also compare our strategy portfolio with the returns of the stocks in
the other filter groups. To be more specific we want to prove that stocks with larger
lagged price changes and decreasing volume exhibit the greatest reversals. We will look
at whether stocks with less extreme price changes also exhibit reversals and we will
compare our portfolio with filter groups with high trading volume.

For the strategy portfolio and each control group analyzed we will find the mean and
standard deviation of the weekly returns for the 10 year period examined. T-statistics will
be computed to measure the significance of the difference between two means. We will
also put our strategy returns up against the returns of a buy and hold strategy of the 300
largest cap stocks as well as the returns from holding the broader market index.


We also plan to examine whether stocks with two consecutive weeks of losses exhibit
larger reversals then do ones with only one week of losses. Two articles, McQueen and
Thorley and DeBondt and Thaler suggest this may be the case. According to them it is
possible to obtain more accurate directional forecasts by conditioning on the information
contained in two consecutive one-week lagged returns rather than a single week lagged
return. We will compute a t-statistic for the difference between two means using our
strategy portfolio and a similar portfolio whose only difference is that it filters stocks
based on two week lagged returns.

The next attribute that we will analyze is how consistent profits are when the holding
period is extended. Lehman and Cooper both found that as the extreme loser and winner
long/short portfolio is held for longer horizons, the basic pattern that emerges is a greater
consistency of profitability.

I’d also like to mention that our strategy does explicitly take transaction costs into
consideration do to the large differences between private and institutional investors.
However, we do realize that they have an important implication for our results. We plan
to weigh the strength of our strategy returns against a range of transaction costs.
Data (1 minute)

To determine the effects of our filters on contrarian profits, we will examine a sample
consisting of Wednesday-close to Wednesday close weekly returns and weekly volume
for the 300 largest market capitalization N Y S E and AM EX individual securities in C R
S P between January 1, 1994 and December 31, 2006.

Data Analysis (3 minutes)

Yuri has discussed our filters earlier but I think they could use some further description.

We form portfolios in week t by including stocks that meet the appropriate lagged filter
constraints. Securities with the most extreme price movements will exhibit an extreme
movement in the opposite direction. Securities with decreasing volume experience the
greatest reversals. Although we have not yet defined our breakpoints, I will illustrate our
plan using the filter rules suggested by Cooper in a similar study. According to him
stocks with more than a 10% change in price and a 75% decrease in volume exhibit the
greatest reversals. Therefore, anticipating we will use similar filter rules we would buy
any stock with a 10% past week loss and who’s trading volume has decreased by 75%.
We will short any stock who experiences a past week return of 10% and who’s trading
volume has decreased by 75%.

Each stock that meets the filter break points will be given an equal weight in the
portfolio. If there are no successful stocks, our strategy will invest in the risk-free rate for
that week. We hold securities for one week and then liquidate the positions.


We will look at the mean returns for weeks in which the portfolios hold equity positions.
If the portfolio’s mean returns are significantly different from zero it will be taken as
evidence in favour of return predictability. The null hypothesis of no predictability is that
the mean return of a portfolio equals zero. Mean-equal-to zero t-statistics will be
calculated.

In addition we will also compare our strategy portfolio with the returns of the stocks in
the other filter groups. To be more specific we want to prove that stocks with larger
lagged price changes exhibit the greatest reversals. We will look at whether stocks with
less extreme price changes also exhibit reversals. To verify that stocks with the
decreasing volume are more likely to be reversed, we will compare our portfolio with
filter groups with high trading volume. For the strategy portfolio and each control group
analyzed we will find the mean (%) and standard deviation of the weekly returns for the
10 year period examined. Test t-statistics will be computed to measure the significance
of the difference between two means. By comparing our strategy portfolio with the stocks
in the other filter groups we will be able to judge the strength of our hypothesis. We will
also compare our strategy returns against the average returns of a buy and hold strategy
of the 300 largest cap stocks and as well as holding the broader market index.



We also plan to examine whether stocks with two consecutive weeks of losses exhibit
larger reversals then do ones with only one week of losses. Two articles, McQueen and
Thorley and DeBondt and Thaler suggest this may be the case. According to them it is
possible to obtain more accurate directional forecasts by conditioning on the information
contained in two consecutive one-week lagged returns rather than a single week lagged
return. It is their belief that if reversals are evidence of overreaction, then markets may
overreact to a greater degree for stocks that have experienced relatively longer periods of
gains or losses. To measure this, we will compute a t-statistic for the difference between
two means of our strategy portfolio and a similar portfolio whose only difference is
screening out stocks based on two week lagged returns.

The next attribute that we will analyze is how consistent the filter strategy profits are
when the holding period is extended. Lehman and Cooper both found that as the extreme
winner and loser filters are extended to longer horizons, the basic pattern that emerges is
a greater consistency of profitability.

Our strategy does not take transaction costs into consideration do to the large differences
between private and institutional investors. However we do realize they have an
important implication on our results. We will weigh the strength of our strategy returns
against a range of transaction costs.

								
To top