Dynamic Relations between Order Imbalances, Volatility and Return of Top Losers
Date Issued
2007
Date
2007
Author(s)
Kuo, Po-Hsin
DOI
en-US
Abstract
For many years, investors have been looking for a reliable indicator to predict the movement of stock prices. Many researches show that order imbalances have a significant relationship with stock returns, especially in speculative stocks. In this paper, we want to examine the relations between order imbalances, volatility and stock returns. Then, we try to find the predictability. Finally, we develop a trading strategy and see if it can earn profits.
First, we apply GARCH (1,1) model with and without volatility to test whether it can fit our time series data. In our research, we find that GARCH (1,1) can capture the properties of our sample stocks in the two methods. Then, we use multi-regression model to see whether contemporaneous and lagged order imbalances have significant influences on stock returns. We find contemporaneous order imbalances have positive effect and lagged–one order imbalances have negative effect on stock returns. While controlling for contemporaneous order imbalances, only lagged- one order imbalances have significant and negative effects.
Then, we want to test if there is a small firm effect on our data. After our empirical test, we can see that order imbalance and market capitalization have positive relation. However, the relation is not significant, thus we can’t say the small firm effect exist from our test.
Finally, we develop a trading strategy and wish it can make profits. We short the stocks when order imbalance is negative and buy back when order imbalance is positive. In our test, we notice that if we don’t select the volume, there will be no abnormal returns. However, if we sift our data from trading volume, that is, above 99% volume, we can find that we will earn profits. Therefore, we can say that our trading strategy is useful.
Subjects
最大跌幅投機型個股
order imbalance
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-96-R94723088-1.pdf
Size
23.31 KB
Format
Adobe PDF
Checksum
(MD5):f3d6b8f3bbd918b6e66f560392c2391b