I'm ready to do an experiment mainly anazlyzing the volume-price relation,
But the purpose may be a little different as the past.
I'm trying to discern this phenomenon:
on Date T, stock SHA's volume increased by 2 times than last day and price increased accordingly.
In the following 5 trading days,stock SHA's price may continue rise or fall remarkably.
why this happened ?
my experimental steps:
I. test of the volume-price relation during bear market
test data from china’s stock market (shenzhen and shanghai both);
only sample the records during this period (from 2009-8-1 to 2009-10-1);
the chinese stock market jumped sharply during this sample period.
The shanghai composite index drops from around 3470 to around 2650;
So,I think it’s a good sample for bear market.
II. volume increase candidates;
select those records whose volume amount is 2 times bigger than the last day;
their later performance differentiated remarkably,part of their price rise amazingly while
part of their price fall desperately;I focus only on the good performance ones and try to analyze why
they vary from others gaining a better price even during the bear market;
III. data collecting and extracting :using my own program to fetch data from internet and process them on local computer.
The generated data are stored in an excel document named vol_bearmarket.xls;
IV. furthere research will be done on this data file;
primarily usual data mining algorithms.
V. the conclusion of the analysis will be published in the near future;
what variables I have to collect ?
now,I already collect nearly all the possible number datas.
open price,close price,high price,low price;
transaction numbers between certain money amount( under 1millon, between 1-5 million.......);
the details of transactions during each trading time .(9:30-10;30........);
the news of each listed company ;
the structure of shareholders;
But I wonder what else I need to take into considerations?
which variables may be more significant than others?
your past experience may tell the answer.
I need your suggestions because I know the pure quantitative method
may be futile, some subjective factors need to be taken into account .
thanks for all of your advice from the bottom of my heart!
you can directly send an e-mail to me.