20.11.2012 Views

recent developments in high frequency financial ... - Index of

recent developments in high frequency financial ... - Index of

recent developments in high frequency financial ... - Index of

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

256<br />

simply aggregated. Even after aggregation we have a large number <strong>of</strong> observations<br />

for each stock, rang<strong>in</strong>g from 572,261 (BA) to 1,170,608 (GE). F<strong>in</strong>ally, note that,<br />

unlike many studies that work with <strong>in</strong>traday stock prices, we do not remove the<br />

diurnal patterns that are present <strong>in</strong> durations. This is a more critical issue if the<br />

focus <strong>of</strong> the study <strong>in</strong>volves the dynamics <strong>in</strong> duration data, but our focus is on<br />

the contemporaneous relationship between durations and price changes, and so<br />

we choose to exam<strong>in</strong>e the data with as few adjustments as possible. In addition,<br />

we do not estimate the distributions at a f<strong>in</strong>e enough grid on durations for the<br />

removal <strong>of</strong> diurnal patterns to affect our results <strong>in</strong> any important way.<br />

In Fig. 1 we show histograms <strong>of</strong> price changes, duration, and signed volume for<br />

IBM. Also displayed is the histogram for signed duration (duration multiplied<br />

by the sign <strong>of</strong> the trade). The distribution <strong>of</strong> price change appears to be very<br />

symmetric. The distribution <strong>of</strong> signed durations is symmetric, so the distribution <strong>of</strong><br />

durations is similar for both buyer- and seller-<strong>in</strong>itiated trades. The histogram for<br />

trad<strong>in</strong>g volume is not <strong>in</strong>formative except to <strong>in</strong>dicate the presence <strong>of</strong> a few outliers.<br />

However, apart from these the distribution is also fairly symmetric. A more detailed<br />

picture can be obta<strong>in</strong>ed from Table 2. Here we show every 10th percentile <strong>of</strong> price<br />

change, (unsigned) duration, duration for buyer- and seller-<strong>in</strong>itiated trades, (unsigned)<br />

volume, and volume for buyer- and seller-<strong>in</strong>itiated trades. For all stocks the<br />

distribution <strong>of</strong> duration and volume across buyer- and seller-<strong>in</strong>itiated trades are<br />

very similar to the overall distribution. In fact there is almost no correlation between<br />

the sign <strong>of</strong> trade and duration. Volume is substantially larger for the more<br />

x 105<br />

3.5<br />

3<br />

2.5<br />

2<br />

1.5<br />

1<br />

0.5<br />

0<br />

0 50 100 150 200 250 300<br />

x 105<br />

5<br />

4<br />

3<br />

2<br />

1<br />

0<br />

-150 -100 -50 0 50 100 150<br />

1.5<br />

1<br />

0.5<br />

x 105<br />

2<br />

0<br />

-300 -200 -100 0 100 200 300<br />

a Duration b Signed Duration<br />

x 105<br />

10<br />

8<br />

6<br />

4<br />

2<br />

0<br />

-12000 -10000 -8000 -6000 -4000 -2000 0 2000 4000<br />

c Price Change d Signed Volume<br />

Fig. 1 Histograms <strong>of</strong> duration, price changes, and volume (IBM)<br />

A. S. Tay, C. T<strong>in</strong>g

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!