Role of Information and Order imbalance on Stock Price Movements: Visualization Challenge Entry

Financial theories argue that stock prices move mainly due to arrival of new information or stock inventory imbalances (Order mbalances). However, it is hard for students or investors to visualize and assess the magnitude of these effects on stock prices. Hence, we propose to develop a visualization tool to demostrate these effects for better understanding of stock price dynamics. We use Lee and Ready (1991) trade assignment algorithm on the cleaned data to distinguish buyer-initiated and seller-initiated trades. And we arrive at order imbalance measure by taking the difference on buyer-initiated and seller initiated trades . We compute order imbalance for each minute and for each stock. We then remove those order imbalance data that overlaps with any news announcements. This will reduce the contamination while explaining the role of order imbalance on the prices. The data used for the analysis was the data provided by SIRCA for the competetion. The order imbalance was calculated using the number of quotes for buy and sell and matched with the trades. These figures were then plotted on high resolution graphs using Gnuplot. To identify possible locations where there are no overlaps of with news items, the data for whole days and in some instances for the whole period were plotted on very high resolution graphs at 8000x4800 pixels and visualized using an optiportal. Given the volatality of the market during the period of data provided, it was hard to identify a large number of samples that met our criteria. Once the periods were identified, graphs were generated for a two hour window (one hour before and one hour after) around the time under study. This was made easy by a generalized program that was given the stock in concern, the date and time as parameters. Utilizing these graphs it was easy to visualize the changes on a normal PC, the effect of order imbalance on the movement of price. The combination of the use of very high resolution graphs over the whole period on larger optiportals enabled us to see the bigger picture and easily identify the scenarios which would have otherwise taken other complicated analysis and programming to achieve. Once these points are identified, it was much easier to analyze the details around these scenarios