Historical Transactions Dataset
We have used the analytical results from the HFD of individual traders’ historical transactions to evaluate the collective behaviour of the agent-based models of traders in reproducing, to a certain extent, actual FX market trading activity.
This HFD represents individual traders’ physical transactions made available on an anonymous basis spanning 2.25 years, from January 1, 2007 to March 5, 2009. The HF dataset contains about 147 million transactions carried out by 45,845 different accounts trading in forty-eight different currency pairs under the same terms and conditions. Each transaction includes the following information: the trade type, the trade timestamp, the traded currency pair, the executional price, the units and the amount traded. However, in this study we used the historical transactions of EUR/USD only.The data is filtered to remove any possible unreliable transactions in terms of the traders’ actual activity. Filtering HFD must be customized, based on the structure of the data. The primary step in the filtering procedure that we conduct relies in two things: studying OANDA’s internal system storage procedures that can affect the consistency of the data, and understanding the nature and structure of the data. The major issue concerning the dataset is the extraordinary and sudden drop of transactions flow in October 2007 with an increasing number of accounts. Additionally, OANDA’s system storage procedures, storing one executed order in several transactions, and internal interest payment procedures creating dummy transactions. Carrying out several filtering steps, the number of transactions in the dataset was reduced to a total of 114 million transactions. To validate the filtering process, we have tracked the individual traders’ transactions from opening to closing their positions on each traded currency pair. In addition, we compared the intraday seasonality of the trading activity pattern against the reported seasonality in the literature. For a fully detailed description of the transaction dataset filtering process, refer to Masry et al. (2010).
The individual traders’ historical transactions dataset is very valuable for many reasons. Firstly, there is the existence of a large number of different traders represented by individuals, banks, central banks, hedge funds, and others. Secondly, there are a large number of high frequency trading volumes, which are rich in terms of observations, and some interesting patterns that are valuable for analysis, in order to understand the forces that drive market behaviour. Thirdly, there is the existence of different types of trades following different currencies. Finally, there is the existence of a wide variety of trading strategies, which have a great impact on the analysis and understanding of traders’ behaviour.
6.
More on the topic Historical Transactions Dataset:
- Historical Development
- Global Outbreaks from an Historical Perspective
- Historical surveys and descriptive works
- The historical record: how traditional is local government in Asia?
- Section 6 Emerging Trends
- Observations on the Problem of “Historical” and “ Non-historical ” Nations
- Simulating Social Life After Prehistory
- 17 Tin Traders
- Traders and Labourers
- Data Collection