Abstract:
We have built an agent service-based enterprise infrastructure: FTRADE.
With its online connectivity to huge real stock data in global markets,
it can be used for online evaluation of trading strategies and data mining
algorithms. The main functions in the F-TRADE include soft plug-and-play,
and back-testing, optimization, integration and evaluation of algorithms. In this
paper, we'll focus on introducing the intelligent plug-and-play, which is a key
system function in the F-TRADE. The basic idea for the soft plug-and-play is to
build agent services which can support the online plug-in of agents, algorithms
and data sources. Agent UML-based modeling, role model and agent services
for the plug-and-play are discussed. With this design, algorithm providers, data
source providers, and system module developers of the F-TRADE can expand
system functions and resources by online plugging them into the F-TRADE.
