[R-sig-finance] R and exchanges

[Adrian Trapletti]

>
>
>I sent a question to the general Help. I received an answer from Patrick
>Burns, and he suggested me to post it here. The question is how to use batch
>files that call R in an effective way. More:
>
>What I need is to interact with other applications(a connection to
>exchanges, like Eurex, Liffe, CBOT and eventually to something like ESB in
>FX). So, if I get a price from the exchanges, I want that R makes a
>calculation. But to get R connected to the exchanges is complex, so what I
>have thought is to have an external program that connects to the exchanges,
>then gets the prices and saves them to a txt file, then R is opened, the txt
>file is read, the calculations are performed, and the results are saved into
>another txt file. Then, another application reads the new txt file and sends
>a message to the exchanges.
>
>This is what I want to do.
>
>Does anybody have some ideas about it?
>
>Thank you!
>
>Jordi
>
>
>  
>
The companies I (used to) work for typically use(d) the following setup: 
Data receiver/collector gets the data over some proprietary API from the 
data provider (Reuters, Bloomberg,...), attaches filter information 
(e.g., a number between 0 and 1 representing the "credibility" of the 
data; even todays electronic markets produce strange information from 
time to time) and stores it in a database (e.g., mysql, binary 
files,...). Applications like trading systems connect to the data base, 
get the needed information (historical and realtime; for realtime 
distribution of information it might useful if the DB notifies the 
applications if new data arrives), compute something, and react somehow, 
e.g., send a trading order through another API (FIX or proprietary) to 
the broker/exchange for execution. The frameworks I have seen so far 
used mainly C, C++ and Java, but it would be no problem to use, e.g, R 
for the computation engine and use its DB interfaces or Java interfaces etc.

Best
Adrian