[dba-Tech] Rails Etc.

There are hundreds if not thousands of hedge funds. The one I worked for was
based in London and specialized in Euro stocks and commodities. Currently
Europe is in the same sort of turmoil that the US is experiencing. But
across the pond they seem to prefer throwing themselves in front of a train
rather than jumping out of buildings. I think it's quaint :)

Actually it's a good thing that I was just the developer rather than one of
the speculators. We would have got killed on the oil market. When a barrel
went over $140 I was sure it was on the way to $200. How do you spell
"smithereens"?

I did learn quite a bit about longs and shorts and puts and gets and that
some people have really a lot of money. The way it worked was this: a web
site written in C# was the FE to a SQL 2005 database. About 1000 users, all
traders located in various parts of the world, are called Participants. They
post Ideas, which essentially are "go long (or short) on stock XYZ. These
ideas are evaluated accrording to a rule base, and either acted upon or
ignored. "Acted Upon" means one of two things: we either bet on the idea as
presented, or in some cases do the opposite. This works by track the past
performance of the given Participant. Each Participant is playing with a
fictional million Euros. So the database itself is comprised of two worlds
-- fiction and fact. An idea is executed in the fictional world, and may or
may not be Acted Upon in the real world. Participants who do well (their
ideas pan out and some or even all of them are executed in the real world)
are rewarded for their ideas (in real Euros), and some rewards can be
substantial by my standards (i.e. hundreds of thousands of Euros).

When I got there, we received all our data about trades and prices via FTP
from Bloomberg, but we were in transition to a live feed.  I didn't see the
end of the project, but much of it. We received an average of 15 million
messages per day. Not all of these were deemed interesting so we discarded a
bunch of them (stocks we don't trade, informational messages, etc.) We were
primarily interested in about 1200 (eventually 2000) and the quantity and
price of the buys and sells on a given day. A C# app read the feed and wrote
the interesting bits to a series of text files. That's where I came in. I
wrote a SSIS package to peruse the directory containing the text files, and
then use Bulk Insert to slam them into a staging table, and finally do
(mostly) updates to the actual data. The reasoning behind this approach is
based on the sheer volume of incoming data. To update the actual data tables
directly from the feed would bring the system to its knees. In my
experiments I was totally floored by the speed of Bulk Insert.

Anyway, I'm not there anymore so this is all academic. It was fun while it
lasted.

A.

On Fri, Jan 16, 2009 at 10:57 AM, Susan Harkins <ssharkins at gmail.com> wrote:

> Um.... not THE hedge fund?????
>
> Susan H.
>