Trading Prerequisites - Doing it Right

You will have to keep grinding a little while longer, but don't worry, those things are only a few lines in R, almost everything in R is 'vectorized', meaning that most functions work on whole vectors. Quants usually work with time-series objects (zoo or xts), but you will get to that stuff later...

Here is an example of what can be done...
http://blog.fosstrading.com/2009/07/...ternative.html

Since it has not been mentioned yet in this thread, I'll throw in one more recommendation of something to learn a bit about: XML. It is not an analytic tool, but the vast majority of financial data today moves via (compressed, then encrypted) XML. For example, there is FPML, the schema which move swap and derivative trades and data between dealers, FIX, which moves forex, RML, which moves risk management data, and the list goes on. Any time you see ??ML, it's a set of specifications for the content of xml nodes. In fact, what you're now reading is a text node of an xml document, because html is a specific xml scheme.

If you have an account with a broker such as ducascopy, and use their automated facilities, data moves in XML (in fact, their feeds to your desktop do so, even though you don't realize it).

Excel understands and will natively display XML, and Microsoft, Oracle, and other relational database products natively import and export records in XML.

It's also used extensively in systems' internal "quasi data base" activities, where the number of entries is too small to need the big relational databases, but too big for programs to do sequential search. Here's an example. Say you have a price data event handler, which is receiving prices as they come from an internet source, like your broker. Every time a packet arrives at the event handler, it produces an xml "node", that might look like:

<price pair='GBPUSD' val=1.51234 side='bid' ssm70=1294534214 ms=476/>

That's pretty self-explanatory, except ssm70, which is the number of seconds since 00:00:00 GMT on 1/1/1970. (This is a standard time stamping particularly in Java apps) when the data arrived, and ms is the fraction of that second.

There is a query syntax called XPath, which can be invoked from any of the languages mentioned in this thread, which allows you to effectively query xml structures. For example, if you wanted all the bid price quotes larger than 1.5135, an XPath syntax would look something like:

xPrices.SelectNodes("./price[pair='GBPUSD' and @val>1.5135 and @side='bid']");

You don't really need any books on XML; you can find lots and lots of examples, blogs, and code on the web. The entity that maintain the standards is the World Wide Web Consortium, at w3c.org, and they have a lot of information as well.

As far as languages go, I second the Python, as well as R. But I'd also say, don't flush Visual Basic just yet. There is still a huge number of Excel-based applications out there, both as furnished API offerings, and throughout the front office operations of the financial industry. Excel still, and will continue to, run VBA - "..for applications". Interactive Brokers, for example, provides their event-driven API in all the high-levels (Java, .Net, C++), but also in VB6 and in Excel itself.

By now, your hair in probably blowing back from the breeze caused by all the information dumped down in this thread. That's okay, because ultimately, you just need to sift through it all, and pick something that (a) you are comfortable with, and (b) you can "lift" lots of examples and snippets of from the net. Choosing a language and a platform is a little like a choosing a puppy at the pound: for reasons you can't explain before or after, one of them will just be the right one.

EDIT: If you end up with R as your choice of platform (and it's a good one!), an xml interface can be found in reference at rss.acs.unt.edu/Rdoc/library/XML/html/xmlTreeParse.html

i think ull need to have a fast server for this,i have watched a doco on utube
and the companies that trade they usually trade in miliseconds eg,brokers and banks,their servers are very close by i think in the same place where they keep the exchange keeps their data,what they do i heard is see the orders and then buy to lift the price up a bit then sell u the stock at a higher price

i also saw in the documentary that its not 100% win rate,its about 60%

Hey OldQuant,

Thank you for chiming in. You're just about right with the wind blown hair idea. I've had an interesting last couple weeks experimenting with differant applications, half the time I have no idea what I'm looking at. I feel in many ways like I did when I turned 16 back in the day and didn't have money to fix my car so i had to learn it all myself. The good thing about it is that I'm a mechanical thinker. Much of this journey is simply learning the mechanics of it all.

You actually answered a question I've been meaning to ask, simply, "in what format does data arrive?" Because I'm a complete newbie on this side of things I need to understand what needs to be there on my side to receive that data so I can process it. It's gonna take quite a while for me, but I love it.

Welcome to FF, BTW, doesn't look like you've been around too long, but I've enjoyed reading what you have to say.

Thanks again,

Jim

I beleive you are describing high frequency dealing (HFD), and yes time would be a critical factor in such a system. I'm more into trading obscure patterns that occur within price data, and also orderflow trading that leans on data about orders that have already traded. The act of automating such trading systems as the ones I'm describing is about getting the best prices and overall efficiency, so time is still a crucial factor but not to the extent of HFD. Once the program recognizes a pattern an order is sent out immeadiatley--faster than any human could do by hand or voice. I can explain it, I just can't carry it out...yet. Thanks for the post.

Hey Guys,

Is there any indicator library out there so that I can use it in c#? Thanks.

There are some C# libraries.

But you would have access to a much wider list of indicators by writing an adapter for FXCM Strategy Trader or NinjaTrader, which both use C#.