Hi,
dukascopy is one of the best ECN. I have a friend that is trading with them and he is really happy to deal with them.
One day maybe I will go with them.
ECN for me is still a dream, you need big bucks.

Wish you all the best in trading.

Have a nice day!

Do you have a link on IB API? Many API are proprietary and my goal with ECN programming is to support a common protocol. MAN supports FIX 4.2 it seems, so does Dukascopy. Both groups require a high deposit. I'm looking at MB Trading as my initial ECN target because the deposit requirement is $400 and they seem to have a decent system for execution, plenty of volume. (Anonymous trades, volume hidden, Level 2 very active with 5000 full lots per 10 seconds visible non-news.)

I'm rounding up programmers that are interested. We might be able to have some MT4 compatibility for source files but my initial testing will be working on a rewritten EA implementing the protocol library directly. The protocol library is written in C++, I'm thinking of Python for EA development with a .Net version implemented automatically through IronPython -> IL -> C#.

I'm asking now about FIX versions, MBT might use Fix 4.4 with proprietary extensions. FIX 4.2 seems the most common.

Minimum account size w/MBT: $400.

Speaking of which, here is a post from EFX/MBT

http://www.nondealingdesk.com/showpo...46&postcount=1

Justo go to http://www.interactivebrokers.com/

Choose "software" tab. Inside there are many API links.

__________________

Quick Elite Links:TSD Elite Manual Trading | TSD Portfolio | Quick Links: TSD Calendar | Suggestions to improve the forum

I don't like the requirement of running their little jar package for the actual order entry. The CTCI requirement to avoid the jar package (TWS) is not too bad for a dedicated connection.

What really bothers me are these people saying one place "we're compliant with FIX 4.1 and 4.2", then later somewhere else, "We have pieces of 4.3 and 4.4 protocols" and then, "we have proprietary extensions to the FIX protocol." I suppose that means as a developer I'd start with FIX 4.2 and then merge in some of the 4.3/4.4 features. The proprietary 'ambiguities and extensions' part will probably be the most annoying, least documented and implementation specific part.

The functionality from 4.3 is probably swap support.
The functionality from 4.4 is probably multileg orders and position reporting.

If IB and FIX are used, we'd probably save time by supporting CCTI at $100/mo. I am not thrilled with the idea of using FIX for quotes and TWS for orders.

Here is the current plan:
Write an implementation of the FIX protocol in Python, with MySQL storing the data as if it were a charting application. I'm working on this with one other programmer, we won't release anything until we have some code. A few programmers have offered to help, we will be doing everything in Python and MySQL.
Quote Protocol
Login
Interpreted data
Writes to MySQL Database

Order Protocol
FIX 4.4 base
Proprietary Extensions
Generics for handling order types
Can fetch data on position sizes, balance, etc.

Indicators
Written in MySQL (SQL, possibly some Transactional SQL, Possibly stored procedures)
Python Indicator Interface (PyIndicator)
Indicators can also be written in Python
Can be time sensitive and historically time sensitive. (In the past X weeks of data, how many Tuesdays at 8am NY time did the price go up if 8am London time went up?)

Automated Trading
Uses Indicators and Quotes to place orders
Written in Python, makes PyIndicator calls
Sends orders to OrderProtocol
Has internal data structures, objects

Account Information
Fetches information from OrderProtocol
Used to pick lot sizes
Historical performance records
Historical record of trades

Daraknor,
Will customers of your hosting datacenter service have access to the indicators you write to utilize your "current plan" as you refer to it?
Hope so!

Yes that is indeed the purpose. I'm going to start with the EA that have very few indicators and then slowly add more exit strategies and indicators. We might also be able to do MQL to ECN conversions, and suggestions to increase profitability based on the advantages of the ECN.

We get to see how many orders are stacked up above the ask and below the bid (tens or hundreds of thousands mini lots) and we can look at the streaming data to determine if the activity is on the bid or the ask. (Which side dropping their price, and how often?) This could also lead to some interesting see saw style calculations.

The main advantage of course using MySQL over a chart is that the data can be selected and managed in more interesting ways, like the price action divided up by market. Make 100 graphs of the last JPY, London and NY open side by side, etc.

Best part in my mind: there is no broker to fight you. And of course, having spreads of 1 pip or less is also nice

Hi daraknor,

You might fly over my reply because it's my first post but I stumbled across this thread and found it really interesting because we have the same goals.

So I'll keep it short and concise.

I developped a system based on woodie's CCI with some other indicators relying on volume.

So I needed to find the best broker with the biggest volume and came across dukascopy. So i started demotrading my system with them.

But as it is based on the metatrader platform, I had bad results depending on which metatrader brokers signals I used. Because as you said some brokers distort their tick data way too much.

So I called dukas which provides a FIX API to trade through and also some agents and wanted to integrate their datafeed into metatrader...and of course I would not get support for that...but they have an agent or dll which allows metatrader to post trades on their system so an expert advisor could do the job.

I began then comparing dukas 1m chart data with that of many metatrader brokers. I wrote some perlscripts to do the job and finally came across a broker whose datafeed is close enough to use and demotested it manually for some time with much better results.

In conclusion for me the best bet is to use a metarader broker demo datafeed to trade live on an ECN without rewriting the whole code in java or whatever. Also there might be pros and contras to dissociate different brokers interest.

I really like your approach going the python way but maybe perl is a simpler option and I don't really see the need of using a database like MySQL to store the tick data because the candles data is just so straightforward and not linked in a way you would need a DB to spare redundant data.

I hope you read on to now and would like to hear from you.

best regards

So Svifx,
Which metatrader broker's data measured up the best against Dukas.