Hi all,
I am looking for a trader from Western Europe (a priority, since I expect not only remote interaction, but will also consider offers from the USA, Canada, Australia) interested in the joint development of an automatic trading system. Priority is given to traders registered on this portal >10 years ago.
I have been developing an automatic trading system since 2004. I started working 20 years ago using the Oanda API, the Esignal data provider using DDE technology (there was complete automation of data uploading through programming keystroke events and mouse buttons without human intervention). Around 2006, I abandoned the Oanda API and switched to the IB API in terms of the availability of much more tools for trading in IB at that time and the cost of trading. All developments at Oanda were transferred to the IB API, but with IB I began developing a system for trading futures contracts of currency pairs and commodities. Over time, I switched to forex instruments (currently CFDs in IB) and shares. At the moment there is a large automatic trading system without human intervention (only the weekly launch of TWS, taking into account mandatory two-factor authentication is required) starting from uploading data in real mode, on several servers from several different TWS (currently the system works with 3 instances of TWS), to if one data source is disconnected, there is a switch to another, data filtering (data is filtered using mathematical analysis in terms of “defective” ticks), synchronization (merging) of ticks from three data sources (InteractiveBrokers has a problem with data quality in real mode, on three Different TWS may receive slightly different data; as a result, ticks from three sources per unit of time are analyzed and the optimal one is selected), data preparation to reduce the load and volume for trading algorithms. Trading applications operate in automatic mode, there are options for automatic start, stop, notification to the phone about all events, balance control, automatic transition to summer/winter time, exclusion of holidays and weekends, etc. To minimize the load on the server (and from the point of view of a large number of ticks), there is a special module for caching the necessary ticks, and outdated and irrelevant data is transferred to auxiliary database tables, which are accessed when necessary. On weekends, data for the week is downloaded through history (in case of force majeure during the week), as well as automatic archiving of the state of the entire system (applications, settings, database) to several backup servers for quick recovery in the event of a breakdown, shutdown, or other force majeure. There is an archive of data with the system development process including source code, ticks, trading results and etc. for the entire time since 2004, from which you can study the process, dynamics and progress of development. Back in the day, reverse testing applications were written. The system works with a MySQL database, C++ code, the load on not the most powerful servers is up to 10-15%, even during “rush hours” when there is news or some events occur on the stock exchanges. The results at the moment are a stable plus for currency pairs over the past year, even on small volumes (even when trading 1 lot for 20,000-35,000 units, depending on the pair, taking into account the commission for entering/exiting a transaction of about $4).
This is brief, I'll tell you more detailed information in personal communication.
