ACT: Arthika's Algorithmic Currency Trading Platform

A complete, end-to-end, plug and play platform that enables traders to develop, execute, monitor and analyze complex algorithmic trading strategies with a very low time to market.

A scalable, state-of-the-art low-latency architecture

FIX and trading API servers connected through a proprietary high performance bus. [More...].

High performance and focus on execution

Real-time, continuous, detailed and systematic performance logging and analysis.[More...].

Real time risk management

Mark to market values of open positions at all levels. [More...].

Liquidity access and management

Arthika’s ACT platform connects to FIX interfaces provided by all major venues in the market. [More...].

Flexible, full accounting model

Reconciles trading activity, cash and exposures with the prime broker’s back office accounting. [More...].

Powerful yet easy to use API

C-Language API for hosted strategies or multi-language, mobile ready, state-of-the-art REST API. [More...].

Advanced backtesting and analytics

Real time logging of full book real time ticks, exposures, and orders. [More...].

Architectural overview of the platform

A scalable, state-of-the-art low-latency architecture. FIX core and trading API servers connected through a proprietary high performance bus

The central component of the platform is the core server, which is hosted and administered by Arthika but is used and configured by the client through a GUI to which the client connects from a standard browser.

The core platform server is the piece that connects to FX venues such as Currenex or LMAX, typically through cross-connects or dedicated lines to minimize latency. It maintains the streaming and trading sessions, keeps the prices and order books, the accounting model, and the risk management system, making all available for client programs when the connect to it.

Clients then can hire hosted client servers (one or several), connected to the core sever through a fast UDP proprietary interface, to host and run their trading programs achieving maximum performance. These clients servers are hosted in the same LAN where the core is, but they are administered and maintained by the clients (or by Arthika upon request), so the client’s intellectual property is protected. The platform provides as well a RESTful API to interact with the core platform server through webservices, which can reduce the development effort (e.g. in Java or C#) at the expense of lower performance.

Liquidity access and management

Arthika’s ACT platform can connect right away to FIX interfaces provided by all major venues in the market

Prices are received and modeled on a full book basis, receiving multiple quotes in real time, and attributed to individual liquidity providers when available. Price quotes are timestamped and evaluated individually, and they are related to previous quotes by the same liquidity providers effectively forming “quote chains”. This way, the transient evolution of the prices published by individual liquidity providers or market participants can be observed, modelled and used for complex trading strategy in real time.

Arthika’s platform can maintain several FIX connections at the same time – both different venues on the same prime broker account, or in different accounts. All price quotes from all the different venues are available to the user, both differentiated by venue and also aggregated in a single synthetic consolidated full book, where all available price quotes are sorted in real time. Aggregated full book and top of book levels are also maintained at all times.

When different venues are available, smart routing functionality is provided to route market and pseudo-market (i.e. limit orders chasing a given set of available quotes in the book) orders optimally at a lowest possible cost, based on the available prices in real time. Heuristic liquidity pre-trade characterization is used to estimate and ensure executability.

The screen above shows a full consolidated book with 7 simultaneous trading interfaces, all with different top of book levels, and with several quotes at each price.

Arthika’s ACT platform can connect right away to FIX interfaces provided by all major venues in the market, such as Currenex, Integral, Hotspot, LMAX, FXCM, Liquid, and FxOne (*). Arthika can develop FIX connections to additional venues upon request, usually in about 2-3 weeks.

(*)This statement does not represent a formal endorsement from these companies

High performance and focus on execution

Arthika’s platform systematically measures performance on a detailed manner and on a continuous basis, in real time.

Tick-to-order is systematically measured through generating “dummy” test trades on price ticks, at a user-specified pace (e.g. 1 second). Histograms are built and rendered in the administration GUI, as shown in the following screen (the horizontal axis is in microseconds).

The total tick-to-order total latency (* from the core platform’s input port to its output port, i.e. not counting cross-connections to the venue) that can be achieved with the platform depends on the exact configuration, but medians can be below 100us when strategies are compiled as libraries and uploaded to the core server. Medians in the 200us range are achieved when user strategies run in different servers (hosted by Arthika within the same LAN).

Besides tick-to-order, the ACT platform provides extensive trade profiling and liquidity characterization / analytics capabilities, useful for subsequent liquidity trimming: execution timing and last look provision detection, fill rations, slippage measurement, etc. Sample strategy libraries provide as well monitoring and fail-safe standard mechanisms, that can be enabled by the user to keep trading activity between safe limits – e.g. maximum drawdown, too many trades in a given time, too many failed or rejected trades, too high latency / delayed prices, minimum and maximum order sizes, margin calls, etc.

Flexible, full accounting model

ACT provides a very powerful and flexible accounting model to reconcile trading activity, cash and exposures with your prime broker’s back office accounting system.

The system provides a full accounting model of the client’s prime brokerage account, consisting in i) a set of cash positions in all different currencies (both initial client funds and cash resulting from settlement of positions upon trading activity), ii) a set of open positions in different currency pairs at a given price (modeled as single swap positions for each currency pair), and iii) a set of live (pending) orders in the venue but not yet fully executed. This “prime broker” accounting level should reconcile with the client’s broker backoffice view.

The client can then define one or more “equity pools” (EPs), containing only a part of the cash and the positions in the prime broker. EPs are used to confine users and their trading activity to only a part of the funds available for trading in the whole system, thus avoiding contamination between them.

Accounting model-Prime Brokers and equity Pools

Trading strategies are then modeled as individual “accounting units” (AUs), which hold the individual positions generated through trading by each strategy, plus the cash generated upon settlement of these positions. Mark-to-market values of open positions and cash amounts add up to the individual strategy’s P&L since inception to a given moment.

As a result, different trading strategies are tracked separately as individual accounting units. Strategies P&Ls can then be set up to compound when they trade on the same pool, or can be separated into differentiated pools to avoid contaminating each other

The diagram outlines the accounting model used throughout the platform:

Real time risk management

Mark to market values of open positions at all levels

ACT provides real time risk management functionality right away. It calculates mark to market values of open positions at all levels (accounting unit, equity pool, and broker account), using representative system prices for all currencies and currency pairs based on the prices available from the different trading venues.

Based on leverage values as configured by the client, the systems calculates used and available margins at all levels, and determines margin call situations when they occur. Such situations generate alarms, and automatic actions can be triggered to correct (e.g. stopping all unsupervised automatic trading activity and close all positions).

This real time risk management capability is one of the key advantages provided by Arthika’s platform, as it combines powerful venue connectivity with an advanced accounting and risk management layer avoiding costly / lengthy backend integrations for real time trading.

A powerful yet easy to use API

C-Language API for hosted strategies or multi-language, mobile ready, state-of-the-art REST API

Arthika Currency Trader platform is fuly available to traders through a complete set of API's.

For latency-sensitive, high performance applications, the API is delivered as C wrapper with direct access to the strategy map. This wrapper, which runs in a hosted client server within the same LAN as the Arthika core server, delivers callbacks on price tick and execution reports, typically hundreds or thousands of times per second.

For less demanding applications, the API is delivered through webservices as a RESTful interface. In this case, applications can be delivered in webservice-friendly languages such as Java or C#, and the strategy programs can run either hosted or in the client’s premises.

In both cases, the API provides real time access to all key data structures: Top of book and full book quotes with price, liquidity levels, provider attribution and # of updates.

Sorted arrays of top of book and full book prices Exposures to cash, swaps and equivalent cash risk Mark to market valuations of equity and positions, and used, free and reserved margins – List of trades alive and recent historical trades, with extensive timestamping Key configuration data.

The API also provides advanced, high level functionality such as: Sending, canceling and replacing live orders Smart routing including heuristic executability estimates Execution time statistics and last look provision estimates High level “macro commands”: auto hedging, close all positions at minimum cost, cancel all pending orders, …

Advanced backtesting and analytics

Real time logging of full book real time ticks, exposures, and orders

ACT implements a powerful database that records in real time all the information flowing through the system, including full book real time ticks, exposures, and orders.

Prices are specific for each client, depending on the liquidity providers they receive prices from. Prices are recorded with the maximum possible granularity, i.e. on a disaggregated full book basis.

Quotes are timestamped both with the original venue timestamp and with the Arthika platform timestamp.

The platform implements as well a playback interface that shows, tick-by-tick with a microsecond resolution, past trading behavior at the user’s command – like a “step by step” debugging session. Trading strategies can then be recreated on a tick-by-tick basis, which is useful for debugging purposes.

Historical information can be queried by clients, who can download right away very detailed information for backtesting and analytics purposes. Detailed reports can be downloaded right away on screen and on CSV and excel files through the GUI.