How to master the special challenges of implementing an algo trader

Publish date:

This article provides an overview on the implementation specifics for purchased algo traders and is divided into the definition of the trading strategy, a chapter on IT-Architecture, followed by an explanation of backtesting, the organizational implementation, the go-live and finally, information on regulatory and legal aspects.

Algo traders provide a state-of-the-art solution for efficient trading activities (see blog algo trader – success factors). Trading companies aiming to implement this technology have to master various challenges by deciding whether to make or buy an algo trader (see blog algo trader – make or buy). This includes adapting their trading strategies, followed by a complex IT implementation, which involves people and a change of business processes, as well as the consideration of legal and regulatory aspects.

Definition of Trading Strategy

Possible strategies for algo trading, in terms of mathematical methods to generate trading signals, includes mean reversion, trend following, market making and statistical arbitrage. Furthermore, artificial intelligence (AI) can be applied to increase the accuracy of price predictions. Mean reversion and trend following are based on rather simple algorithms and do not require large historical data sets to calibrate. Market making and statistical arbitrage in combination with AI are especially complex calculation strategies, which need comprehensive historical input data.

The first challenge is to identify a trading strategy suitable for the organization and its existing digitalization state. The IT system landscape must meet the specific requirements for algo trading. In addition, the trading and IT personnel must establish the processes and know-how needed to execute algo trading.

Within this context it is recommended to start with a simple algo trading strategy, which does not require an overhall of the IT-Infrastructure and business processes. Furthermore, the statistical significance of simple strategies, which is required to prove the positive impact, can already be tested and proven via Excel calculations. This supports a low initial investment together with a reduction of the risk of trading losses due to calibration failures. This approach is particularly suitable for risk averse companies.

Technical Implementation of an Algo Trader

The technical implementation starts with the architectural setup of the of the algorithmic trading platform (ATP). The primary components include the market data adapter, which receives historical and life market data and converts it into an appropriate format for the complex events processing engine, containing the algo trader and the backtesting engine. Finally, the order routing system converts and encrypts the trading signals and sends these to an exchange (1,2).

Figure 1 provides a conceptual IT architecture of the ATP. Most data layers and interfaces require adjustments with respect to the specifics of an algo trader. The data source layer needs to be enriched with historical data of the granularity expected by the algo trader. Network lines and interfaces need to be capable of maintaining the ongoing availability and transfer of trading and market data.

Figure 1 Exemplary architectural design embedding an algo trader
Figure 1: Exemplary architectural design embedding an algo trader

It is recommended to ease into algorithmic trading by implementing a simple and comparably secure algorithm (e.g. mean reversion) and to start with a limited set of deals. Complex strategies like statistical arbitrage (especially those utilizing AI) require significantly more historical data for calibration and an infrastructure capable of handling high data volumes. Hence, intense investments are needed at the project start and the risk of significant trading losses exists if the algo trader and the periphery systems are not set up correctly.


A major aspect of algo trading is an ongoing risk and performance evaluation to constantly identify if the chosen strategy is effective. This is where a backtesting of the strategy comes into consideration. Backtesting is a historical simulation of the algorithmic trading strategy to evaluate past performance. Most algo trading tools already provide backtesting functionality.

Organizational Implementation

Automated trading has shifted the focus of human intervention from the process of trading to a more behind-the-scene role. In terms of algo trading a focus on market and operational risk monitoring becomes more important. Any failure of technology, network or data streams can be disastrous. Multiple level checks for data are required to capture any anomalies and stop the strategy instantly if something is wrong.

To run a complex algo trading strategy successfully an interdisciplinary team of professionals continuously running the trading desk and the related IT-Infrastructure is needed. These are traders include strategists, IT professionals, network managers, risk managers and legal teams, who all need to work together. However, it is recommended to start with a simple strategy for which a team of IT professionals and traders/strategists is sufficient.


To prove the concept a test sequence based on a dedicated test system should be performed as part of the go-live. This test needs to include all operation modes of the algo trader based on real life scenarios including emergency situations.

For the go-live it is recommended to step into a parallel operation mode by generating the trading activities for the algo trader. The trading personnel validates the algo trader’s suggestion prior to executing an order manually.

Once the algo trader runs successfully, the switch to fully automated trading can be made. In case the ongoing post-trade analysis by means of backtesting does not ensure the expected results, it may be required to (re-)calibrate the algo trader to a new market situation and/or to adjust the price signals initiating the order events.

Legal, Regulatory and Compliance Aspects

All legal and regulatory aspects applicable to manual trading are also relevant for algo traders. The legislation in Europe supports further automation of the trading business to improve transparency and accountability (3). Current algorithmic trading should take into account certain aspects related to the REMIT and MAR regulations. However, more comprehensive legal and regulatory frameworks as an adaptation of the respective financial regulations also to the energy business are expected. (Please follow our next blog on compliance.)

Furthermore, companies applying algo traders must undergo regular audits. For the audit, it is required to maintain order logs, trade logs, control parameters, etc. for the past few years.

It is recommended that trading companies build up a network and exchange up-to-date legal and regulative information and invest in appropriate measures to set up the algo trader accordingly.

Conclusion / Recommendations

  • Start small and step into algorithmic trading to follow the learning curve.
  • Validate if the chosen strategy provides the expected results before beginning any major technical and organizational transformations.
  • A rather simple trading strategy, such as mean reversion, can already be implemented without extensive IT investments.
  • Perform a stepwise go-live from manual, via semi-automatic, to full algorithmic trading.
  • Take the necessary legal, regulatory and compliance measures into consideration.


  1. Algorithmic Trading System Architecture
  2. How Does An Algorithmic Trading System Work?
  3. The Growth And Future Of Algorithmic Trading

Weitere Posts


Agile IT? Nicht ohne die passende Organisationsstruktur!

Philipp Rass
Date icon Januar 16, 2019

Die Schaffung einer effizienten und agilen IT ist nur möglich, wenn die...

Capgemini Invent

Die Zukunft des SCM – Träumt nicht davon, erschafft sie

Ralph Schneider-Maul
Date icon Januar 14, 2019

Die Disruptionen im Supply Chain Management – Eine Sicht von Capgemini Invent...

Capgemini Invent

Cybersecurity Awareness – ein Schlüsselthema der Digitalisierung

Marlene Nowakowski
Date icon Januar 9, 2019

Wieso wird Cybersecurity Awareness dank zunehmender Digitalisierung noch wichtiger?


Mit dem Fortsetzen des Besuchs dieser Website akzeptieren Sie die Verwendung von Cookies.

Für mehr Informationen und zur Änderungen der Cookie-Einstellungen auf Ihrem Computer, lesen Sie bitte Privacy Policy.


Cookie Information schließen