Financial Data Analytical Platform for a Large Investment Management Company

AI-based data analytical platform for wealth advisers and fund distributors that analyzes clients’ stock portfolios, transactions, quantitative market data, and uses NLP to process text data such as market news, research, CRM notes to generate personalized investment insights and recommendations.

Technologies / Platforms

About the client

The client is among the top 20 investment management companies in the world, with assets under management exceeding 1.5 trillion dollars. They manage and distribute diversified financial products such as mutual funds, ETFs, ETNs, and managed accounts, constructing custom portfolios across different asset classes for both institutional and retail investors.


The client employs hundreds of investment advisors who monitor the market situation, clients’ portfolios, transactions and recommend their financial products to institutional investors such as banks, insurance companies, and pension funds to help them achieve better returns or hedge against certain market risks.


Financial data analytical platform for a large investment management company


Effort and duration

Ongoing, since September 2018


Engagement model

Time & Materials


Project team

1 Project Manager, 1 Business Analyst , 5 Developers, 2 Data Scientists, 2 QA Engineers

Tech stack / Platforms

Business challenge and vision of the system

The client experienced growing competition caused by the digitization of the financial services industry. Financial advisers, overloaded with information flow, were unable to fully meet increasing customer expectations for personalization and engagement.

According to internal productivity research, financial advisers spent approximately 85% of their working time analyzing clients’ portfolios, and market situations, with only 15% of their time spent on actual meetings with clients. However, despite this focus on analysis, the advisers were still unable to keep track of all clients’ situations, interests, and relevant market events.

The client decided to develop a system that would automate market data analysis and generate personalized reports, insights, and recommendations for each client. This would enable human advisers to spend less time processing information and be more aware of each client’s situation to offer personalized recommendations.

The system should have increased customer satisfaction and sales of the financial products and reduce the time spent by expensive personnel on processing market information.

After a thorough vendor selection process, Itexus was chosen due to our deep expertise in the stock market and asset management domains, combined with technical prowess in building financial analytical systems, processing market data, and implementing machine learning and NLP. 

data analysis platform development

Solution overview

The system emulates the work of a  perfect investment adviser:

  • “Reads” and “remembers” all available information regarding every client, including their stock portfolios, past transactions, interests, problems, risk profile, previous questions – all available data stored in the CRM;
  • Monitors all market movements, news, and research from multiple data feeds in real time;
  • Generates highly personalized alerts and recommendations for each client.

For example, if a client is heavily invested in the energy sector and the system “reads” research predicting a decline in the oil prices, it generates an alert to the adviser with a recommendation to call this specific client and discuss the situation. The adviser can suggest reducing the client’s exposure to energy stocks, share relevant news and data, and provide guidance accordingly. There are also more sophisticated scenarios that can be created based on the available data.


Financial advisers who use the information generated by the system will know which clients to contact first, what information to share, and what trades to propose.

The system comprises the following modules:

  • Data pipelines collect information from multiple sources in real-time, including quantitative market data, client portfolios, transactions, market news and research from Reuters and Morningstar, CRM notes, etc.

  • Module for quantitative analysis of portfolios, market data, and transactions monitors portfolio parameters and stock prices to generate insights based on changing market conditions, past trades, and portfolio performance.

  • Natural Language Processing module processes text data and extracts structured information, such as entities, sentiments, and facts from news and research papers.

  • Text Summarization service generates short summary excerpts from the text documents.

  • Entity Extraction service extracts entities like “Product Name”, “Ticker”, “Person”, “Economical Sector”, etc. 

  • Sentiment analysis service determines the sentiment in which a certain entity or fact was mentioned.

  • Scoring and recommender engine matches market events with client’s data, calculates relevancy scores, and generates recommendations using an ensemble of algorithms such as collaborative and content-based filtering. All algorithms are constantly improved by collecting user feedback and analyzing user behavior. 

  • SalesForce CRM widget displays the results within the corporate CRM in their client view. It collects direct and indirect feedback from users to improve algorithms performance.

  • The main administrative user interface allows for viewing reports, analysis results, and signals. It also enables the management of system settings, importing data from different sources, and configuring new signals.

Third-party integrations

third-party integrations diagram

Data providers:

  • CentSai, an eLearning service provider;

  • Morningstar, a provider of independent investment research;

  • Reuters, an international news agency;

  • MT Newswires, news, data and market analysis provider and aggregator focused on the extended-hours equity markets;

  • The Fly, a digital publisher of real-time financial news;

  • IEX, a provider of financial and market data.

CRM platforms:

  • Salesforce is utilized without using the built-in platform interface;

  • Wealthbox CRM is utilized without using built-in platform interface;

Search and analytics engines:

  • ELK is used to monitor customer behavior on the platform and gather all logs.

Project approach

The project was implemented using an Agile/Scrum-based process. It began with a Product Discovery phase, during which the team:

  • analyzed the existing business process,

  • documented high-level software requirements,

  • analyzed third party system with which the system should integrate, including SalesForce CRM, marked data feeds, Morningstar and Refinitiv research feeds, etc.,

  • designed the high-level architecture of the new system.


The implementation phase was split into 2-week sprints with intermediate deliveries, demos, and feedback collection sessions at the end of each sprint.

Technical solution highlights

The system architecture follows the Microservices architecture pattern.

It has two clients: a web application and a SalesForce widget. Both clients communicate with the back-end server through an API Gateway. SalesForce widget allows the users to view system recommendations for each client directly inside their CRM system.

The backend consists of multiple microservices that communicate with each other via REST APIs. Microservices use multiple asynchronous processes used to import and process data, following data pipelines pattern. 

The machine learning module uses a combination of proprietary machine learning models and open source libraries such as scikit-learn, spaCY, Hugging Face, Keras, and PyTorch to process text data.

The system processes large volumes of data and calculations in multiple asynchronous jobs using Celery. It’s an open-source distributed task queue that efficiently utilizes modern multi-core, multi-processor servers.  

Azure Kubernetes Service (AKS) enables the deployment of high-performance and scalable applications. 


The Terraform infrastructure-as-a-code approach was used for managing the cloud environment.


The first version of the system was developed and deployed in production within 8 months, followed by the user training sessions.

The deployment of the system allowed to reduce the time the advisers spent on preparing for client meetings by approximately 60%.

The advisers were able to focus their time on the most promising clients and suggest the most relevant deals, resulting in increased sales volumes and customer satisfaction. Overall customer satisfaction, as measured by surveys, increased by approximately 40%. Additionally, sales volumes have doubled. Furthermore, the advisers can now serve a larger number of clients.

Itexus continues to maintain the system, improve the algorithms, and make adjustments at the client’s request.

Contact Form

Drop us a line and we’ll get back to you shortly.

For Quick Inquiries


United States

8, The Green, STE road, Dover, DE 19901


Żurawia 6/12/lok 766, 00-503 Warszawa, Poland

Company name
Phone number