What's "under the hood" of the FNA Platform

FNA Platform Architecture

FNA Platform has been designed from the bottom up to replace myriad commercial and open source tools typically used by data scientists and software developers for modelling, network analytics & simulation. It is self contained and requires no additional software such as databases or visualisation tools.

  • Self contained Java application that runs on a variety of operating systems including Unix/Linux, Mac and Windows
  • Fully functional out of the box. No additional software licenses required
  • Native time series analysis

Network Algorithms

FNA has deep experience with transaction and risk reporting data standards within the financial services industry. Our understanding of the business context and message standards in specific domains such as payments, clearing and settlement, accelerates clients’ ability to deliver rapid and useful answers to specific business challenges.

  • Network centrality algorithms including: Eigenvector, PageRank
  • Network cluster algorithms including: spanning tree, PMFG, core periphery
  • Machine learning includes multiple training algorithms, cost and activation functions

Enterprise Integration

The FNA Platform is available as an on-premise enterprise deployment, on a private cloud, or as a shared Software-as-a-Service (SaaS) instance at Estimated time to deployment is measured in minutes, not days. The Platform is self-contained and is not dependent on additional software licences such as databases or middleware so can be implemented rapidly.

  • WAR or Docker container deployments on-premise, private cloud or shared cloud.
  • New FNA Platform users can be trained to be productive investigating their own data within hours.
  • Established FNA security credentials and integration with SAML means access controls and security requirements can be met rapidly.

Data Formats

Efficient and pre-configured data extraction, import and cleansing tools accelerate access to specific financial network standards and message formats including key financial services transaction formats ISO 20022, SWIFT and their variants.

  • Functionality to read and write file formats including: .csv, .xlsx, etc.
  • Database connections to read and write including: SQL, Teradata, Cloudera, and Neo4J.
  • Read / write REST and websocket APIs are included and pre configured API read access to popular online dataset such as Quandl, BIS, Companies House.
  • Our deep experience importing and cleansing data streams from within financial services networks means rapid delivery of results.

Finance specific algorithms & features

A key differentiator of FNA is that much of the software has been co-developed with our regulatory and FMI clients and this work has been put back into the platform. At the same time we contribute and follow academic research and implement domain specific algorithms when we see them

  • Systemic importance: SinkRank, DebtRank, SRISK, Syrin, PSLI
  • Ability to generate representative payments/trade data
  • FMI simulation & CCP stress testing


FNA’s clients include central banks, FMIs and government regulators with some of the world’s most stringent security requirements. We have passed all of their requested scans and testing. We work tirelessly to address such topics as data encryption at rest and in transit, integrity of all resources, and user authentication.

  • Trusted by Central Banks, CCPs and FMIs and tested for data security, integrity and confidentiality.
  • Role based access via secure access control through LDAP or SAML.
  • Security for data at rest with AES-128 encryption and SSL encryption for data is transit.


The FNA Platform includes a sophisticated set of data visualisation algorithms. Real-time mapping and filtering capabilities enable users to gain meaning from their data easily and to focus on the signal and remove the noise. Network displays are automatically optimised to reduce link overlap to eliminate “hairballs”.

  • Multiple visualisation layouts including: circle, force directed, ego distance, bipartite, bowtie, circle & circle tree, column, core periphery, ego distance, Harel Koren, hub and spokes, radial tree. Layouts can be calculated on server or adaptively on client side.
  • Nodes and link mappings can be updated in real time including node or link size/colour, link width, node/link label font size/colour, arrow style, etc.
  • Filters: Nodes & links can be filtered with multiple filters based on their property values


The FNA Platform can be configured to simulate any global FMI transaction settlement scheme including payments and trades. Additionally, Opening and closing times, number of participant members, daily volume and value settled can also be implemented

  • Settlement schemes include: Real-time gross settlement, netting, central counterparty clearing house (CCP)
  • Configurable liquidity savings mechanisms including multiple queues such as jumbo payments, bilateral and multilateral netting cycles,
  • Generation of representative payments including SWIFT MT103

Machine Learning and Artificial Intelligence

The FNA Platform includes configurable artificial neural network functionality to enable complex analytics enhancing the 300+ network algorithms

  • Machine learning includes multiple training algorithms, cost and activation functions


Contact our Experts

FNA is a UK-Registered company:
4-8 Crown Place, London EC2A 4BT
See on Map

By clicking Submit I agree to Privacy Policy

Subscribe to our Newsletter

Get the news of the next generation analytics, AI and machine learning in RegTech and SupTech.

By signing up you are agreeing to our Privacy Policy.