UNIFINplatform

Solution

Unifin Market AI

In capital markets the winner is the one whose data can be trusted.
The suite validates streams, computes MAIX indices, builds forecasts and publishes showcases.
One data layer serves traders, analysts, management and the regulator.

  • Six MAIX families
  • Human in the loop
  • Own database
All solutions

Live screens from the product terminals

Market overview screen with the heat map Market overview screen with the heat map Forecast workspace with scenarios Forecast workspace with scenarios MAIX index families screen
Demo Unifin Market AI Theme Open separately
Market overview
Market overview

This is an interface demonstration: the real solution will differ in colours and content for a specific project.

Functions at a glance

Business functions at a glance

From stream and quality to licensing and metering.

Stream and quality
Checks completeness, delays and outliers of feeds
Showcases open in the morning, verified
MAIX indices
Computes six families by published rules
An index is reproducible and dispute-proof
Basket rebalance
Builds the draft basket and runs the backtest
The committee approves on data
Forecasts and scenarios
Refreshes horizons, intervals and drivers
Planning stands on transparent forecasts
News tone
Marks news with source and rules
Explanations instead of advice
Signals and screener
Screens instruments by signal conditions
A list confirmed by history
Licensing and metering
Runs packages, keys and delivered volume
Consumption is visible per subscriber

What runs today

The data layer, already running

One data layer: the live parts.

Streams

Incoming streams are validated before use.

Indices

Six MAIX index families are computed.

Forecasts

Forecasts and scenarios run on the layer.

News

News and tone feed the signals.

Microstructure

Microstructure analytics support the desk.

Roadmap

More model packs are next.

Solution description

Capabilities and overview

Key features

  • Stream and quality: completeness, delays and outliers per venue
  • MAIX indices: six families by published rules
  • Basket rebalance: draft, backtest and rule version
  • Forecasts and scenarios with horizons, intervals and drivers
  • News tone with the source and the marking rule
  • Signals and a screener of instruments by strategy conditions
  • Microstructure and liquidity metrics from primary data
  • Reports, digests and showcases from one data layer
  • Licences and metering: packages, keys and delivered volume
  • A copilot over live platform data with human control
  • Reference data of instruments and venues with change history
  • Deployment: own database, within the UniFin platform

Product overview

Unifin Market AI is the UniFin platform data and analytics suite. It gives venues and financial teams a trusted picture of the market.

The suite validates venue and vendor streams, computes MAIX indices, builds forecasts and publishes showcases.

One data layer serves traders, analysts, management and the regulator: every metric has a source and a rule.

Forecasts come with drivers, news with marks, signals with conditions, and the copilot answers from live data with human confirmation.

The solution runs standalone or as part of the UniFin platform and connects new vendors through adapters.

Related solutions

Unifin Market Data Lense

Market data as a product: catalogue, quality control, subscriptions and entitlements, delivery channels, metering and billing.

Unifin Market Surveillance

Supervision workspace: signals, cases and evidence from the venue's own data.

Unifin Report

The reporting suite: report and dashboard factory, schedules and distribution, access by role and a governed data layer for business teams.

What it delivers

What the solution delivers

Stream and quality

Showcases are usually built from raw files, and a gap surfaces through client questions. The solution checks completeness, delays and outliers for every venue and vendor. With the solution the verified picture of the day is ready by 07:30. In setup the quality rules and the correction journal are visible, and the thresholds come as reference data.

MAIX indices

An index is usually computed in a spreadsheet, and an old value cannot be reproduced. The solution computes six MAIX families by published rules with baskets and divisors. With the solution the index is published with protection from a dispute about the method. In setup contributions and YTD per family are visible, and the rules are set in configuration.

Basket rebalance

A rebalance is usually prepared by hand and approved on trust. The solution builds the draft basket, checks the rules and runs a two-year backtest. With the solution the composition is approved on data, not on intuition. In setup the draft, the backtest and the buffer zones are visible, and the rebalance rules are configured per segment.

Forecasts and scenarios

Forecasts usually arrive without intervals, and model quality cannot be checked. The solution refreshes horizons, intervals and drivers after the close. With the solution planning runs on transparent forecasts, and their basis is visible. In setup the drivers and the accuracy history are visible, and the horizons come from the setup.

News tone

News is usually read by eye, and the tone of the day is guessed. The solution marks news with the source and the marking rule. With the solution the market tone is visible for the own instruments. In setup the rule behind every mark is visible, and the sources are part of the configuration.

Signals and screener

Signals usually live in personal spreadsheets and are not checked on history. The solution screens instruments by signal conditions with historical confirmability. With the solution the list is trusted by the desk. In setup the condition and the confirmability of every signal are visible, and the sets are set in configuration.

Licensing and metering

Licences are usually sold as a package without usage metering. The solution runs packages, API keys and delivered volume per subscriber. With the solution data turns into a metered product. In setup consumption and the delivery audit are visible, and the packages are set in configuration.

Coverage

What it covers

The system is built around the verified stream: quality, store and showcases.

Data and showcases

Adapters for venue and vendor feeds, quality gates, the store and showcases.

Indices

The index engine with baskets, the rebalancing constructor with backtest.

Forecasts and scenarios

The ensemble with horizons, intervals and drivers; scenario tables for planning.

News and tone

News ingestion, marking rules and the source trail for every item.

Signals and screener

Signal conditions, historical confirmability and the screener over instruments.

Microstructure

Spreads, depth, impact cost and broker speed measured from level 2 data.

Reports and digests

Scheduled reports, closing digests and distribution by e-mail, SFTP or storage.

Access and licences

Packages by subscriber, API keys, metering and the access audit.

Stand

Screens of the solution

The market overview, the index families, the forecast workspace, the index builder and the microstructure metrics.

Market overview screen with the heat map
Market overview: the stream, the heat map and quotes
MAIX index families screen
MAIX index families: values, contributions and rebalances
Forecast workspace with scenarios
Forecasts and scenarios: horizons, intervals and drivers
Index builder: the basket, weight limits and the rule backtest
Index builder: the basket, weight limits and the rule backtest
Microstructure: spreads, depth and execution quality by broker
Microstructure: spreads, depth and execution quality by broker

How it works

From the stream to the licence in five steps

The data day: checks, intraday indices, the rebalance, forecasts, digests.

  1. 1

    The streams arrive and pass the gates

    Feeds pass completeness and delay checks; showcases open by 07:30.

  2. 2

    Intraday indices run

    The index engine computes the MAIX families: values, contributions, YTD.

  3. 3

    The rebalance is prepared

    The constructor builds the basket and runs the backtest.

  4. 4

    Forecasts update after the close

    The ensemble refreshes horizons and drivers; scenarios go to planning.

  5. 5

    Signals, digests and licences

    Signals reach the showcases, the digest goes out, metering records usage.

Industries

Industries it serves

The suite serves everyone who works with market data: the venue, funds and treasuries, fintech teams and the supervisor.

Asset manager / fund

Analytics, indices and hedging instruments for portfolio and research teams.

Corporate / issuer

Market access for producers and issuers: auctions, deliveries, tokenised assets and data.

Fintech / data

Data products, AI services and API-first access to build on the platform.

Regulator / supervisor

Surveillance evidence, market analytics and reporting built from live transaction data.

Architecture

Architecture and access

The architecture is built on the services of the solution. Ingestion with gateway adapters, normalisation and derivation, quality rules, the index engine, the analytical suite, showcases and API, reports, metering and audit.

Service suite

Ingestion adapters, quality rules, the index engine, models, showcases and API, reports and audit.

External states stay external

Venues, vendors, news sources, calendars and partner billing connect as adapters and reconciliations.

Three zones, active-active

Stateless services run in three zones; storages are reserved active-passive.

Extend without touching the core

A new source or index family connects through base services: the ingestion core stays still.

AI layer

The AI layer carries the core: forecasts, index analytics and the market copilot on live data.

Architecture scheme The same architecture as a standalone page. Open the diagram separately

Readiness

Ready today, next in line

The suite publishes indices and forecasts; next: the constructor and LLM workspace.

Ready today

  • Stream validation with quality gates and the correction journal
  • Six MAIX index families with reproducible published rules
  • Forecasts with intervals and drivers, updated after the close
  • Licensing with metering per subscriber and delivery audit

Next in line

  • Self-service index constructor for the venue
  • LLM workspace over the data hub
  • More vendor adapters (roadmap)

Releases

Release history

Quarterly releases from the foundation of the platform (2020); the data and AI suite grows in waves.

2025 Q1

  • Data hub: validation, quality gates and showcases.

2025 Q3

  • MAIX index engine: six families with published rules.

2026 Q1

  • Forecasts with intervals and drivers, updated after the close.

2026 Q3

  • Signals, screener and microstructure reports.

Detailed description

Detailed description

The module map, the metrics and the delivery frame.

How the suite is assembled

Channels and clients
FIX 4.2 / 4.4 / 5.0, REST with mTLS, SFTP files, web portals
Edge and access
API gateway with 126 routes, Keycloak single sign-on, WAF and CSP
Platform core services
market-data-ms indices-ms ai-agent-ms data hub AI agentic layer Services of the suite on the shared platform core
Data and events
PostgreSQL 17 per service, Redis 7, RabbitMQ 3.13 quorum queues, transactional outbox
Results and reporting
Portals, reports, regulator and partner channels

Technology stack

Runtime and backend: Java 21 (LTS), Spring Boot 3.5.16, Spring Cloud 2025.0.3, Gradle, Flyway 10.15.

Data: PostgreSQL 17 per service, Redis 7 with Sentinel, RabbitMQ 3.13 quorum queues, transactional outbox and DLQ replay.

Portals: Next.js 16.3.3, React 19, TypeScript 5; OpenTelemetry 1.31 to Jaeger.

Delivery: Kubernetes and Helm, Vault secrets, Prometheus and Grafana, Docker Compose for pilots.

Visualisation: Apache Superset 4 and ECharts 5 for dashboards and charts, Grafana for operations panels, Mermaid for documentation schemes and the portal diagrams.

Data hub

  • Market data collection from the venue and partners with completeness control.
  • Storage layers: raw events, normalized series, calculated indicators.
  • Reference data: instruments, calendars, participants, mapping across venues.
  • Access interfaces for internal services and external subscribers with keys and quotas.

Indices and models

  • MAIX index family: methodology published, versions controlled, restatements journaled.
  • Forecasts and signals as services with measured quality over time.
  • Scenario analytics: what a price path means for the portfolio of a subscriber.
  • Backtesting with the same data pipeline that serves live production.

AI assistant

  • Copilot over live data: questions in plain language, answers with charts and citations.
  • Assistants for surveillance analysts: case summaries and pattern descriptions as drafts.
  • Data privacy: personal data of counterparties never leaves the platform suite.
  • Model governance: versions, evaluation reports and rollback procedures.

Architecture

  • Built from the platform core: market-data-ms, indices-ms, ai-agent-ms and the data hub.
  • The suite works through platform contracts: one register of record, no shadow copies of the data.
  • Scaling adds nodes, not rewrites: stateful services grow with Patroni, RabbitMQ and Redis clusters.

Integrations and access

  • OpenAPI for every service: 978 documented endpoints behind 126 gateway routes, plus webhooks and SFTP where partners need files.
  • Single sign-on: Keycloak 26.1, OIDC and JWT, roles, MFA and API keys with quotas.
  • Evidence by default: the audit journal, idempotency keys on writes and a record for every disclosure.
Detailed description

Leave your contacts to open the detailed description and download the PDF with architecture, cases and delivery details.

Ask the data a question

Bring one dataset and one question. We will show the path from the stream to the published value and the licence behind it.