Personal project · markets · 2026

TRADEVALUE

A market research terminal: specialised agents read prices and macro data, and a safety system is always on.

Role
Concept and development
Period
2026
Stack
PythonFastAPIReactTypeScriptSSESQLite

live demo Terminal with synthetic data

Strategies, levels, prompts and results are private. The demo uses fictional assets and synthetic data: no investment advice.

TRADEVALUE · research workspaceSynthetic data · not investment advice
TRADEVALUEresearch terminal
UTC --:--:--

tradevalue terminal · live

Buongiorno.

Three agents watch two synthetic assets and only flag what changes the picture.

ASSET-A ▲ bullishASSET-B ■ flat
Simulated positions
2
Levels watched
12
Last event
0s
Events buffered
124
Stream
SSE · live

Agent constellation

  • Orchestratorlistening

    Coordinates and decides what deserves attention

    Active thesisTwo signals waiting for macro confirmation before becoming an alert.

  • Levels analysisanalysing

    Reads price, volume and key zones

    Active thesisASSET-A above pivot: constructive structure while the zone holds.

  • Macroidle

    Frames the economic context

    Active thesisCooling inflation, contained volatility: neutral regime.

Live event stream

    Tap an event to inspect the raw payload.

    A rebuild of the interface with data generated in your browser. Strategies, real levels and agent prompts stay private.

    A — The problem

    Following markets means combining charts, price levels, macro data and news — and remembering why you made a decision. Platforms give you the data, not the reasoning, and rarely help you understand afterwards what worked.

    B — What I built

    TRADEVALUE is my analysis workspace. An orchestrator coordinates agents that study price levels and the macro picture, write reasoned theses and send alerts on Telegram. A terminal-style dashboard shows charts, live events, macro indicators and a log of simulated trades. A sentinel module can halt everything with a switch, and a calibration view compares stated confidence with outcomes.

    3coordinated agents
    4chart timeframes
    10+macro indicators
    1kill switch per asset

    C — How it works

    1. 01

      Streaming data

      Prices, levels and macro indicators arrive in real time.

    2. 02

      Agents with one job

      Each produces a thesis with its reasons, not a bare signal.

    3. 03

      Safety first

      Sentinel with a per-asset kill switch and an audit log.

    4. 04

      Memory and review

      Post-mortems and calibration measure how reliable the forecasts were.

    D — The system

    A simplified map of the components: not the technical documentation, but it shows how the parts talk to each other.

    E — Why it works

    Explicit reasoning

    Every thesis is readable and comparable over time.

    A switch always within reach

    The halt is deterministic and does not depend on the model.

    Learning from mistakes

    Post-mortems compare what happened with the alternatives.

    Terminal density

    Lots of information, readable, with tabular numbers and shortcuts.