← Back to projects

Tradiumz / Fullstack

Tradiumz

A crypto market intelligence platform connecting live market data, explainable analysis, risk planning, and a trading journal.

Backend-connected product · validation and hardening in progress

Scope & current limitations

Tradiumz supports analysis, planning, and journaling. Exchange execution remains manual. Deployment tooling is implemented; release readiness and trading-model performance depend on validation evidence, rather than an assumed win rate.

The challenge

Market monitoring, trade analysis, and performance review often live in separate tools. Tradiumz brings these steps into one traceable workflow so a user can inspect the evidence behind a setup, prepare a plan, and review the outcome against the original decision.

Outcome

An integrated decision-support workflow from market ingestion to post-trade review, with backend-connected screens and an auditable record of the data and decisions behind each plan.

My contributions

  • Built a Next.js interface for market discovery, scanner results, analysis, trade plans, journal entries, and validation.
  • Developed the Go API and background ingestion workers, with durable market history in PostgreSQL and live state in Redis.
  • Connected analysis, exact-decimal risk calculations, versioned plans, and manual trade records through typed API contracts.
  • Added advisory AI integration, research validation, observability, and container deployment workflows.
Explore the engineering & delivery

Product surfaces, workflows, architecture decisions, and verification details.

The product, end to end

Backend-connected web app

Market & decision workspace

Responsive dashboards, scanner filters, score breakdowns, symbol analysis, and trade-plan detail pages make market context and decision evidence accessible in one interface.

Implemented workflow

Journal & review

Versioned plans, manual entries and exits, fee-adjusted outcomes, and daily reviews connect the intended trade with what actually happened.

Implemented services

API & market workers

A Go API and separate workers handle symbol discovery, candle backfill, WebSocket ingestion, gap repair, feature snapshots, and scheduled scanning.

Evidence-gated capabilities

Research & advisory layer

Backtests, paper signals, calibration records, and optional AI interpretation support investigation while preserving the underlying quantitative decision fields.

How the workflow fits together

  1. Ingest and reconcile market data

    Discover eligible Binance symbols, backfill closed candles, and receive live updates. Missing intervals trigger repair; data freshness is explicit.

  2. Scan and inspect a setup

    Create multi-timeframe feature snapshots, apply scanner eligibility rules, and expose component scores for inspection in the web interface.

  3. Prepare and record a trade plan

    Resolve an analysis into a plan, calculate price and position constraints, and track its lifecycle. The user executes externally and records the entry manually.

  4. Close the loop with review

    Record the exit, calculate the backend outcome, compare plan versus execution, and review the evidence in the journal and validation workspace.

From product design to delivery

01 / Product & interaction design

Organized the product around a repeatable decision loop: market overview, scanner, analysis, plan, execution record, and review. Explicit offline and empty states keep missing market information visible.

02 / Frontend & API integration

Implemented Next.js and React screens with TypeScript and Tailwind CSS. Server-side reads and an allow-listed backend proxy centralize API access, preserve idempotency keys, and keep backend configuration out of browser code.

03 / Backend & processing

Separated HTTP request handling from ingestion and scheduled work in Go. Domain modules cover indicators, scanning, analysis, risk, journaling, notifications, and research validation; SSE delivers update events to the browser.

04 / Data & consistency

Used PostgreSQL for closed candles, immutable snapshots, analysis records, and plan history, with Redis for live state. Decimal-string API values and precise database numerics avoid floating-point drift in price and position calculations.

05 / Reliability & verification

Implemented reconnecting market feeds, gap recovery, stale-data indicators, durable webhook handling, and health/readiness checks. Tests cover domain calculations, scanner behavior, integration boundaries, and validation gates.

06 / Delivery & operations

Defined Docker Compose services and GitHub Actions workflows for building images, publishing to GHCR, and deploying to a VPS. Operational documentation covers metrics, recovery, backups, and rollback; release gates require recorded validation evidence.

Key engineering decisions

Keep market history and live state separate

Persist closed candles as durable records while using Redis for rapidly changing state. This supports repeatable analysis without treating partial market updates as final history.

Preserve the evidence behind a decision

Immutable snapshots and versioned plans make later review traceable to the information available when an analysis was created.

Make AI advisory and uncertainty visible

Optional AI interpretation cannot rewrite quantitative plan fields. Scanner scores are not presented as win probabilities, and probability serving remains subject to evidence gates.

Testing & operational readiness

  • The repository includes race-enabled Go checks, domain tests, PostgreSQL integration tests, and frontend chart checks.
  • Health endpoints distinguish a live process from ready dependencies, while worker freshness is reported separately.
  • Load, recovery, paper-validation, and release gates are documented; this case study does not claim those operational gates have all passed.

Building something similar?

Let’s discuss the users, workflows, and priorities for your product.

Discuss Your Project