Skip to content
Sableframe
Sableframe
LIVEFlagship

Prisma Trade-Ups

Market intelligence for the CS2 economy.

A platform for float-aware trade-up analysis, cross-market opportunities, price and scarcity signals, portfolio tracking, and accountable historical replay.

Status
LIVE
Source
Private
Capabilities
  • Market intelligence
  • Data pipelines
  • Decision tools
Visit live product
Prisma Trade-Ups home page, signed out: the headline “Trade smarter. See the edge.” with counts of items tracked, live contracts, and arbitrage opportunities beside game artwork, above cards for trade-up contracts, market signals, arbitrage, and portfolio, and a live feed.
HomeThe public entry point to Prisma’s market-intelligence views.

The problem

The CS2 item economy is spread across several marketplaces that price the same item differently. Trade-up contracts, cross-market spreads, and scarcity only become sound decisions when the prices underneath them can be trusted.

Prisma Trade-Ups brings those prices together and turns them into ranked, explainable opportunities, so traders state their constraints instead of searching each market and section by hand.

The system

  • Navigator

    Ranked trade-up and cross-market routes filtered by a trader’s constraints, each with a score and a separate confidence figure.

  • Forge

    Builds trade-up contracts from constraints, with alternatives and a breakdown of each outcome’s probability-weighted value. Float is currently estimated from wear bands.

  • Market signals

    Pulse, Atlas, and Depth cover price movement, volume, scarcity, how the economy connects, and liquidity and exit options by venue.

  • Arsenal

    Portfolio tracking with Steam sign-in. Anyone can preview an inventory, while stored holdings stay private to their owner.

  • Replay

    Scores past signals only after their window closes, using the prices available at the time, and reports insufficient data instead of an early number.

How it was built

A Next.js and TypeScript frontend reads from a FastAPI service. Scheduled Celery workers collect marketplace prices into PostgreSQL with TimescaleDB for price history, and separate engines compute trade-ups, cross-market spreads, and market signals into tables the API serves.

Redis backs the task queue, response caching, and sessions. Replay reads from append-only history tables that capture jobs extend over time.

  1. Collect

    • Scheduled Celery workers
    • Marketplace price collectors

  2. Store

    • PostgreSQL + TimescaleDB history
    • Latest-price tables
    • Replay history

  3. Analyze

    • Trade-up engine
    • Cross-market spreads
    • Market signals

  4. Serve

    • FastAPI and live feed
    • Redis cache and sessions

  5. Interface

    • Next.js App Router
    • React + TypeScript
Prices flow from scheduled collectors through analysis engines into stored results. The web app reads what the engines produced.

Technology

  • Next.js
  • React
  • TypeScript
  • FastAPI
  • PostgreSQL/TimescaleDB
  • Redis
  • Celery

Design decisions

  1. One source of truth for every figure

    Navigator and Forge filter and score what the scheduled engines already computed. They never recalculate value or spreads, so one opportunity cannot show different numbers on different pages.

  2. Verified or hidden

    A price is shown only when it is corroborated across sources or cross-checked. Otherwise the interface says the price is unavailable, because no number is better than a wrong one.

  3. No look-ahead in Replay

    Outcomes are judged only after their window closes, using prices up to that cutoff, and a result needs a minimum sample before any figure appears.

  4. Measured and modelled stay separate

    Without completed-sale data, time-to-sale is labelled as a modelled estimate, and low confidence hides the estimate instead of widening it.

Selected views

  • Prisma Navigator with route parameters for capital, strategy, risk, and markets; an optimal buy, transfer, and sell route with its Prisma score and expected return; the best route’s score, confidence, and value breakdown; and ranked alternate routes.
    NavigatorRanked routes filtered by constraints, each with a score and a confidence figure. Live market data as of September 14, 2026.
  • Prisma Forge with build parameters, a recommended jackpot trade-up contract showing total cost, expected value, profit and jackpot probabilities, its best outcome and outcome summary, the ten input skins, and alternative builds.
    ForgeBuilding a trade-up contract from constraints, with alternatives and an outcome breakdown. Live market data as of September 14, 2026.
  • Prisma Depth for an AWP skin: a liquidity score of 70 out of 100 with the factors behind it, and what it would sell for immediately, at the current ask, and within 24 hours or 7 days, with the timed ranges labelled as estimates.
    DepthWhat an item would realize if sold now or later, net of fees, with modelled ranges labelled as estimates. Live market data as of September 14, 2026.
  • Prisma Trade-Ups home page on a phone, signed out, with the headline, counts of items tracked, live contracts, and arbitrage opportunities, and the first feature card.
    On a phoneThe signed-out home page at a 390px viewport.

Current state

Live. Navigator, Forge, Atlas, Depth, and Replay are shipped. Replay’s history is still young, so its track record is intentionally sparse. The source code is private.