CS2 GRP Case Study

A case study ranking cross-venue buy→sell routes in CS2 skins by cash netback after FX basis, fees, lockups, and liquidity haircuts, with optional pricers for floats/patterns and trade-up manufacturing anchors.

Why it’s relevant to commodities / physical trade

If you strip the paint off the skins:

• marketplaces = exchanges / brokers
• Steam Wallet = restricted settlement currency (Steam USD $\neq$ Cash USD)
• fees = transaction costs
• trade holds = inventory carry / settlement delay
• Covert $\to$ Gold = input $\to$ output manufacturing spread
• regional basis (West vs China) = geographic arbitrage / basis trading

This repo demonstrates the same system design decisions: normalize, de-bias, score, and rank.

What it does

1. Ingest price prints and listings from multiple venues.
2. Normalize:
   • net-of-fees
   • currency to USD (ECB FX by default)
   • optional implied FX basis per venue (treating restricted usage currencies like Steam Wallet separately)
3. Compute a Global Reference Price (GRP) via weighted median: $P^{GRP} = \text{wMedian}({P_{net}}, {Weights})$
4. Apply Feature Multipliers to compute a Model Price: $P^{model} = P^{GRP} \cdot M_{float} \cdot M_{pattern} \cdot M_{stickers}$
   • Float: Tri-phase curve (Perfection, Wear Buckets, High-Float).
   • Pattern: Tier-based multipliers (e.g., Blue Gem).
   • Stickers: Scrap value + Synergy premium.
5. Score Routes (Netback Edge): $Edge = \frac{SellProceeds_B - BuyCost_A}{BuyCost_A}$
   • Haircuts: Time-to-sell, inventory lockup, and venue risk.

Full mathematical details in docs/quant-model.md.

Quickstart

1) Install

This project uses uv for dependency management.

uv sync

2) Run the demo

uv run cs2arb demo run --prices data/demo_prices.json --listings data/demo_listings.json

3) Feature Scoring

Compute the precise model price for an item with specific attributes (e.g., low float):

uv run cs2arb features score-listing --base-price 100 --float-val 0.001

4) Manufacturing Analysis

Analyze input/output spreads (e.g., Trade-Up profitability):

uv run cs2arb manufacturing analyze --input-prices "10,10,10" --output-prices "50,12"

Documentation

• docs/case-study.md – context + framing
• docs/architecture.md – data-flow diagrams
• docs/data-model.md – listings, prints, and feature schemas
• docs/usage.md – detailed CLI usage guide

Repo map

• src/cs2_arb/ – package
  • fx/ – ECB FX + implied FX basis estimator
  • markets/ – adapters (demo, steam, buff163)
  • pricing/ – GRP, fee normalization, float curves, manufacturing
  • scoring/ – route scoring (netback & basis)
  • storage/ – sqlite cache
• data/ – fixtures (safe, fake-but-plausible numbers)

Disclaimer

This project is for engineering demonstration and research. It does not automate trades, does not bypass platform protections, and is not financial advice.