solsticestudioai's picture
Initial upload of ARC-T DeFi Execution sample
f502b3d verified
metadata
license: cc-by-4.0
task_categories:
  - reinforcement-learning
  - tabular-classification
language:
  - en
tags:
  - synthetic
  - finance
  - defi
  - crypto
  - mcts
  - trading-agents
  - execution-intelligence
pretty_name: Solstice ARC-T DeFi Execution Intelligence

Solstice ARC-T DeFi Execution Intelligence (Sample)

Synthetic risk-to-execution decision loops for autonomous DeFi agents. This dataset captures 3,000 multi-step decision traces from autonomous trading agents operating in a simulated Decentralized Finance (DeFi) ecosystem.

Built by Solstice AI Studio as a free sample of a larger commercial pack. 100% synthetic — no real wallet addresses or on-chain history used.

What's in the box

Each record in the JSONL stream represents a full execution loop, including:

  • Market State: Simulated oracle prices, liquidity depths, and volatility triggers.
  • Risk Assessment: Whale movement signals and oracle staleness metrics.
  • Agent Reasoning: MCTS-style strategy selection (Monte Carlo Tree Search) with success/failure probability weights.
  • Execution Outcome: Final trade result including gas costs, slippage, and PnL.

Use Cases

  • Autonomous Agent Training: Train models to select optimal trading strategies based on simulated market stress.
  • Risk Model Evaluation: Benchmark protocol safety against extreme market scenarios and oracle failures.
  • Anomaly Detection: Identify malicious or inefficient trading patterns in high-frequency DeFi telemetry.

Data Provenance

Generated using Solstice’s PhantasOS / SIMA simulation engine. The simulation uses MCTS (Monte Carlo Tree Search) to explore the decision space of a trading agent, recording both the chosen path and the explored alternatives.

Get the Full Pack

Scale this dataset to 1M+ decision loops, including cross-chain bridge logic and complex liquidity provider (LP) scenarios. www.solsticestudio.ai/datasets