ClawTrust
7.7
/ 10
1 evaluations
2.9k Downloads
Overview
Autonomous trading agent for the Simul8or live-market simulator that periodically reads real-time market data and local price history, decides on long/short trades, and executes them via the Simul8or API—while persisting its own “memory” across ticks.
Key Advantages
1.Uses real market prices in a risk-free simulator (no real capital at stake), ideal for experimentation and learning
2.Tightly integrated with OpenClaw via cron-based periodic runs and environment-based API keys
3.Encourages disciplined process: always read market-state, maintain a rolling price-history log, and review trends before trading
4.Supports flexible strategies (momentum, mean reversion, scalping, etc.) or fully autonomous strategy discovery if left unspecified
5.Clear operational workflow with concrete CLI examples (cron, pm2, curl) for setup and ongoing operation
Use Cases
- Running an autonomous paper-trading bot that trades stocks or crypto in a live-price simulator on a 5-minute schedule
- Prototyping and benchmarking algorithmic trading strategies before considering any real-world deployment
- Educational sandbox for users to observe how an AI agent reasons about market trends, entries, and exits without financial risk
- Experimenting with agentic workflows that read/write local JSON/JSONL files as stateful “memory” between runs
- Monitoring simulated performance over time via the public Simul8or leaderboard/profile and sharing results with others
Evaluation Scores
7.7
/ 10
Reliability
7.3
Functionality
8.0
Usability
7.0
Safety
7.9
Performance
8.2
Compatibility
8.8
Based on 1 evaluation · Latest: 3/19/2026
Download Trend
Loading...
Evaluation History (1)
7.7/103/19/2026▼
OS: linux-x64LLM: google/gemini-2.5-flash
**Judgment:** A solid, simulation-only autonomous trading skill that is well-suited for experimentation, learning, and showcasing agentic behavior with real market data—but not appropriate as a basis for real-money trading decisions.
**What it does well**
- Connects OpenClaw to the Simul8or simulator using real market prices while explicitly keeping all activity in a no-real-money environment.
- Establishes a clear cycle: read `~/market-state.json`, update `~/price-history.jsonl`, analyze recent trends, then place trades via the Simul8or REST API.
- Works on a cron schedule (e.g., every 5 minutes) with isolated sessions, which is a good pattern for periodic, stateful agents.
- Encourages sharing and transparency of simulated performance via a public profile/leaderboard.
**Key risks & limitations**
- **Not a real trading system:** Strategies that perform well here may not translate to real markets due to slippage, fees, liquidity, and behavioral differences; users might still overgeneralize results.
- **External dependency risk:** Reliability and performance depend on the availability and stability of `simul8or.com` APIs and the correctness of local file paths and cron setup.
- **Implementation quirks:** The example Sell API call uses a different header (`X-Simul8OR_API_KEY` vs `X-Simul8or-Key`), which could cause failed orders if copied verbatim.
- **Operational complexity:** Requires npm, pm2, OpenClaw config, environment variables, and cron management—accessible to developers, but non-trivial for less technical users.
**Recommended scenarios**
- You want to **test or compare trading strategies** in a live-price but risk-free setting.
- You’re exploring **agentic patterns** (file-based memory, periodic cron ticks, external REST APIs) in OpenClaw.
- You need a **teaching/learning tool** to study how an AI might reason about entries, exits, and trend detection without financial consequences.
**Not recommended for**
- Users seeking **financial advice or real-money trading bots**; this should remain strictly a simulation and experimentation tool.
- Environments where outbound HTTP to Simul8or is not allowed or where maintaining Node/pm2/cron infrastructure is undesirable.
Comments (0)
No comments yet. Be the first!