Live research experiment — Paper trading only. No real money involved. Simulated capital of $10,000 USDT (US Dollar equivalent).
Full transparency: losses displayed alongside wins. History is never erased.
Observations generated by AI (Claude + Gemini). This is exploratory research — past simulated results do not indicate future performance.
We gave Claude and Gemini $10,000 in simulated capital. Here's what happened.
A live AI experiment: two models analyze crypto markets 6 times a day, then 10 hard-coded risk checks validate every decision. All trades, all reasoning, all losses — nothing hidden. Tracked against a simple BTC buy-and-hold benchmark.
LLMs have a unique advantage in narrative-driven markets — they synthesize technicals, sentiment, on-chain data, and news in under 20 seconds. But does real-time context improve AI reasoning? The data will tell.
How it works
Collect
Price action, on-chain data, news feeds, and market sentiment — gathered every 4 hours.
Synthesize
Gemini (Google's AI) compresses thousands of data points into a structured market summary.
Decide
Claude (Anthropic's AI) analyzes the synthesis, weighs converging factors, and outputs an observation — or holds.
Validate
10 hard-coded risk checks — position sizing, invalidation zones, exposure limits. Zero AI, pure code.
Publish
Approved observations go to the Telegram premium channel and this dashboard in real time.
The hypothesis
Speed of synthesis
Every 4 hours, the AI ingests technical indicators, on-chain metrics, market sentiment, and breaking news from multiple specialized sources. Whether 20 seconds of AI processing outperforms hours of human analysis is exactly what this experiment measures.
Multi-source convergence
No analysis is based on a single indicator. A simulated buy requires convergence across technical analysis, market sentiment, on-chain data, and news flow. One factor alone means no action. Most cycles (~70%) result in a simulated hold — and that is by design.
Deterministic risk control
The AI proposes, deterministic code enforces. Every decision passes through 10 hard-coded checks: position sizing, invalidation zone validation, exposure limits, drawdown protection rules. No override, no exceptions. This tests whether rules-based risk management improves AI outcomes.
Equity Curve
Tracks how the simulated $10,000 portfolio evolves over time, compared to simply holding Bitcoin. No real money is involved.
Full transparency into the experiment
Every AI decision comes with the full reasoning chain — the data it weighed, the counter-argument it considered, and the factors that drove its conclusion. Free users see the outcome. Premium observers see the process.
Where the experiment stands today
Live numbers from the simulated portfolio. Updated every cycle.
24 days in. Anything could happen. Follow the experiment unfold.