Investment thesis
AI / LLM-driven — economic intuition. Large language models act as a flexible signal-extraction layer over unstructured inputs (filings, earnings calls, macro commentary). The edge is not price prediction directly but higher-fidelity feature engineering on text that would otherwise require a team of analysts — a cost-arbitrage against traditional discretionary funds.
Risk-adjusted performance — live track record
Forward-tested daily against live market data. Metrics derived from end-of-day portfolio marks; methodology documented on the Due Diligence and About pages.
| Return | Value | Risk-adjusted | Value | |
|---|---|---|---|---|
| Current portfolio worth | $10543.95 | Sharpe ratio | 0.82 | |
| Total return | 4.95% | Sortino ratio | 1.27 | |
| CAGR | 16.29% | Calmar ratio | 2.50 | |
| Volatility (annualised) | 17.38% | Profit factor | 1.18 | |
| Days live | 96 | Maximum drawdown | -6.51% |
Process consistency
| Positive months | 50.0% |
| Best month | 10.74% |
| Worst month | -4.47% |
| Recovery from max drawdown | still underwater |
Market independence
Correlation and beta versus passive benchmarks, computed over the full live series.
| Benchmark | Correlation | 90-day rolling correlation | Beta |
|---|---|---|---|
| S&P 500 (SPY) | 0.66 | 0.00 | 1.01 |
| Bitcoin (BTC-USD) | 0.37 | 0.37 | 0.18 |
A correlation materially below 1.0 to both benchmarks indicates the strategy’s returns are not a simple re-expression of long equity or long crypto beta.
Equity curve
Live track record — forward-tested performance from the strategy's production start date.
Drawdown profile
Underwater curve — percentage below the running high-water mark. Institutional allocators read this before the equity curve.
Current holdings
| Symbol | Quantity |
|---|---|
| AAPL | 2.8611 |
| AMZN | 2.6264 |
| BTC-USD | 0.0051 |
| ETH-USD | 0.1268 |
| GLD | 1.0966 |
| GOOG | 1.7430 |
| LLY | 0.2896 |
| META | 0.9270 |
| MSFT | 2.2485 |
| NVDA | 5.0714 |
| QQQ | 2.1582 |
| TMF | 14.6050 |
| USD | 0.0000 |
| UUP | 37.6162 |
| VGT | 13.2997 |
Research & documentation
- Strategy deep-dive: DeepSeekToolBot: strategy deep-dive & live performance
- Reference implementation:
tradingbot/aideepseektoolbot.py - Framework: python_tradingbot_framework (open source, fully inspectable)
Related strategies
Other strategies in the AI / LLM-driven family:
- GptBasedStrategyBTCTabased · research note- AIHedgeFundBot · research note Or view the full strategy roster.
For professional investors
Request the investor deck, DDQ, and extended analytics. Firm-gated and reviewed manually.
Request access