Crypto Self-Learning

**Judgement:** A well-scoped, practical skill for disciplined crypto traders who want a structured, data-driven journaling and review loop. It is strong as an *intelligence/insight* layer (logging, analysis, rule extraction) but is **not** a full quantitative research or execution system. **What it does well:** - Enforces consistent trade logging with detailed context (indicators, macro, notes). - Provides straightforward analyses (win rate by direction, day, RSI ranges, leverage, etc.). - Automatically converts trade history into heuristic rules and appends them to an agent-readable MEMORY.md. - Encourages regular, systematic reviews via a weekly report script. **Key risks & limitations:** - **Statistical robustness:** Rules are generated from historical trades but may be based on small sample sizes or noisy correlations (e.g., “SHORT Mondays” with n=9). Without significance testing, this can lead to overfitting and misleading confidence. - **No built-in risk management or portfolio analytics:** It analyzes trade-level patterns but doesn’t model drawdowns, position sizing, or portfolio volatility. - **User-dependent data quality:** If trades are logged inconsistently or selectively (e.g., skipping bad trades), insights and rules will be biased and potentially harmful. - **Behavioral risk:** Users may over-trust generated rules as “alpha,” instead of treating them as hypotheses to be validated with further testing. **Recommended scenarios:** - Intermediate/advanced crypto traders comfortable with Python and CLI who want a rigorous trade journaling and review framework. - Users running agents like tess-cripto who want their agents to “learn” from past trades via a shared MEMORY.md rules section. - Traders focusing on process improvement and self-diagnosis (identifying which setups, times, or leverage levels are hurting/helping performance). **Use with caution when:** - You are a beginner trader looking for an automatic profit-generating strategy—this skill helps analyze your behavior; it does not guarantee profitable signals. - You cannot maintain strict discipline in logging *every* trade; partial logging will produce biased analytics and unreliable rules. - You need statistically rigorous backtesting, optimization, or risk modeling; this skill should then be paired with dedicated quant/backtesting tools.