菜单
Creating PROGRESSION.md documents for research experiment series in Volatio. Captures the evolution of ideas, learnings, and breakthroughs across numbered experiments (001, 002, ...). Use when documenting experiment series or reviewing research evolution.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Market data integrity rules — gap filling, look-forward bias prevention, and stock split handling. Use when writing or reviewing code that loads prices, fills missing data, computes returns, backtests strategies, or makes trading decisions. Applies across Python, TypeScript, SQL, and TimescaleDB.
Analyze Volatio debug logs from /tmp/volatio-debug/. Use when the user asks to check debug logs, analyze problems, or diagnose simulation issues.
Apply bottom-up systematic testing to a multi-layer system. Use when asked to test, debug, or verify a pipeline that spans multiple layers (e.g., Python worker → bridge → processor → API → UI).
Drizzle Kit workflow for Volatio using idempotent migrations (generate + edit + migrate), never use push. Idempotent patterns prevent partial failure issues, enable safe re-runs, and work cleanly in CI. Use when schema changes, migration errors, or database sync issues occur.
Template for creating Volatio research experiments with Optuna optimization and portfolio backtesting. Includes proper train/test splits, lookahead bias prevention, results saving, and validated patterns from Exp 051-053.
Financial research look-ahead bias verification for Volatio ML pipelines, mandatory baseline comparisons, timezone handling, feature leakage detection. Use when model accuracy seems too good or validating prediction experiments.
| name | research-progression-docs |
| description | Creating PROGRESSION.md documents for research experiment series in Volatio. Captures the evolution of ideas, learnings, and breakthroughs across numbered experiments (001, 002, ...). Use when documenting experiment series or reviewing research evolution. |
Research experiments are numbered sequentially (001, 002, ...) where each builds on previous learnings. A PROGRESSION.md captures:
# [Series Name]: Experiment Progression
> **Series:** [Full description]
> **Goal:** [What we're trying to achieve]
> **Status:** ✅ ACTIVE | ⚠️ DEPRECATED | ✅ COMPLETED
---
## Progression Summary
| Phase | Experiments | Focus | Outcome |
|-------|-------------|-------|---------|
| 1 | 001-005 | Signal Discovery | [Key outcome] |
| 2 | 006-010 | Validation | [Key outcome] |
| 3 | 011-015 | Optimization | [Key outcome] |
---
## Phase 1: [Phase Name]
### 001 - [Experiment Name]
**Idea:** [What hypothesis are we testing?]
**Finding:** [What did we learn?]
**Progression:** [How does this inform the next experiment?]
---
### 002 - [Experiment Name] ✅ (BREAKTHROUGH)
**Idea:** [Hypothesis]
**Finding:** 🎯 **[Major discovery in bold]**
**Progression:** [Next steps]
---
## Key Learnings
1. **[Learning 1]** — [Explanation]
2. **[Learning 2]** — [Explanation]
---
## Evolution of Best Strategy
| Experiment | Strategy | Sharpe |
|------------|----------|--------|
| 001 | Baseline | 0.50 |
| 005 | +Optimization | 1.20 |
| 010 | +Filter | **2.50** |
---
## Production Configuration
```python
SIGNAL = "..."
ENTRY = "..."
EXIT = "..."
---
## Entry Format for Each Experiment
Each experiment entry follows this pattern:
```markdown
### NNN - Experiment Name [Status Emoji]
**Idea:** One sentence describing the hypothesis or question being tested.
**Finding:** What was discovered. Use 🎯 for breakthroughs, tables for data.
**Progression:** How this informs the next experiment or changes direction.
---
| Emoji | Meaning |
|---|---|
| ✅ | Successful, confirmed |
| ❌ | Failed, disproven |
| ⚠️ | Partial success, needs caveat |
| 🎯 | Major breakthrough |
| (none) | Standard experiment |
When an experiment changes the direction of research, mark it clearly:
### 010 - Validated SHORT Signal ✅ (BREAKTHROUGH)
**Idea:** Lock in the best SHORT signal with validation.
**Finding:** 🎯 **CORE SIGNAL DISCOVERED**
| Metric | Value |
|--------|-------|
| Sharpe | 12.67 |
| Win Rate | 83% |
**Progression:** Core signal validated. Now optimize.
Failed experiments are as important as successes — they prevent repeating mistakes:
### 003 - Earnings Reversal
**Idea:** Big moves revert the next day (mean reversion hypothesis).
**Finding:** Mean reversion appeared to work with high returns.
❌ *Later found to have look-ahead bias (see experiment 023)*
**Progression:** Invalidated — do not use.
When insights from one series apply to others, capture this:
## Key Learnings
| Insight | Discovered In | Applied To | Result |
|---------|---------------|------------|--------|
| Signal Exit | Laplace 016 | Extrinsic | ❌ Doesn't work |
| Signal Exit | Laplace 016 | Risk-Rotation | ✅ Works |
| Regime Filter | Extrinsic 036 | Risk-Rotation | ❌ Hurts |
Track how the best strategy improved across experiments:
## Evolution of Best Strategy
| Experiment | Change | Sharpe | Δ |
|------------|--------|--------|---|
| 008 | Baseline | 0.50 | — |
| 027 | +Production params | 0.70 | +0.20 |
| 031 | +Net Put-Call | 1.62 | +0.92 |
| 033 | +DTE 5-14 | 1.99 | +0.37 |
| 035 | +Trail50 | **4.72** | +2.73 |
End with actionable configuration for the final strategy:
## Production Configuration
```python
# Signal
SIGNAL = "Net Put-Call z-score > 1.0"
DTE_RANGE = (5, 14)
# Entry
ENTRY = "T+1 Open"
# Exit
EXIT = "trail50_7d"
MAX_HOLD = 7 # days
# Filters
REGIME_FILTER = "STABLE only"
---
## Examples in Codebase
| Series | Location | Experiments |
|--------|----------|-------------|
| SEC 8-K Original | `research/experiments/sec-8k-signals/PROGRESSION.md` | 001-023 |
| SEC 8-K V3 | `research/experiments/sec-8k-signals/v3/PROGRESSION.md` | 001-023 |
| Extrinsic Flow | `research/experiments/options/extrinsic/PROGRESSION.md` | 001-038 |
| Risk-Rotation | `research/experiments/options/extrinsic-risk-rotation/PROGRESSION.md` | 001-017 |
| Laplace | `research/experiments/options/laplace/PROGRESSION.md` | 001-021 |
---
## Checklist for New Progression Doc
- [ ] Header with series name, goal, and status
- [ ] Progression summary table with phases
- [ ] Individual entry for each experiment
- [ ] Breakthroughs clearly marked with 🎯
- [ ] Failed experiments marked with ❌ and explanation
- [ ] Key learnings section with cross-experiment insights
- [ ] Evolution table showing strategy improvement
- [ ] Production configuration if strategy is ready