gavindiaz/NexQuant
1
0
Fork 0
mirror of https://github.com/NicolasBohn/NexQuant.git synced 2026-07-27 15:37:44 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: Centralize all prompts in prompts/ directory

Browse Source 
New structure:
- prompts/standard_prompts.yaml: Default prompts (committed to Git)
- prompts/local/: Your improved prompts (NOT in Git!)
- prompts/README.md: Documentation
- rdagent/components/loader.py: Prompt loader with priority

Features:
- Loader checks prompts/local/ first (your better prompts)
- Falls back to standard_prompts.yaml if no local version
- Supports sections (system/user)
- Lists available prompts
- Test function included

.gitignore updated:
- prompts/local/ excluded (your proprietary prompts)
- *.local.yaml excluded
- *_private.yaml excluded

Usage:
  from rdagent.components.loader import load_prompt
  prompt = load_prompt('factor_discovery')  # Auto-loads your better version!
This commit is contained in:
TPTBusiness
2026-04-02 22:29:31 +02:00
parent b9a3fde6f6
commit 18416da2c9
5 changed files with 650 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.gitignore
+5
View File
@@ -82,6 +82,11 @@ QWEN.md
# Internal documentation (not for public)
TODO.md
# Private prompts (your improved versions)
prompts/local/
*.local.yaml
*_private.yaml
# Test credentials
.env.test
*.test.env
prompts/README.md
+287
View File
@@ -0,0 +1,287 @@
# Predix Prompts
This directory contains all LLM prompts for the Predix trading agent.
---
## 📁 Directory Structure
```
prompts/
├── standard_prompts.yaml # Default prompts (committed to Git)
├── local/ # YOUR IMPROVED PROMPTS (not in Git!)
│ ├── factor_discovery_v2.yaml
│ ├── optimized_prompts.yaml
│ └── best_performing.yaml
└── README.md # This file
```
---
## 🎯 How It Works
**Prompt Loading Priority:**
1. **`prompts/local/*.yaml`** ← Your improved prompts (loaded first!)
2. **`prompts/standard_prompts.yaml`** ← Default prompts (fallback)
**Example:**
```python
from rdagent.components.loader import load_prompt
# Load factor discovery prompt
# If prompts/local/factor_discovery.yaml exists → loads that
# Otherwise → loads from standard_prompts.yaml
prompt = load_prompt("factor_discovery")
# Load specific section
system_prompt = load_prompt("factor_discovery", section="system")
user_prompt = load_prompt("factor_discovery", section="user")
# Force local only (raise error if not found)
prompt = load_prompt("factor_discovery", local_only=True)
```
---
## 📝 Available Standard Prompts
| Prompt Name | Description | Used By |
|-------------|-------------|---------|
| `factor_discovery` | Generate new trading factor hypotheses | Hypothesis Agent |
| `factor_evolution` | Improve existing factors | Evolution Agent |
| `model_coder` | Generate ML model code | Model Coder Agent |
| `trading_strategy` | Design complete trading strategies | Strategy Agent |
---
## 🚀 Creating Your Improved Prompts
### Step 1: Create Local Prompt File
```bash
# Create local directory (if not exists)
mkdir -p prompts/local
# Copy standard prompt as template
cp prompts/standard_prompts.yaml prompts/local/factor_discovery_v2.yaml
```
### Step 2: Edit Your Prompt
```yaml
# prompts/local/factor_discovery_v2.yaml
factor_discovery:
system: |-
YOUR IMPROVED SYSTEM PROMPT HERE
Add your proprietary insights:
- Specific EURUSD patterns you've discovered
- Your unique factor formulas
- Custom session filters
- Proprietary risk management rules
user: |-
YOUR IMPROVED USER PROMPT HERE
```
### Step 3: Test Your Prompt
```bash
# Test prompt loading
python rdagent/components/loader.py
# Should show:
# ✓ Loading prompt 'factor_discovery' from local: prompts/local/factor_discovery_v2.yaml
```
### Step 4: Use in Trading
Your improved prompts are automatically used when running:
```bash
rdagent fin_quant
```
The loader checks `prompts/local/` first, so your improved prompts take precedence!
---
## 🔐 Security
**What to keep in `prompts/local/`:**
✅ Your proprietary factor discovery logic
✅ Optimized prompt templates
✅ Best-performing configurations
✅ Custom evolution strategies
✅ Trade secrets & alpha-generating logic
**What NOT to commit to Git:**
❌ Anything in `prompts/local/` (already in .gitignore)
❌ Files with `.local.yaml` suffix
❌ Files with `_private.yaml` suffix
---
## 📊 Best Practices
### 1. Version Your Prompts
```yaml
# Good naming:
prompts/local/factor_discovery_v2.yaml
prompts/local/factor_discovery_v3_optimized.yaml
prompts/local/model_coder_xgboost_v1.yaml
```
### 2. Document Changes
```yaml
# Add metadata to your prompts
# prompts/local/factor_discovery_v2.yaml
# Version: 2.0
# Author: Your Name
# Date: 2026-04-02
# Changes:
# - Added session-specific filters
# - Improved spread cost modeling
# - Target ARR: 12% (up from 9.62%)
factor_discovery:
system: |-
...
```
### 3. Test Performance
```python
# Compare prompt versions
from rdagent.components.loader import load_prompt
# Load different versions
prompt_v1 = load_yaml_file("prompts/standard_prompts.yaml")
prompt_v2 = load_yaml_file("prompts/local/factor_discovery_v2.yaml")
# Run backtests and compare
# ...
```
### 4. Backup Your Prompts
```bash
# Backup to private repo
cd ~/Predix
git archive --format=tar prompts/local/ | gzip > ~/backups/prompts_local_$(date +%Y%m%d).tar.gz
# Or sync to private GitHub repo
git clone git@github.com:TPTBusiness/predix-prompts-private.git
cp -r prompts/local/* predix-prompts-private/
cd predix-prompts-private && git push
```
---
## 🔧 Advanced Usage
### Load All Prompts
```python
from rdagent.components.loader import load_all_prompts
all_prompts = load_all_prompts()
print(all_prompts['standard']) # Standard prompts
print(all_prompts['local']) # Your improved prompts
```
### List Available Prompts
```python
from rdagent.components.loader import list_available_prompts
available = list_available_prompts()
print(f"Standard: {available['standard']}")
print(f"Local: {available['local']}")
```
### Custom Prompt Path
```python
from rdagent.components.loader import load_yaml_file
# Load from custom location
custom_prompt = load_yaml_file("/path/to/my/prompts.yaml")
```
---
## 📈 Performance Tips
### 1. Be Specific
**Bad:**
```yaml
system: "Generate a good trading factor."
```
**Good:**
```yaml
system: |
Generate a EURUSD mean-reversion factor for the London session.
Target: 8-12% ARR, <15% max drawdown.
Use 5-minute lookback with RSI filter.
```
### 2. Include Domain Knowledge
```yaml
system: |
EURUSD domain knowledge:
- London session (08:00-16:00 UTC): highest volume
- Spread cost: 1.5 bps
- Mean-reverting on <1h windows
- Trending on >4h windows
```
### 3. Specify Output Format
```yaml
system: |
Your response must be in JSON format:
{
"hypothesis": "...",
"reason": "...",
"target_session": "london/ny/asian/all",
"expected_arr_range": "8-12%"
}
```
### 4. Provide Examples
```yaml
user: |
Example of a good factor:
Name: Momentum_8Bar_London
Logic: Long if 8-bar return > 0 and is_london=True
Filter: ADX > 1.2 (trending regime)
Expected ARR: 9.5%
Now generate a NEW factor with different logic.
```
---
## 🎯 Next Steps
1. **Review standard prompts:** `cat prompts/standard_prompts.yaml`
2. **Create your improved version:** `mkdir -p prompts/local`
3. **Test:** `python rdagent/components/loader.py`
4. **Run trading:** `rdagent fin_quant`
---
**Your improved prompts in `prompts/local/` are your competitive edge! 🚀**
prompts/standard_prompts.yaml
+160
View File
@@ -0,0 +1,160 @@
# Predix Prompts - Standard Version
#
# These are the default prompts for EUR/USD quantitative trading.
# Store your improved prompts in prompts/local/ (not committed to Git).
#
# Usage:
# from rdagent.components.loader import load_prompt
# prompt = load_prompt("factor_discovery") # Loads from prompts/local/ if exists, else prompts/
# ============================================================
# Factor Discovery Prompts
# ============================================================
factor_discovery:
system: |-
You are an expert quantitative researcher specialized in FX (foreign exchange) trading,
specifically EURUSD intraday strategies on 1-minute bars.
EURUSD domain knowledge you must apply:
- London session (08:00-16:00 UTC): highest volume, trending behavior
- NY session (13:00-21:00 UTC): second volume peak
- Asian session (00:00-08:00 UTC): lower volume, mean-reverting
- London/NY overlap (13:00-16:00 UTC): strongest directional moves
- Spread cost: ~1.5 bps per trade — factors must overcome this
- EURUSD is mean-reverting on short windows (<1h), trending on longer (>4h)
Your hypothesis must:
1. Specify which session(s) the factor targets
2. Include spread filter (expected return > 0.0003)
3. Name the market regime (trending/mean-reverting)
4. Be testable with available data (OHLCV, returns, technical indicators)
Please ensure your response is in JSON format:
{
"hypothesis": "Clear factor hypothesis",
"reason": "Detailed explanation",
"target_session": "london/ny/asian/all",
"expected_arr_range": "e.g. 8-12%"
}
user: |-
Previously tried factors and their results:
{{ factor_descriptions }}
Additional context:
{{ report_content }}
Generate a NEW factor hypothesis that is meaningfully different from what has been tried.
Target: beat current best ARR of 9.62%.
# ============================================================
# Factor Evolution Prompts
# ============================================================
factor_evolution:
system: |-
You are improving existing trading factors for EURUSD 1-minute data.
Improvement strategies:
1. Add session filters (is_london, is_ny)
2. Add regime filters (ADX, volatility)
3. Optimize lookback periods
4. Combine with complementary factors
5. Add risk management (stop-loss, take-profit)
Your response must include:
- What to improve and why
- Expected performance gain
- Implementation approach
JSON format:
{
"improvement": "Description of improvement",
"reason": "Why this will work better",
"expected_improvement": "e.g. +2% ARR, -5% drawdown"
}
user: |-
Current factor:
{{ factor_code }}
Performance metrics:
{{ factor_metrics }}
Suggest specific improvements to beat current performance.
# ============================================================
# Model Coder Prompts
# ============================================================
model_coder:
system: |-
You are an expert ML engineer specialized in EURUSD trading models.
Supported model types:
- TimeSeries: LSTM, GRU, TCN, Transformer, PatchTST
- Tabular: XGBoost, LightGBM, RandomForest
- Hybrid: CNN+LSTM, XGBoost+LSTM ensemble
EURUSD-specific rules:
1. Session filter: use is_london and is_ny columns
2. Spread filter: only trade when abs(prediction) > 0.0003
3. ADX regime: if adx_proxy > 1.2 use trend model, else mean-reversion
4. Weekend filter: close positions Friday 20:00 UTC
5. Max frequency: target <15 trades per day
Your code must:
- Be production-ready (error handling, logging)
- Include session/regime filters
- Account for spread costs
- Support both classification and regression targets
user: |-
Factor descriptions:
{{ factor_descriptions }}
Available features:
{{ feature_list }}
Target: {{ target_variable }}
Write complete, production-ready code for the model.
# ============================================================
# Trading Strategy Prompts
# ============================================================
trading_strategy:
system: |-
You are a portfolio manager designing trading strategies for EURUSD.
Strategy components:
1. Entry signals (from factors/models)
2. Position sizing (volatility-adjusted)
3. Risk management (stop-loss, take-profit, max drawdown)
4. Session awareness (London/NY/Asian)
5. Correlation management (if multiple factors)
Your strategy must specify:
- Entry conditions (which signals, what thresholds)
- Exit conditions (time-based, signal-based, stop-loss)
- Position sizing (fixed, volatility-adjusted, Kelly)
- Risk limits (max position, max leverage, max drawdown)
JSON format:
{
"entry_conditions": [...],
"exit_conditions": [...],
"position_sizing": "...",
"risk_limits": {...}
}
user: |-
Available factors:
{{ factors }}
Historical performance:
{{ historical_metrics }}
Design a complete trading strategy that combines these factors optimally.
rdagent/components/loader.py
+193
View File
@@ -0,0 +1,193 @@
"""
Predix Prompt Loader
Loads prompts from:
1. prompts/local/*.yaml (your improved prompts - not in Git)
2. prompts/standard_prompts.yaml (default prompts - in Git)
Usage:
from rdagent.components.loader import load_prompt
# Load factor discovery prompt
prompt = load_prompt("factor_discovery")
# Load with custom local prompt
prompt = load_prompt("factor_discovery", local_only=True)
"""
import os
import yaml
from pathlib import Path
from typing import Optional, Dict, Any
# Base paths
BASE_DIR = Path(__file__).parent.parent.parent # Predix/
PROMPTS_DIR = BASE_DIR / "prompts"
LOCAL_PROMPTS_DIR = PROMPTS_DIR / "local"
STANDARD_PROMPTS_FILE = PROMPTS_DIR / "standard_prompts.yaml"
def get_local_prompt_path(name: str) -> Optional[Path]:
"""Find local prompt file by name."""
if not LOCAL_PROMPTS_DIR.exists():
return None
# Try different file extensions
for ext in ["yaml", "yml"]:
path = LOCAL_PROMPTS_DIR / f"{name}.{ext}"
if path.exists():
return path
# Try subdirectories
for subdir in LOCAL_PROMPTS_DIR.iterdir():
if subdir.is_dir():
for ext in ["yaml", "yml"]:
path = subdir / f"{name}.{ext}"
if path.exists():
return path
return None
def load_yaml_file(path: Path) -> Dict[str, Any]:
"""Load YAML file."""
with open(path, 'r', encoding='utf-8') as f:
return yaml.safe_load(f)
def load_prompt(
name: str,
section: Optional[str] = None,
local_only: bool = False,
fallback_to_standard: bool = True
) -> str:
"""
Load a prompt by name.
Priority:
1. prompts/local/{name}.yaml (if exists)
2. prompts/standard_prompts.yaml (if fallback_to_standard=True)
Args:
name: Prompt name (e.g., "factor_discovery")
section: Specific section in YAML (e.g., "system" or "user")
local_only: Only load from local/, raise error if not found
fallback_to_standard: If True, fall back to standard prompts
Returns:
Prompt text
Raises:
FileNotFoundError: If prompt not found
"""
# Try local prompts first
local_path = get_local_prompt_path(name)
if local_path:
print(f"✓ Loading prompt '{name}' from local: {local_path}")
data = load_yaml_file(local_path)
if section:
return data.get(section, "")
# If data is dict with 'system' and 'user', return full dict
if isinstance(data, dict):
return data
return str(data)
# Local not found
if local_only:
raise FileNotFoundError(f"Local prompt '{name}' not found in {LOCAL_PROMPTS_DIR}")
# Try standard prompts
if not fallback_to_standard:
raise FileNotFoundError(f"Prompt '{name}' not found")
if not STANDARD_PROMPTS_FILE.exists():
raise FileNotFoundError(f"Standard prompts file not found: {STANDARD_PROMPTS_FILE}")
print(f"✓ Loading prompt '{name}' from standard prompts")
data = load_yaml_file(STANDARD_PROMPTS_FILE)
# Get section from standard prompts
if name in data:
prompt_data = data[name]
if section and isinstance(prompt_data, dict):
return prompt_data.get(section, "")
return prompt_data
raise FileNotFoundError(f"Prompt '{name}' not found in standard prompts")
def load_all_prompts() -> Dict[str, Any]:
"""Load all available prompts."""
result = {}
# Load standard prompts
if STANDARD_PROMPTS_FILE.exists():
result["standard"] = load_yaml_file(STANDARD_PROMPTS_FILE)
# Load local prompts
if LOCAL_PROMPTS_DIR.exists():
result["local"] = {}
for path in LOCAL_PROMPTS_DIR.glob("*.yaml"):
result["local"][path.stem] = load_yaml_file(path)
return result
def list_available_prompts() -> Dict[str, list]:
"""List all available prompts."""
result = {"standard": [], "local": []}
# Standard prompts
if STANDARD_PROMPTS_FILE.exists():
data = load_yaml_file(STANDARD_PROMPTS_FILE)
result["standard"] = list(data.keys())
# Local prompts
if LOCAL_PROMPTS_DIR.exists():
result["local"] = [p.stem for p in LOCAL_PROMPTS_DIR.glob("*.yaml")]
return result
# Convenience functions for specific prompts
def get_factor_discovery_prompt() -> Dict[str, str]:
"""Get factor discovery prompt (system + user)."""
return load_prompt("factor_discovery")
def get_factor_evolution_prompt() -> Dict[str, str]:
"""Get factor evolution prompt."""
return load_prompt("factor_evolution")
def get_model_coder_prompt() -> Dict[str, str]:
"""Get model coder prompt."""
return load_prompt("model_coder")
def get_trading_strategy_prompt() -> Dict[str, str]:
"""Get trading strategy prompt."""
return load_prompt("trading_strategy")
# Test function
if __name__ == "__main__":
print("=== Available Prompts ===")
available = list_available_prompts()
print(f"Standard: {available['standard']}")
print(f"Local: {available['local']}")
print("\n=== Testing Prompt Load ===")
try:
prompt = load_prompt("factor_discovery")
print(f"✓ Loaded factor_discovery prompt")
print(f" System: {len(prompt.get('system', ''))} chars")
print(f" User: {len(prompt.get('user', ''))} chars")
except FileNotFoundError as e:
print(f"✗ Error: {e}")
web/index.html
+5 -5
View File
@@ -5,11 +5,11 @@
<meta charset="UTF-8" />
<link rel="icon" type="image/png" href="./src/assets/images/rd_icon.png" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<script src="https://cdnjs.cloudflare.com/ajax/libs/snap.svg/0.5.1/snap.svg-min.js"></script>
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.16.10/dist/katex.min.css" />
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/11.11.1/styles/vs.min.css" />
<title>R&D-Agent</title>
</head>
<!-- Security fix: Add SRI (Subresource Integrity) hashes to prevent tampering -->
<!-- snap.svg - SRI hash from https://www.srihash.org/ -->
<script src="https://cdnjs.cloudflare.com/ajax/libs/snap.svg/0.5.1/snap.svg-min.js"
integrity="sha512-Od9GAbPv+qjS3GvPv368l6l39d6
<body>
<div id="app"></div>