How luckyone7777/llm-trading-lab works
LLM Trading Lab combines a documented live micro-cap trading experiment and its evaluation with a newer multi-model IPO experiment framework. The IPO workflow selects a cadence, prepares prompts and candidate data, obtains model-generated orders, validates buys, and delegates portfolio processing and persistence to LIBB. The README documents the historical experiment’s accounting, stop-loss, benchmark, and analytics capabilities. Some of those paths are represented by directories rather than sampled implementations, so their internal wiring is not inferred.
Main entry points: workflow.py Cadence workflow · order_verification.py Order verification · prompts Prompt modes · main_functions.py IPO universe enrichment
Components 22
IPO Experiment
- Cadence workflow
Experiments/multi_model_ipo/workflow.py - Prompt modes
Experiments/multi_model_ipo/prompts - LIBB portfolio engine
- Order verification
Experiments/multi_model_ipo/miscellaneous/order_verification.py - Rejection logging
Experiments/multi_model_ipo/miscellaneous/csv_conversion.py - Research outputs
Experiments/multi_model_ipo
Decision Inputs
- IPO universe enrichment
Experiments/multi_model_ipo/prompt_orchestration/main_functions.py - Market data fetching
Experiments/multi_model_ipo/prompt_orchestration/get_prompt_data/fetching.py - Eligibility utilities
Experiments/multi_model_ipo/prompt_orchestration/get_prompt_data/utilities.py - Snapshot cache
Experiments/multi_model_ipo/miscellaneous/order_verification.py
Micro-Cap Record
- Live trading engine
Experiments/chatgpt_micro-cap/scripts/processing/trading_script.py - Portfolio processing
Experiments/chatgpt_micro-cap/scripts/processing/ProcessPortfolio.py - Daily portfolio records
Experiments/chatgpt_micro-cap/csv_files/Daily Updates.csv - Trade log
Experiments/chatgpt_micro-cap/csv_files/Trade Log.csv - Research and decisions
Experiments/chatgpt_micro-cap/collected_artifacts
Performance Analysis
- Research evaluation
Experiments/chatgpt_micro-cap/evaluation - Portfolio metrics
Experiments/chatgpt_micro-cap/scripts/metrics - Performance charts
Experiments/chatgpt_micro-cap/graphing
Other components
- Experiment operator
- Financial data services
- Market data services
- Market benchmarks
Connections 24
- Experiment operator → Cadence workflow: runs
- Cadence workflow → Prompt modes: selects mode
- Prompt modes → IPO universe enrichment: requests inputs
- IPO universe enrichment → Market data fetching: enriches candidates
- Market data fetching → Financial data services: fetches profiles
- Market data fetching → Market data services: fetches prices
- IPO universe enrichment → Eligibility utilities: validates eligibility
- Prompt modes → LIBB portfolio engine: builds portfolio prompt
- Cadence workflow → LIBB portfolio engine: processes portfolio
- LIBB portfolio engine → Cadence workflow: returns report
- Cadence workflow → Order verification: filters orders
- Order verification → Financial data services: checks order data
- Order verification → Market data services: checks prices
- Order verification → Snapshot cache: reads and writes cache
- Cadence workflow → Rejection logging: saves rejections
- Cadence workflow → LIBB portfolio engine: saves orders and analysis
- LIBB portfolio engine → Research outputs: persists outputs
- Live trading engine → Daily portfolio records: records accounting
- Live trading engine → Trade log: records trades
- Portfolio processing → Daily portfolio records: processes portfolio data
- Research and decisions → Research evaluation: documents experiment
- Daily portfolio records → Portfolio metrics: supplies portfolio data
- Portfolio metrics → Performance charts: provides analytics
- Performance charts → Market benchmarks: compares performance