LXL·QuantAxis

AI-Native Quantitative Investment Research Platform
Convert investment ideas into validated research — without writing strategy code.

---

What It Does

You write:  "AI servers will benefit from cloud CAPEX growth"
                            │
            ┌───────────────┼───────────────┐
            ▼               ▼               ▼
     Investment Thesis   Factor Model    Strategy DSL
            │               │               │
            └───────────────┼───────────────┘
                            ▼
                    Backtest Results
                            │
                            ▼
                 Institutional Report
                    (.md + .html)

LXL·QuantAxis bridges human investment intuition and systematic quantitative research. You provide the thesis. The platform handles factor mapping, strategy construction, backtesting, and report generation — through a safe DSL that never executes AI-generated code.

Quick Demo

# One command, 7 stages, 0 code to write
$ python demo/demo_ai_research.py "AI servers benefiting from cloud CAPEX"

[1/7] Thesis Extraction       [OK]  → Note #1: AI服务器产业链看多
[2/7] Factor Mapping          [OK]  → momentum_score: 30%, trend: 25%
[3/7] Strategy Generation     [OK]  → Entry: momentum > 0.6 AND trend > 0.5
[4/7] Validation              [OK]  → All checks passed
[5/7] Backtest                [OK]  → Sharpe: 1.25
[6/7] AI Analysis             [OK]  → Viable strategy with moderate Sharpe
[7/7] Report Generation       [OK]  → reports/AI_growth_strategy.md

Complete: 7/7 stages passed

→ More examples

Quick Start

pip install -r requirements.txt

# Web UI (recommended)
python web_modern.py                # → http://127.0.0.1:5000

# CLI demo
python demo/demo_ai_research.py     # Built-in example
python demo/demo_ai_research.py "Your investment thesis"

# Interactive CLI
python main.py                      # 20+ functions menu
python main.py --research list      # View research notebook

Web Pages

Page	URL	What It Does
Research Center	/research	Type an idea → get a full research report
Classic Dashboard	/classic	Backtest, strategies, factors, diagnostics
Trading Studio	/studio	Real-time K-line charts + signal alerts
Paper Trading	/game	¥1M simulated portfolio with leaderboard
Admin	/admin	User management

Core Innovation

Safe AI Strategy DSL

AI-generated strategies use declarative rules, never executable code:

entry: "momentum_score > 0.6 AND trend_strength > 0.5"
exit:  "max_drawdown > 0.10"

Rules pass through 3 safety layers: token blocklist → AST allowlist → factor whitelist. Zero exec/eval.

Research Pipeline (7 stages)

1. Thesis Extraction — Natural language → structured investment thesis
2. Factor Mapping — Thesis → 28-factor registry with style templates
3. Strategy Builder — Factors → safe DSL rules
4. Validation — Syntax, factors, parameters, risk checks
5. Backtest — T+1 execution, A-share cost model, benchmark metrics
6. AI Analysis — Metrics → strengths, weaknesses, suggestions
7. Report Generation — 8-section institutional report (Markdown + HTML)

Research Cases

Three detailed walkthroughs showing the complete workflow:

#	Case	Style
1	AI Infrastructure Supply Chain	Growth
2	Consumer Sector Value Recovery	Value
3	Semiconductor Cycle Bottom	Macro-Momentum

Architecture

┌──────────────────────────────────────────────────┐
│           Web UI / CLI / Desktop GUI              │
├──────────────────────────────────────────────────┤
│  V2 Research Layer (src/lxl_quantaxis/)           │
│  AI Pipeline · Strategy DSL · Portfolio Intel     │
├──────────────────────────────────────────────────┤
│  V1 Quant Engine (src/)                           │
│  28 Factors · 16 Strategies · Backtest Engine     │
├──────────────────────────────────────────────────┤
│  Data: akshare (A-share) · yfinance (US/HK)       │
│  Storage: SQLite ×10 · CSV Cache                  │
└──────────────────────────────────────────────────┘

Full architecture documentation →

Features

Category	Capabilities
AI Research	Thesis extraction, factor mapping, strategy DSL, backtest analysis, report generation
Factors	28 factors (trend, momentum, volatility, volume, pattern, sentiment, fundamental)
Strategies	16 strategies (7 classic, 5 advanced, 4 factor-composed) with V2 compiler
Backtest	T+1 execution, A-share cost model, benchmark metrics, signal lag queue
Portfolio	4 allocation models, walk-forward, factor exposure, diversification scoring
Risk	Pre-trade gate (6 checks), trailing stop, circuit breaker, Kelly sizing
Research	Immutable notebook, thesis builder, AI parser, correlation analyzer

Tech Stack

Python 3.12 · Flask · Pandas/NumPy · Plotly · SQLite · JWT/bcrypt · akshare/yfinance · SciPy

Roadmap

Version	Focus
v2.0.0 (current)	AI pipeline, safe DSL, research notebook, portfolio intelligence
v2.1	Real fundamental data (financial statements, macro), live paper trading
v2.2	Multi-step AI agent with iterative refinement, Docker support
v3.0	Collaborative research, multi-user notebooks, cloud deployment

Documentation

Document	Content
Architecture V2	System design with Mermaid diagrams
AI Pipeline	7-stage pipeline detail
System Design	Engineering principles
Research Cases	3 complete research walkthroughs
Contributing	Dev setup, conventions
Changelog	All versions
Release Notes	v2.0.0 details

License

MIT · Ryhs666