Institutional Distributed Quantitative Alpha Mining & Portfolio Orchestration Engine for WorldQuant BRAIN
Built natively in Rust. High-throughput, concurrent, and mathematically rigorous.

Quickstart · CLI Reference · Methodology · Architecture · Knowledge Graph

---

📌 About

AlphaFind is an institutional-grade quantitative alpha discovery and portfolio orchestration framework engineered natively in Rust for the WorldQuant BRAIN algorithmic trading platform.

Quantitative research at scale requires exploring thousands of mathematical formulations, optimizing decay parameters, eliminating self-correlation, and adhering to strict platform risk constraints. Performing this through manual web interfaces or ad-hoc scripts leads to quota starvation, portfolio dilution, and execution errors.

AlphaFind provides a high-performance CLI engine that coordinates scalable asynchronous concurrency (up to 9 parallel workers), sweeps decay ranges, applies convex non-linear transformations, audits pairwise Pearson correlation in microseconds, and pre-validates all 8/8 mandatory In-Sample platform checks prior to Out-of-Sample submission.

---

⚡ Why AlphaFind

Capability	Ad-hoc Scripts / Web UI	AlphaFind Engine
Execution Architecture	Single-threaded, synchronous	9-Worker Tokio Async Event Loop
Worker Orchestration	Single session, manual queueing	Multi-Session Async Pooling (Configurable workers)
Transfer Safety	Manual copy-pasting across tabs	Atomic Transfer Lock for Main verification
Correlation Auditing	Slow web UI / manual calculations	Microsecond Vector Pearson Matrix ($\le 0.15$ target)
Platform Compliance	Trial-and-error submission rejections	Automated 8/8 Pre-Validation (alphafind check)
Decay & Convexity Tuning	Manual parameter guessing	Autonomous Decay Sweeps & Power Transformations
Memory Footprint	Heavy Python / Electron runtime	Native Rust Binary (~15 MB RAM, zero GC pause)

---

📊 Supported Factor Pillars & Universes

Factor Pillars

Pillar	Economic Focus & Underlying Data	Default Universe	Neutralization
OPTIONS	Volatility Surface, Term Structure, Skew, VRP	TOP1000 / TOP3000	MARKET / SUBINDUSTRY
ANALYST	Consensus EPS Revisions, Price Targets, Revisions Drift	TOP500	SECTOR / MARKET
MICRO	VWAP Slippage, Order Imbalance, Liquidity Shocks	TOP1000	SUBINDUSTRY
QUAL	Fundamental Quality, Accruals, Cash Flow Spreads	TOP1000 / TOP500	SUBINDUSTRY
RISK	Idiosyncratic Risk, Unsystematic Volatility Curvature	TOP1000 / TOP500	SUBINDUSTRY
SHORT	Short Interest Dynamics, Borrow Supply Constraints	TOP2000 / TOP1000	MARKET

Equity Universes

Universe	Description	Institutional Role
TOP3000	Broad-market US equities	Micro/small-to-large cap idiosyncratic signals
TOP1000	Liquid large and mid-cap US equities	Core institutional testing universe (default)
TOP500	Large-cap US equities (S&P 500 equivalent)	High-capacity fundamental & analyst revisions
TOP200	Mega-cap US equities	Highest liquidity, lowest market impact

---

🚀 Quickstart

1-Line Automated Install

Deploy instantly on macOS or Linux with live progress tracking:

curl -# -fsSL https://raw.githubusercontent.com/vtp772002/alphafind/main/install.sh | bash

  [1/5] Verifying toolchain        [██████████████████████] 100% Rust 1.85.0 (OK)
  [2/5] Downloading repository     [██████████████████████] 100% Complete
  [3/5] Compiling release engine   [██████████████████████] 100% Optimized build ready
  [4/5] Installing binary to PATH  [██████████████████████] 100% ~/.cargo/bin/alphafind
  [5/5] Configuring environment    [██████████████████████] 100% .env initialized

3-Step Workflow

# 1. Configure BRAIN credentials
cp .env.example .env && nano .env

# 2. Authenticate all configured accounts
alphafind auth

# 3. Launch distributed 9-worker screening
alphafind screen --pillar OPTIONS --universe TOP1000 --min-fitness 1.50

👉 For manual compilation and detailed prerequisites, see Quickstart Guide.

---

📖 CLI Commands at a Glance

Command	Syntax	Primary Purpose
auth	alphafind auth	Authenticate all configured BRAIN accounts concurrently
sync	alphafind sync	Synchronize active Out-of-Sample portfolio into local cache
portfolio	alphafind portfolio	Simulate merged multi-alpha Out-of-Sample portfolio Sharpe & risk
sim	alphafind sim --expr "..."	Simulate an individual FastExpr formula backtest directly
screen	alphafind screen --pillar <P>	Launch distributed candidate screening with async concurrency
check	alphafind check <ALPHA_ID>	Verify the 8 mandatory In-Sample platform submission checks
audit	alphafind audit <ALPHA_ID>	Run 1,236-day Pearson correlation audit against active portfolio
submit	alphafind submit <ALPHA_ID>	Submit an Alpha to Out-of-Sample (OS) after 8/8 PASS verification

👉 For exhaustive options, flags, and workflow examples, see CLI Command Reference.

---

🏛️ System Architecture

┌────────────────────────────────────────────────────────────────────────┐
│        TIER 1: TAXONOMY & ORTHOGONAL HYPOTHESIS ENGINE                 │
│   ├── src/taxonomy.rs   : 6 Economic Factor Pillars & Hypothesis Gen   │
│   ├── docs/BRAIN_KNOWLEDGE_GRAPH.md : 66 Operators & Diagnostics       │
│   └── data/             : Local Field Schemas & Profiling Metadata     │
└───────────────────────────────────┬────────────────────────────────────┘
                                    │
┌───────────────────────────────────┴────────────────────────────────────┐
│        TIER 2: DISTRIBUTED COMPUTE ENGINE (Multi-Worker Concurrency)   │
│   ├── src/screener.rs   : Asynchronous Parallel Tokio 9-Worker Engine  │
│   ├── src/client.rs     : Async BrainClient + Session Cookie + Backoff │
│   ├── src/autotuner.rs  : Autonomous Decay Sweeps & Convex Optimizer   │
│   └── transfer_lock     : Thread-Safe Verification Lock on Main        │
└───────────────────────────────────┬────────────────────────────────────┘
                                    │
┌───────────────────────────────────┴────────────────────────────────────┐
│        TIER 3: CORRELATION ENGINE & AUDIT GUARD                        │
│   ├── src/correlation.rs: Microsecond Pearson Matrix & Merge Simulator │
│   ├── ~/.cache/alphafind: Disk-Cached Daily PnL Baseline               │
│   └── portfolio_os.json : Master Active Out-of-Sample Portfolio Cache  │
└───────────────────────────────────┬────────────────────────────────────┘
                                    │
┌───────────────────────────────────┴────────────────────────────────────┐
│        TIER 4: SUBMISSION VERIFICATION & DISPATCH                      │
│   ├── src/main.rs       : Unified High-Performance CLI (alphafind)     │
│   ├── GET /check        : Official 8/8 Mandatory In-Sample PASS Engine │
│   └── POST /submit      : Direct API Submission & Institutional Tagging│
└────────────────────────────────────────────────────────────────────────┘

👉 For full concurrency diagrams and engine specifications, see Architecture Guide.

---

📚 Documentation Index

All technical and quantitative documentation is organized within the docs/ directory:

Document	Description
🚀 Quickstart Guide	Installation options, .env setup, toolchain requirements, and first run
📖 CLI Command Reference	Exhaustive parameter descriptions, CLI syntax, and practical examples
📐 Quantitative Methodology	Orthogonal Variance Law math, factor taxonomy, and 8 In-Sample checks
🏛️ System Architecture	9-worker concurrency model, atomic transfer lock, and SIMD Pearson engine
🧠 FASTEXPR Knowledge Graph	Curated selection of core platform operators and diagnostic heuristics

---

🛡️ Security & Privacy

1. Zero Credential Persist:
   • Credentials in .env are loaded strictly into memory at runtime and never logged or committed.
   • .env is permanently excluded via .gitignore.
2. Ephemeral Research Sandbox:
   • Disposable exploratory scripts are restricted to the git-ignored scratch/ workspace and purged post-discovery.
   • Proprietary alpha ledgers and active portfolio cache files are strictly git-ignored.
3. Institutional OPSEC:
   • Repository code contains zero hardcoded account IDs, leaderboard metrics, or private portfolio data.

---

⭐ Star History

---

⚖️ License & Disclaimer

This project is licensed under the MIT License — see the LICENSE file for details.

Disclaimer: AlphaFind is an independent research framework designed for automated quantitative analysis and workflow efficiency on the WorldQuant BRAIN platform. Users are solely responsible for compliance with WorldQuant BRAIN platform terms of service, competition guidelines, intellectual property rules, and submission quotas.