174 AI Experts + 200 Citizen Agents. Zero API Cost. Predict Anything — On Your Laptop.

Where the crowd is wrong, the alpha lives.

Quick Start • Live Demo • How It Works • The Comparison • Unique Agents • Contribute

Every prediction tool tells you what the crowd thinks. BlackSwanX tells you where the crowd is wrong.

We don't seek consensus. We seek the widest gap — the Cognitive Dissonance between what the masses believe and what the experts fear. That gap is where the alpha lives.

The Comparison

BettaFish MiroFish BlackSwanX

Cost

Setup time 30+ min + PostgreSQL 15 min + Zep account 2 min, zero config

Expert agents 5 0 (generic personas) 174 domain experts

Citizen agents 0 ~100 per run (OASIS) 200 per run (Shadow Swarm)

Citizen simulation None OASIS framework Shadow Swarm (200 diverse citizens)

Live simulation feed None Static graph Real-time Twitter-like feed

Self-learning No No SONA Auditor (stateful, persists across runs)

Adversarial testing No No Kill-Switch + BlackSwan Assassin

Cognitive Dissonance No No The Trap / The Blindspot / The Chaos

Stress testing No No BlackSwan What-If Injector

Antifragile Play No No How to profit from collapse

User Profiles No No Personalized predictions based on YOUR life

Auto Agent Routing No No NEXUS picks relevant agents per topic

Financial Agents No No Monte Carlo, Quant, Boom/Bust, Value Investor, Whale Tracker, Panic Seller

Speed modes No No Turbo (25, ~2 min) / Standard (50, ~5 min) / Deep (200, ~25 min)

Data sources Chinese social media Chinese social media DuckDuckGo + Reddit + HN + YouTube (5 free sources)

Runs 100% locally No No Yes — your data never leaves your machine

Database needed PostgreSQL required Zep Cloud required None — SQLite built into Python

File upload No Seed documents PDF, TXT, CSV, MD

Coverage Chinese only Chinese only Global (any language)

Live Demo

Clone it, run bash start.sh, and try it yourself in 2 minutes.

What It Looks Like

Dashboard — Hardware auto-detection, user profile, turbo/standard/deep mode, pipeline progress:

Pipeline in Action — Real-time stage tracking (Crawling → Compressing → Assassin → Swarm → Synthesis):

Live Simulation Feed — Watch citizens post opinions in real-time with emotions, hot takes, and sentiment:

174 Agent Library — Browse all experts by division, search by keyword, click to chat:

Decision-Ready Map — Linchpin, 20 Active Agents, Kill Shot with cascade, Cognitive Dissonance Matrix:

Predictions + NEXUS DAG — Antifragile Play, 3 scenarios (Best/Likely/Worst), interactive force graph:

Shadow Swarm Citizens — 200 diverse personas react with emotions, hot takes, and sentiment scores:

Quick Start (2 Minutes)

# 1. Clone git clone https://github.com/Kalki-M/BlackSwanX.git cd BlackSwanX

# 1. Clone git clone https://github.com/Kalki-M/BlackSwanX.git cd BlackSwanX

2. Pull models (one-time download, ~25GB total)

ollama pull llama3.2:3b # The Swarm — fast citizen simulation ollama pull phi4:14b # The Assassin — deep kill shot reasoning ollama pull mistral-small:24b # The Nexus Brain — synthesis & DAG

3. Install & Run

python3 -m venv .venv && source .venv/bin/activate pip install -r requirements.txt bash start.sh

4. Open http://localhost:9100` [blocked]

No API keys. No database setup. No Docker required. No cloud account needed.

That's it.

How It Works

BlackSwanX runs an 8-stage adversarial intelligence pipeline:

YOU  │  ▼  ┌───────────────┐  │ 1. CRAWL │ 5 sources: DuckDuckGo + Reddit + HN + YouTube + Twitter  │ + NOISE │ All free, no API keys needed  │ CHECK │ Auto re-crawls with better terms if data is noise  └───────┬───────┘  ▼  ┌───────────────┐  │ 2. COMPRESS │ 10,000 posts → 5 "Social Personas"  │ (90% token │ via HDBSCAN clustering  │ savings) │ Only the vibes hit the reasoning model  └───────┬───────┘  ▼  ┌───────────────┐  │ 3. ASSASSIN'S │ BlackSwan Assassin identifies the Kill Shot  │ MARK │ BEFORE the swarm runs (sets the target)  │ │ Uses phi4:14b for deep reasoning  └───────┬───────┘  ▼  ┌───────────────┐  │ 4. SHADOW │ Up to 200 citizens in waves of 10  │ SWARM │ Turbo: 25 | Standard: 50 | Deep: 200  │ │ Forced diversity (bull/bear/mixed)  │ │ GPU cache flushed between waves  │ │ Live feed shows each opinion in real-time  └───────┬───────┘  ▼  ┌───────────────┐  │ 5. COGNITIVE │ THE TRAP: crowd euphoria vs expert fear  │ DISSONANCE │ THE BLINDSPOT: consensus ignoring kill shot  │ MATRIX │ THE CHAOS: expert disagreement variance  │ │ Calculated from REAL citizen data, not LLM  └───────┬───────┘  ▼  ┌───────────────┐  │ 6. NEXUS │ Synthesizes everything into a DAG  │ SYNTHESIS │ Finds the LINCHPIN everything depends on  │ │ Generates the ANTIFRAGILE PLAY  └───────┬───────┘  ▼  ┌───────────────┐  │ 7. SONA │ Audits every agent after each run  │ AUDIT │ Boosts agents that caught risks (up to 2.0x)  │ (Stateful) │ Demotes agents that missed threats (down to 0.3x)  │ │ Stores patterns in ReasoningBank (SQLite)  │ │ PERSISTS across runs — system gets SMARTER  └───────┬───────┘  ▼  DECISION-READY MAP  "The Bull Case rests on [Linchpin].  The Swarm is 72% Bullish, but the  Assassin found a 15% probability of  [Kill Shot]. Delta is HIGH.  Action: [Antifragile Play]."

YOU  │  ▼  ┌───────────────┐  │ 1. CRAWL │ 5 sources: DuckDuckGo + Reddit + HN + YouTube + Twitter  │ + NOISE │ All free, no API keys needed  │ CHECK │ Auto re-crawls with better terms if data is noise  └───────┬───────┘  ▼  ┌───────────────┐  │ 2. COMPRESS │ 10,000 posts → 5 "Social Personas"  │ (90% token │ via HDBSCAN clustering  │ savings) │ Only the vibes hit the reasoning model  └───────┬───────┘  ▼  ┌───────────────┐  │ 3. ASSASSIN'S │ BlackSwan Assassin identifies the Kill Shot  │ MARK │ BEFORE the swarm runs (sets the target)  │ │ Uses phi4:14b for deep reasoning  └───────┬───────┘  ▼  ┌───────────────┐  │ 4. SHADOW │ Up to 200 citizens in waves of 10  │ SWARM │ Turbo: 25 | Standard: 50 | Deep: 200  │ │ Forced diversity (bull/bear/mixed)  │ │ GPU cache flushed between waves  │ │ Live feed shows each opinion in real-time  └───────┬───────┘  ▼  ┌───────────────┐  │ 5. COGNITIVE │ THE TRAP: crowd euphoria vs expert fear  │ DISSONANCE │ THE BLINDSPOT: consensus ignoring kill shot  │ MATRIX │ THE CHAOS: expert disagreement variance  │ │ Calculated from REAL citizen data, not LLM  └───────┬───────┘  ▼  ┌───────────────┐  │ 6. NEXUS │ Synthesizes everything into a DAG  │ SYNTHESIS │ Finds the LINCHPIN everything depends on  │ │ Generates the ANTIFRAGILE PLAY  └───────┬───────┘  ▼  ┌───────────────┐  │ 7. SONA │ Audits every agent after each run  │ AUDIT │ Boosts agents that caught risks (up to 2.0x)  │ (Stateful) │ Demotes agents that missed threats (down to 0.3x)  │ │ Stores patterns in ReasoningBank (SQLite)  │ │ PERSISTS across runs — system gets SMARTER  └───────┬───────┘  ▼  DECISION-READY MAP  "The Bull Case rests on [Linchpin].  The Swarm is 72% Bullish, but the  Assassin found a 15% probability of  [Kill Shot]. Delta is HIGH.  Action: [Antifragile Play]."

The Three-Model Strategy

BlackSwanX uses three specialized local models. Auto-detected based on your hardware.

Role Model Purpose RAM

The Swarm llama3.2:3b Up to 200 biased, emotional citizen agents ~2GB

The Assassin phi4:14b Kill shots, cascade analysis, stress tests ~8GB

The Nexus Brain mistral-small:24b Elite synthesis, DAG construction ~14GB

Models load on-demand and flush between stages (keep_alive: 0). Runs on a MacBook with 16GB RAM.

Hardware Auto-Detection

Your Setup RAM What BlackSwanX Uses

MacBook Air M1 8GB Swarm only (llama3.2:3b for everything)

MacBook Pro M2 16GB Full 3-model pipeline

Mac Studio M2 Ultra 64GB Full pipeline + larger models

Linux + RTX 4090 24GB VRAM Full pipeline (NVIDIA detected)

You don't configure anything. BlackSwanX detects your hardware and picks the optimal models automatically.

174 Expert Agents

BlackSwanX includes the complete agency-agents library plus custom agents, organized into 13 divisions:

Division Count Key Agents

Specialized 41 CFO, Fundraising, Tax, Monte Carlo, Quant Analyst, Boom & Bust Historian, Value Investor, Whale Tracker, Panic Seller, Vedic Astrologer, Chaos Mathematician

Marketing 27 Growth Hacker, SEO, TikTok, Reddit, Instagram, LinkedIn, Content Creator

Engineering 23 Backend Architect, AI Engineer, Security Engineer, DevOps, Mobile

Game Dev 20 Unity, Unreal, Godot, Roblox, Narrative Design, Audio

Design 8 UX Architect, Brand Guardian, UI Designer, Visual Storyteller

Sales 8 Deal Strategist, Sales Coach, Pipeline Analyst, Outbound Strategy

Testing 8 Reality Checker, Performance Benchmarker, API Tester, Evidence Collector

Paid Media 7 PPC, Programmatic, Paid Social, Creative Strategy, Tracking

Support 6 Analytics Reporter, Infrastructure, Finance Tracker, Legal Compliance

Project Mgmt 6 Project Shepherd, Studio Producer, Jira Steward, Sprint Prioritizer

Spatial 6 VisionOS, XR, Metal, WebXR, Terminal Integration

Product 5 Product Manager, Feedback Synthesizer, Trend Researcher

Academic 5 Psychologist, Historian, Anthropologist, Narratologist, Geographer

Plus custom elite prompts: Agent Provocateur, Sentiment Whale, Catalyst, Chaos Mathematician, Street Smart Hustler, Chief Economist, Market Analyst, VC Partner, Geopolitical Strategist, Tech Analyst, Regulatory Analyst, Crypto Strategist, Floor Trader, Therapist, NEXUS Orchestrator, and the BlackSwan Assassin.

Agents You Won't Find Anywhere Else

These are the agents that make BlackSwanX different from every other tool:

Agent What It Does Why It Exists

🦋 Chaos Mathematician Finds tipping points, cascade failures, butterfly effects, power laws Linear models miss the nonlinear dynamics that actually move markets

💪 Street Smart Hustler No-MBA reality checks from someone who bootstrapped from zero "Your pitch deck is pretty. Now show me your bank account."

🔮 Vedic Astrologer Planetary transits, Saturn returns, cosmic cycles for timing Markets follow psychology. Psychology follows cycles. Fight me.

🧘 Therapist & Counselor CBT/ACT frameworks for founder anxiety and decision paralysis The psychological dimension that spreadsheets miss

📱 Gen Z Culture Decoder Vibe checks, cringe scores, virality prediction, meme analysis If it's not on TikTok, it doesn't exist

🕉️ Spiritual Philosopher Stoicism, Buddhism, Taoism applied to business decisions Before asking "will it work?" — ask "does it matter?"

🎯 Career Strategist Non-linear career paths, skill stacking, salary negotiation Your career is a portfolio of bets, not a ladder

Unique Features

Kill-Switch (Narrative Stress Test)

Two swarms fight: Fans build the hype case, Assassins find the kill shot. A War Room judge scores the Volatility Score. If the Fans can't defend against the Assassins, the narrative is fragile.

BlackSwan Assassin

Assumes the consensus is wrong. Traces second and third-order cascade effects. Finds the Founder's Blindspot and the Trader's Trap. Outputs the Antifragile Play — how to profit from collapse.

Cognitive Dissonance Matrix

Delta Measures Signal

The Trap Crowd sentiment vs Expert sentiment Crowd euphoric + experts scared = TOP IS IN

The Blindspot Consensus strength vs Kill Shot feasibility Everyone ignoring a viable threat = DANGER

The Chaos Variance among citizen opinions Citizens fundamentally disagree = HIGH VOLATILITY

Calculated from real citizen data — not hallucinated by an LLM.

SONA (Self-Optimizing Neural Auditor) — Stateful & Self-Learning

SONA is why BlackSwanX gets smarter with every run. Unlike MiroFish and BettaFish that start fresh every time, SONA persists learning across all analyses in a local SQLite database.

After every run, SONA:

• Audits every citizen: Did they catch the kill shot? Were they contrarian during high dissonance?

• Boosts winners: Citizens that caught risks others missed get up to 2.0x weight in future runs

• Demotes losers: Citizens that missed critical threats drop to 0.3x weight

• Stores patterns in a ReasoningBank: "Last time dissonance was 80+ on a crypto topic, the kill shot was regulatory"

• Domain-specific learning: Finance, tech, politics, career — each has its own weight table

• Stateful: The more you use it, the sharper it gets. Run 10 crypto analyses and SONA learns which personas catch crypto risks best.

Check what SONA has learned: GET /api/sona/status

Live Simulation Feed

Watch 200 citizen agents post opinions in real-time — like scrolling Twitter during a crisis. Each citizen has an emotion avatar, sentiment badge, hot takes, and confidence scores. The BlackSwan Kill Shot appears mid-feed like a breaking news alert. NEXUS shows which agents it activated and why.

User Profiles — Personalized Predictions

Set up your profile once (age, role, income, savings, risk tolerance, goals). Every prediction is personalized to YOUR life. The Therapist knows your burnout risk. The Career Strategist knows your runway. The Street Hustler knows if you can afford to fail. Profile persists in local SQLite — never leaves your machine.

Auto Agent Routing

NEXUS auto-selects relevant agents per topic. Ask about NVIDIA? It activates Economist, Quant Analyst, Monte Carlo, Boom & Bust Historian, Whale Tracker, and 15 more financial agents. Ask about quitting your job? It activates Career Strategist, Therapist, VC Partner, and Street Hustler. You see exactly which agents are active in the live feed.

Fast & Deep Modes

• Turbo Mode (25 citizens, ~2 min): Quick demos and screenshots

• Standard Mode (50 citizens, ~5 min): Good for everyday analysis

• Deep Mode (200 citizens, ~25 min): Full census, 200 diverse personas from billionaire assistants to submarine captains

5 Free Data Sources (No API Keys)

Source What It Gets

DuckDuckGo Web Current search results

DuckDuckGo News Last 24-72 hours of news

Reddit Public posts sorted by relevance

Hacker News Stories via free Algolia API

YouTube Video titles from search results

Zero-Config Storage

Everything persists in a single SQLite file — no database to install, configure, or manage. Topics, results, user profiles, SONA learning, live feed events — all in one file that grows smarter with every run.

Use Cases

Who Question What They Get

Traders "Will BTC crash after ETF outflows?" Kill Shot + Antifragile Play + Volatility Score

Founders "Should I pivot to AI agents?" 200 simulated users react + VC Partner + Street Hustler reality check

VCs "Is this startup's TAM real?" Chaos Mathematician finds the tipping point + Therapist catches founder ego

PR Teams "Will this scandal destroy us?" Kill-Switch test + Gen Z Culture Decoder predicts virality

Policy Makers "How will voters react to this regulation?" Shadow Swarm census + Geopolitical Strategist + Regulatory Analyst

Career Changers "Should I quit FAANG for a startup?" Career Strategist + Therapist + Vedic Astrologer (why not?)

Anyone "What's the next big thing?" 174 experts + 200 citizens + Assassin stress test

File Upload

Upload any document alongside your topic:

• PDF — annual reports, research papers, pitch decks

• TXT/MD — notes, meeting transcripts, strategy docs

• CSV — financial data, survey results, metrics

BlackSwanX combines your private document with live web data + citizen simulation + adversarial analysis. Your documents never leave your machine.

Project Structure

BlackSwanX/ ├── backend/ │ ├── app_simple.py # Flask server (API + UI + profile + SONA endpoint) │ ├── pipeline_sync.py # 8-stage NEXUS pipeline (the brain) │ ├── hardware.py # Auto-detects your GPU/RAM, picks models │ ├── config.py # Settings (Pydantic) │ ├── signature.py # 👀 dig deep and you'll find the builder │ ├── agents/ │ │ └── registry.py # Auto-loads 174 agents from markdown files │ ├── compression/ │ │ ├── embedder.py # Sentence embeddings via Ollama │ │ ├── clusterer.py # HDBSCAN clustering → Social Personas │ │ └── summarizer.py # Persona compressor (10K posts → 5 vibes) │ ├── crawler/ │ │ ├── duckduckgo.py # Web + news search (free) │ │ ├── reddit.py # Public Reddit JSON API (free) │ │ ├── hackernews.py # Algolia HN API (free) │ │ ├── youtube.py # YouTube search scraper (free) │ │ ├── twitter.py # Nitter-based Twitter scraper (free) │ │ └── unified.py # Parallel crawl orchestrator │ ├── llm/ │ │ ├── client.py # Triple-model Ollama client (swarm/assassin/nexus) │ │ ├── router.py # Semantic Router (routes cheap vs reasoning tasks) │ │ └── prompts.py # 20+ elite prompts + NEXUS + BlackSwan Assassin │ ├── simulation/ │ │ ├── nexus.py # 8-stage async state machine │ │ ├── swarm.py # Wave-based 200-citizen simulation │ │ ├── elite_panel.py # Domain expert analysis with auto-routing │ │ ├── delta.py # Cognitive Dissonance Matrix (Trap/Blindspot/Chaos) │ │ ├── kill_switch.py # Fans vs Assassins → Volatility Score │ │ ├── blackswan.py # What-If stress testing + cascade analysis │ │ └── sona.py # SONA: self-learning auditor + ReasoningBank (stateful) │ ├── report/ │ │ └── generator.py # Decision-Ready Map renderer │ └── templates/ │ └── index.html # Full UI (dashboard + profile + live feed + D3 + report + agents browser + chat) ├── agency-agents/ # 174 agent definitions (auto-loaded markdown files) │ ├── academic/ # 5 agents (psychologist, historian, etc.) │ ├── design/ # 8 agents │ ├── engineering/ # 23 agents │ ├── marketing/ # 27 agents │ ├── sales/ # 8 agents │ ├── specialized/ # 41 agents (CFO, Monte Carlo, Astrologer, etc.) │ ├── testing/ # 8 agents │ └── ... # 13 divisions total ├── start.sh # One-command startup (both backend + frontend) ├── docker-compose.yml # Docker deployment ├── Dockerfile # Backend container ├── requirements.txt # Python dependencies (no database needed) ├── .env.example # Configuration template ├── CONTRIBUTING.md # How to add agents and contribute ├── LICENSE # MIT └── README.md # You are here

BlackSwanX/ ├── backend/ │ ├── app_simple.py # Flask server (API + UI + profile + SONA endpoint) │ ├── pipeline_sync.py # 8-stage NEXUS pipeline (the brain) │ ├── hardware.py # Auto-detects your GPU/RAM, picks models │ ├── config.py # Settings (Pydantic) │ ├── signature.py # 👀 dig deep and you'll find the builder │ ├── agents/ │ │ └── registry.py # Auto-loads 174 agents from markdown files │ ├── compression/ │ │ ├── embedder.py # Sentence embeddings via Ollama │ │ ├── clusterer.py # HDBSCAN clustering → Social Personas │ │ └── summarizer.py # Persona compressor (10K posts → 5 vibes) │ ├── crawler/ │ │ ├── duckduckgo.py # Web + news search (free) │ │ ├── reddit.py # Public Reddit JSON API (free) │ │ ├── hackernews.py # Algolia HN API (free) │ │ ├── youtube.py # YouTube search scraper (free) │ │ ├── twitter.py # Nitter-based Twitter scraper (free) │ │ └── unified.py # Parallel crawl orchestrator │ ├── llm/ │ │ ├── client.py # Triple-model Ollama client (swarm/assassin/nexus) │ │ ├── router.py # Semantic Router (routes cheap vs reasoning tasks) │ │ └── prompts.py # 20+ elite prompts + NEXUS + BlackSwan Assassin │ ├── simulation/ │ │ ├── nexus.py # 8-stage async state machine │ │ ├── swarm.py # Wave-based 200-citizen simulation │ │ ├── elite_panel.py # Domain expert analysis with auto-routing │ │ ├── delta.py # Cognitive Dissonance Matrix (Trap/Blindspot/Chaos) │ │ ├── kill_switch.py # Fans vs Assassins → Volatility Score │ │ ├── blackswan.py # What-If stress testing + cascade analysis │ │ └── sona.py # SONA: self-learning auditor + ReasoningBank (stateful) │ ├── report/ │ │ └── generator.py # Decision-Ready Map renderer │ └── templates/ │ └── index.html # Full UI (dashboard + profile + live feed + D3 + report + agents browser + chat) ├── agency-agents/ # 174 agent definitions (auto-loaded markdown files) │ ├── academic/ # 5 agents (psychologist, historian, etc.) │ ├── design/ # 8 agents │ ├── engineering/ # 23 agents │ ├── marketing/ # 27 agents │ ├── sales/ # 8 agents │ ├── specialized/ # 41 agents (CFO, Monte Carlo, Astrologer, etc.) │ ├── testing/ # 8 agents │ └── ... # 13 divisions total ├── start.sh # One-command startup (both backend + frontend) ├── docker-compose.yml # Docker deployment ├── Dockerfile # Backend container ├── requirements.txt # Python dependencies (no database needed) ├── .env.example # Configuration template ├── CONTRIBUTING.md # How to add agents and contribute ├── LICENSE # MIT └── README.md # You are here

Requirements

• Python 3.11+ (comes with SQLite built-in — no database to install)

• Ollama (brew install ollama or ollama.com)

• 8GB+ RAM (16GB recommended for full 3-model pipeline)

• Apple Silicon (M1/M2/M3/M4), Linux with NVIDIA GPU, or any system that runs Ollama

You do NOT need: PostgreSQL, MySQL, Redis, Docker, API keys, cloud accounts, or any external services.

Contributing

PRs welcome. See CONTRIBUTING.md. Key areas:

• New agents — What expert perspective is missing? Add a .md file and it auto-loads

• Agent-to-agent debate — Citizens that argue with each other across rounds

• 3D force graph — Three.js version of the DAG with animated timeline

• Backtest framework — Compare past predictions vs actual outcomes

• Deeper crawlers — YouTube comments (not just titles), full Twitter threads, LinkedIn

• WebSocket streaming — Replace polling with real-time WebSocket feed

• Mobile responsive UI — Works on phones

• Export reports — PDF/Markdown/JSON export of Decision-Ready Maps

• More citizen personas — Always room for more diversity in the Shadow Swarm

FAQ

Q: Is the Vedic Astrologer serious? A: The astrologer is labeled as entertainment/pattern-recognition, not science. But markets follow psychology, psychology follows cycles, and some of those cycles correlate with cosmic patterns. The Chaos Mathematician is the one doing real nonlinear analysis. The Astrologer is the one getting us on the front page of Reddit.

Q: Can this actually predict the future? A: No tool can predict the future with certainty. BlackSwanX identifies where predictions are most likely to be wrong — which is more valuable than a prediction itself. The Antifragile Play tells you how to position regardless of which scenario wins.

Q: Why 174 agents instead of 5? A: A prediction from 5 agents has 5 blind spots. A prediction that survives scrutiny from 174 domain experts, 200 diverse citizens, and a dedicated Assassin trying to kill it — that's closer to truth.

Q: How is this free? A: Everything runs on Ollama (local inference). No cloud API calls. No subscriptions. Your electricity bill is your only cost.

Credits & Acknowledgments

• agency-agents by @msitarzewski — 156 expert agent definitions (we added 18 more custom agents on top) that form our Elite Panel. Licensed under their original terms.

• Ollama — Local LLM inference that makes zero-cost AI possible.

• OASIS by CAMEL-AI — Inspiration for swarm simulation architecture.

• MiroFish & BettaFish — Projects that inspired BlackSwanX to exist. We built on their vision and took it in an adversarial direction.

• SONA architecture inspired by self-play reinforcement learning and Nassim Taleb's Antifragile philosophy.

Star History

If BlackSwanX helped you see something the crowd missed, star the repo. It helps others find it.

BlackSwanX — 174 experts + 200 citizens, zero cost, adversarial truth.

Where the crowd is wrong, the alpha lives.