APEX

APEX : Autonomous Population of Evolving eXpertsEvolve Your Agents, Adapt Your Strategy and Dominate the Market - a self-evolving trading strategy system powered by Airia AI

Comment

APEX : Autonomous Population of Evolving eXperts

A self-evolving trading strategy system powered by Airia AI

Banner

🧬 What is APEX?

APEX is a real-time evolutionary AI system that manages a population of 40 autonomous trading strategy agents. Every generation cycle, Airia AI scores each agent, culls the weakest, breeds the strongest, and writes plain-English post-mortems — all in under 20 seconds.

Think of it as survival of the fittest, but for trading algorithms.

Product integrated : Airia.ai, Google DeepMind, Datadog

✨ Key Features

  • 🔁 Evolutionary Cycles — Click a button to trigger a full generation: score → cull → breed → mutate
  • 🧠 Airia-Powered Intelligence — Every evolution decision, post-mortem, and genome mutation is driven by live Airia API calls
  • 🌍 Environmental Awareness — Paste any market headline and watch the regime shift in real time
  • 🎛️ Behavioral Override — Override a trade recommendation and watch the agent genome update to reflect your style
  • 📊 Live Radar Charts — Behavioral genome visualized as an animated 5-axis radar
  • 📜 Evolution Log — A live-scrolling feed of every birth, death, and override event
  • 📈 Generation History — Fitness trends, lineage trees, and extinct agent post-mortems across all generations

🏗️ Architecture

┌────────────────────────────────────────────────────────┐
│                     APEX Frontend                      │
│   React 18 + TypeScript + TailwindCSS + shadcn/ui      │
│   Zustand (state) · Recharts (radar + line charts)     │
└─────────────────────┬──────────────────────────────────┘
                      │ API Routes (Next.js 14)
         ┌────────────┼────────────────┐
         ▼            ▼                ▼
   /run-generation  /override      /environment
         │            │                │
         └────────────┼────────────────┘
                      ▼
              ┌───────────────┐
              │   Airia API   │  ← The brain
              └───────────────┘
                      │
              ┌───────────────┐
              │  state.json   │  ← Local persistence
              └───────────────┘

🧩 Tech Stack

Layer Technology
Framework Next.js 14 (App Router)
Frontend React 18 + TypeScript
Styling TailwindCSS + shadcn/ui
Charts Recharts (RadarChart, LineChart)
State Zustand
AI Orchestration Airia API
Storage Local JSON (no DB needed)
Fonts Syne (headings) · Space Mono (data)

📁 Project Structure

apex/
├── app/
│   ├── page.tsx                  # Main dashboard
│   ├── history/page.tsx          # Generation history
│   └── api/
│       ├── run-generation/route.ts   # 🔑 Core evolution endpoint
│       ├── override/route.ts         # Behavioral genome update
│       ├── environment/route.ts      # Market regime classification
│       └── state/route.ts            # Full state read
├── components/
│   ├── AgentCard.tsx             # 40-card agent grid
│   ├── EvolutionLog.tsx          # Sliding log feed
│   ├── BehavioralRadar.tsx       # 5-axis genome radar
│   ├── EnvironmentPanel.tsx      # Headline → regime
│   ├── GenerationButton.tsx      # Hero CTA
│   └── OverrideModal.tsx         # Trade override dialog
├── data/
│   └── state.json                # Seed data (40 agents, gen 3)
├── lib/
│   ├── airia.ts                  # Airia API client
│   ├── store.ts                  # Zustand store
│   └── seed.ts                   # Seed data generator
└── types/
    └── index.ts                  # All TypeScript interfaces

🧬 Agent Anatomy

Each agent carries a genome that encodes its trading personality:

{
  "id": "agent_001",
  "name": "Momentum Alpha",
  "generation": 3,
  "status": "active",
  "fitness_score": 74.2,
  "genome": {
    "entry_logic": "RSI crossover + volume confirmation",
    "exit_logic": "trailing stop 2.5%",
    "risk_tolerance": 0.65,
    "position_sizing": "Kelly criterion",
    "indicator_weights": { "RSI": 0.4, "MACD": 0.3, "volume": 0.3 },
    "momentum_bias": 0.7,
    "drawdown_sensitivity": 0.4
  },
  "performance": {
    "sharpe_ratio": 1.42,
    "max_drawdown": -8.3,
    "win_rate": 0.61,
    "total_return": 14.2
  },
  "parent_ids": ["agent_019", "agent_007"],
  "born_generation": 2,
  "trade_count": 47
}

🤖 Airia Deep Integration

Airia is the cognitive engine behind APEX --- not just a text generator.

Evolutionary Intelligence

  • Scores agents using performance + behavioral alignment
  • Selects weakest strategies for extinction
  • Breeds new agents with synthesized genomes
  • Generates explainable post‑mortems

Behavioral Learning

User overrides update a behavioral genome:

Human decision → Airia reasoning → Genome mutation → Future evolution

Environmental Intelligence

Market headlines become structured signals:

{
  "regime": "risk-off",
  "volatility_level": "high",
  "sentiment_score": -0.6
}

Airia continuously shapes how the system adapts. Airia1 Airia2

APEX makes 3 types of live Airia calls:

1. POST /api/run-generation

Sends top 5 agents → Airia scores, culls 2, breeds 1 child with a new creative name and blended genome.

2. POST /api/override

User describes what they would have done differently → Airia updates the Behavioral Genome to reflect the new preference pattern.

3. POST /api/environment

Market headline pasted in → Airia classifies regime (risk-on / risk-off / choppy) and returns the full Environmental State Vector.

🚀 Getting Started

# 1. Clone the repository
git clone https://github.com/your-username/apex.git
cd apex

# 2. Install dependencies
npm install

# 3. Set up environment variables
cp .env.example .env.local

Add your credentials to .env.local:

AIRIA_API_KEY=your_airia_api_key_here
AIRIA_BASE_URL=https://api.airia.com/v1

# 4. Run the development server
npm run dev

# 5. Open http://localhost:3000

Built With

  • airia
  • airia.ai
  • datadog
  • google-deepmind
  • nextjs
  • recharts
  • supabase
Share this project:

Updates