VARRD — Trading Edge Discovery

VARRD — Trading Edge Discovery

Institutional-grade quant research. Describe any trading idea in plain English, get statistically validated results with exact trade levels.

Any AI can backtest a strategy. VARRD guarantees it was done right — with K-tracking, Bonferroni correction, OOS lock, lookahead detection, and 4 other integrity guardrails enforced at infrastructure level.

MCP Server — 7 Tools, 4 Prompts

Endpoint: https://app.varrd.com/mcpTransport: Streamable HTTP (2025-03-26 spec)Auth: Anonymous access with auto-provisioned credits. No API key required.

{
  "mcpServers": {
    "varrd": {
      "transport": {
        "type": "streamable-http",
        "url": "https://app.varrd.com/mcp"
      }
    }
  }
}

Works with Claude Desktop, Claude Code, Cursor, and any MCP-compatible client.

MCP Tools

MCP Prompts

Quick Start — CLI

pip install varrd

# Research an idea (auto-follows the full workflow)
varrd research "When wheat drops 3 days in a row, is there a snap-back?"

# What's firing right now?
varrd scan --only-firing

# Search your saved strategies
varrd search "momentum on grains"

# Let VARRD discover edges autonomously
varrd discover "mean reversion on futures"

# Check credits
varrd balance

Quick Start — Python

from varrd import VARRD

v = VARRD()  # auto-creates free account

# What's firing right now?
signals = v.scan(only_firing=True)
for s in signals.results:
    print(f"{s.name}: {s.direction} {s.market} @ ${s.entry_price}")

# Research a trading idea
r = v.research("When RSI drops below 25 on ES, is there a bounce?")
r = v.research("test it", session_id=r.session_id)
print(r.context.has_edge)       # True / False
print(r.context.edge_verdict)   # "STRONG EDGE" / "NO EDGE" / etc.

# Get the trade setup
r = v.research("show me the trade setup", session_id=r.session_id)

Authentication & Passkey

First use auto-creates an account. You'll receive a passkey (VARRD-XXXXXXXXXXXXXXXX) saved to ~/.varrd/credentials.

  VARRD account created.
  Your passkey: VARRD-A3X9K2B7T4M8P1Q6
  Saved to: ~/.varrd/credentials

Link to browser: Go to app.varrd.com, click "Link your AI agent", enter your passkey with an email and password. Your agent's strategies and credits merge into your account.

v = VARRD(api_key="your-key")         # Python

varrd --key your-key scan              # CLI
export VARRD_API_KEY=your-key          # Environment variable

What You Get Back

Edge Found

STRONG EDGE: Statistically significant vs both zero and market baseline.
The pattern produces real returns that beat market drift.

  Direction: LONG
  Win Rate:  62%
  Sharpe:    1.45
  Trades:    247
  K:         3 (tests run on this hypothesis)

  Trade Setup:
    Entry:       $5,150.25
    Stop Loss:   $5,122.00
    Take Profit: $5,192.50
    Risk/Reward: 1.5:1

No Edge

NO EDGE: Neither test passed — no tradeable signal found.

This is a valid result. You found out for 25 cents
instead of $25,000 in live losses.

The Research Flow

Your idea (plain English)
    |
    v
  Chart pattern — see the actual signals on real price data
    |
    v
  You approve — sanity check before spending statistical power
    |
    v
  Statistical test — event study or backtest with proper controls
    |                 (K increments, fingerprints logged)
    v
  Edge verdict:  STRONG EDGE — beats zero AND beats market
                 MARGINAL    — beats zero, doesn't clearly beat market
                 NO EDGE     — no signal found
    |
    v
  Trade setup — exact dollar entry, stop-loss, take-profit

A typical session is 3-5 turns and costs ~$0.25.

8 Statistical Guardrails (Infrastructure-Enforced)

Data Coverage

15,000+ instruments total.

Pricing

• $2 free on signup — enough for 6-8 full research sessions
• Research: ~$0.20-0.30 per complete workflow
• ELROND council (8 expert investigators): ~$0.40-0.60
• Multi-market (3+ markets): up to ~$1
• Free tools: scan, search, get_hypothesis, check_balance, reset_session
• Credit packs: $5 / $20 / $50 via Stripe
• Credits never expire

Examples

See examples/:

• quick_start.py — 5 lines to get started
• scan_portfolio.py — Scan all strategies, show what's firing
• research_idea.py — Full multi-turn research workflow
• multi_idea_loop.py — Test many ideas in a loop
• mcp_config.json — MCP config for Claude Desktop

For AI Agents

See AGENTS.md for a structured guide with complete tool reference, response formats, and integration patterns.

Links

• Web app: app.varrd.com
• Website: varrd.com
• MCP endpoint: https://app.varrd.com/mcp
• MCP Registry: registry.modelcontextprotocol.io