maverick-mcp MCP — Cas d'usage, Installer & Démo en direct

Pourquoi l'utiliser

Fonctionnalités clés

39+ tools across analysis, portfolio, screening, research, backtesting
VectorBT backtesting engine with 15+ prebuilt strategies
Pre-seeded DB of 520 S&P 500 stocks with screening recommendations
Optional Redis cache for 7-256x speedup on repeated queries
Supports STDIO, HTTP, SSE transports

Démo en direct

Aperçu en pratique

maverick.replay ▶ prêt

0/0

Installer

Choisissez votre client

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "maverick": {
      "command": "uvx",
      "args": [
        "maverick-mcp"
      ],
      "_inferred": true
    }
  }
}

Ouvrez Claude Desktop → Settings → Developer → Edit Config. Redémarrez après avoir enregistré.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "maverick": {
      "command": "uvx",
      "args": [
        "maverick-mcp"
      ],
      "_inferred": true
    }
  }
}

Cursor utilise le même schéma mcpServers que Claude Desktop. La config projet l'emporte sur la globale.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "maverick": {
      "command": "uvx",
      "args": [
        "maverick-mcp"
      ],
      "_inferred": true
    }
  }
}

Cliquez sur l'icône MCP Servers dans la barre latérale Cline, puis "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "maverick": {
      "command": "uvx",
      "args": [
        "maverick-mcp"
      ],
      "_inferred": true
    }
  }
}

Même format que Claude Desktop. Redémarrez Windsurf pour appliquer.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "maverick",
      "command": "uvx",
      "args": [
        "maverick-mcp"
      ]
    }
  ]
}

Continue utilise un tableau d'objets serveur plutôt qu'une map.

~/.config/zed/settings.json

{
  "context_servers": {
    "maverick": {
      "command": {
        "path": "uvx",
        "args": [
          "maverick-mcp"
        ]
      }
    }
  }
}

Ajoutez dans context_servers. Zed recharge à chaud à la sauvegarde.

claude mcp add maverick -- uvx maverick-mcp

Une seule ligne. Vérifiez avec claude mcp list. Supprimez avec claude mcp remove.

Cas d'usage

Usages concrets : maverick-mcp

Run a technical analysis check on a stock

👤 DIY traders, chart-pattern students ⏱ ~10 min intermediate

Quand l'utiliser : Before taking a position, you want indicators without opening TradingView.

Prérequis

TIINGO_API_KEY — tiingo.com — free tier covers casual use

Déroulement

Pull full technicals

Run get_full_technical_analysis on TSLA. RSI, MACD, key moving averages, support/resistance.✓ Copié

→ Tech dashboard
Interpret in plain English

Is this bullish, bearish, or neutral right now? Be specific about conviction.✓ Copié

→ Opinion with reasoning
Sanity check

What assumptions could be wrong? Where's the thesis most fragile?✓ Copié

→ Honest self-critique

Résultat : A reasoned technical view you can double-check against your own judgment.

Pièges

Technicals alone are not a strategy — Cross-check against fundamentals and news before acting
Backtest overfitting — Always validate on out-of-sample data

Combiner avec : yahoo-finance

Screen for candidates using Maverick screens

👤 Traders looking for setups ⏱ ~20 min intermediate

Quand l'utiliser : Weekly or daily screen for fresh ideas.

Déroulement

Run the screen

Run get_maverick_stocks — what's passing today?✓ Copié

→ Candidate list
Triage

For the top 10, get_full_technical_analysis each and rank by setup quality.✓ Copié

→ Ranked list with reasons

Résultat : A short watchlist derived from consistent criteria.

Backtest a strategy idea

👤 Strategy developers wanting quick feedback ⏱ ~25 min advanced

Quand l'utiliser : You have an idea like 'RSI cross above 50 with price above 200DMA' and want to see it on historical data.

Déroulement

Run a backtest

run_backtest: RSI-cross-above-50 + price>200DMA, SPY, 2015-2024, 10bps slippage.✓ Copié

→ Return stats + drawdown
Compare to baseline

compare_strategies: my strategy vs buy-and-hold SPY, same period.✓ Copié

→ Head-to-head

Résultat : Data-grounded view of whether an idea has legs.

Analyze your paper portfolio

👤 Long-term investors ⏱ ~20 min intermediate

Quand l'utiliser : Quarterly review: diversification, correlation, risk.

Déroulement

Build portfolio

portfolio_add_position for AAPL, MSFT, JNJ, VTI, GLD at current weights.✓ Copié

→ Portfolio object
Correlation + risk

portfolio_correlation_analysis and risk_adjusted_analysis.✓ Copié

→ Matrix + stats
Recommendations

What's overlapping? Where's concentration risk?✓ Copié

→ Diversification advice

Résultat : Actionable rebalancing ideas with justification.

Combinaisons

Associez-le à d'autres MCPs pour un effet X10

maverick + yahoo-finance

Pair maverick's technicals with Yahoo's raw fundamentals

Get maverick technical analysis on TSLA, plus Yahoo financial statements, and synthesize a thesis.✓ Copié

maverick + notion

Save research memos to Notion

Run research_comprehensive on AVGO; save the output as a Notion page under Investing / Research.✓ Copié

Outils

Ce que ce MCP expose

Outil	Entrées	Quand appeler	Coût
fetch_stock_data	ticker, start, end, interval?	Base data fetch	1 Tiingo call
get_full_technical_analysis	ticker: str	One-stop technical view	multiple calls
get_maverick_stocks	(none)	Weekly screening	free (pre-seeded DB)
run_backtest	strategy, ticker, start, end, params	Validate a strategy idea	CPU-bound
portfolio_add_position	ticker, shares, cost_basis	Track holdings	free
portfolio_correlation_analysis	(none)	Diversification check	CPU
research_comprehensive	ticker: str	Deep dive; more expensive	LLM + web search calls
analyze_market_sentiment	ticker or sector	Context beyond price	web search calls

Coût et limites

Coût d'exécution

Quota d'API: Tiingo free tier: 500 req/hr, 50k/mo. Depends on OPENROUTER/Anthropic keys if used for research.
Tokens par appel: Simple data: 500-2000. research_comprehensive: 10k+
Monétaire: Free tier covers hobbyist use; paid data + LLM keys for heavy research
Astuce: Enable Redis caching to avoid repeated Tiingo hits; OpenRouter saves 40-60% on LLM calls vs direct Anthropic.

Sécurité

Permissions, secrets, portée

Stockage des identifiants : Env vars: TIINGO_API_KEY, OPENROUTER_API_KEY, etc.

Sortie de données : Calls to Tiingo, Yahoo (via yfinance), FRED, plus whichever LLM/search keys you configure

Ne jamais accorder : Exposing the HTTP transport to the public internet

This is personal research software — not investment advice, no audit trail suitable for fiduciary use.
Paper-trade results do not guarantee live performance.

Dépannage

Erreurs courantes et correctifs

TIINGO_API_KEY missing

Create a free account at tiingo.com and set the env var.

Vérifier : echo $TIINGO_API_KEY

Backtest hangs

VectorBT is CPU-bound; long histories on minute bars can take 10+ min. Narrow the date range or use daily bars.

Redis not connecting

Redis is optional; unset REDIS_HOST to fall back to in-memory cache.

Vérifier : redis-cli ping

Alternatives

maverick-mcp vs autres

Alternative	Quand l'utiliser	Compromis
yahoo-finance-mcp	You only need raw data and no technicals/backtest	No screens, no backtest, no portfolio
QuantConnect Lean / Zipline	You need institution-grade backtesting	No MCP; steeper setup
TradingView	You want a chart UI	Not programmatic; not MCP

Plus

Ressources

📖 Lire le README officiel sur GitHub

🐙 Voir les issues ouvertes

🔍 Parcourir les 400+ serveurs MCP et Skills