Video summary

I Let Claude AI Control TradingView — Claude Scanned 238 Trades in Seconds! (Game Changer)

Main summary

Key takeaways

Technology

Overview (what the video is about)

The creator (Gaurav) demonstrates how to connect TradingView (desktop) with the Claude/Cloud AI tool using an MCP (Model Context Protocol) setup. The goal is to let Claude/Cloud control TradingView, run automated chart scanning, and accelerate strategy research/backtesting—especially tailored to the Indian market (Nifty) rather than US-focused examples.

Key guide/tutorial: connecting TradingView desktop to Claude/Cloud

Prerequisites (must install)

  1. TradingView Desktop
  2. Cloud/Claude desktop app
  3. Node.js
  4. Git

Setup method (high-level)

  • Copy-paste an end-to-end command/prompt into Cloud to:
    • Clone a GitHub repo that sets up a TradingView MCP
    • Run npm install
    • Replace with the user’s system username
    • Apply rules/config needed for the integration

Common installation/runtime issues mentioned

  • Error like: “MCP server still isn’t loaded / session wasn’t fully restarted”
    • Fix: allow/enable debugging when prompted; rerun after restart
  • Another issue occurs if TradingView is already open
    • Fix: close TradingView first, then let Cloud launch/control it

Connection verification / health check

  • Cloud captures the currently opened TradingView chart symbol + timeframe (example shown: “Sandisk”).
  • It confirms access to:
    • chart/symbol/timeframe
    • API/CDP availability

Dashboard + automation concept

The creator built a dashboard (link provided in the video description) to document:

  • Installation steps
  • Commands to run
  • Troubleshooting/error fixes
  • Prompts for different trading workflows

Capabilities after connection (what Cloud can control in TradingView)

The video claims Cloud can:

Chart control

  • Switch symbol
  • Scroll the chart
  • Set the visible date range
  • Change chart type (e.g., candles → Heikin-Ashi → bars)

Scanner/strategy automation

  • Read trade lists / evaluate strategies
  • Fetch statistics like win rate, strike rate, etc.

Live data + alert workflow

  • Connect to live charts/data
  • Create an alert watch list

Use Case 1: Watchlist scanning + ranking (“A+ / B+ / B-” style)

Problem statement

If the user has a large watchlist (e.g., 15–20, later 29, potentially 100+), Cloud should quickly scan all symbols and return:

  • Top 3–4 picks (or ranked bucket categories)
  • Sector segregation / market breadth summary

Example: 10-point rule set (criteria)

An example rule set includes:

  • Price above 75 EMA, with EMA sloping up
  • Near highs (52-week or multi-month), close to a fraction of 50-bar high
  • Shallow/orderly pullbacks
  • Volume dry-up (example threshold: 5-bar avg volume < 70%)
  • Tight candle structure / narrow range (avoid ugly whipsaws)
  • Patterns like inside bar / narrow range / VCP, tight flags, supply exhaustion
  • RSI momentum between 50 and 75

Scanning behavior notes

  • Initially, scanning a whole watchlist batch could take time.
  • Improvement mentioned: Cloud uses per-symbol permission/step execution (scans by confirming each symbol).

Output example

Cloud produces a dashboard indicating which names qualify, e.g.:

  • Power Grid rated A+ (only some conditions met by each stock)

It also provides an overall market read based on watchlist performance.

Use Case 2: Indicator creation (score-based)

Cloud can generate an indicator/scanner that:

  • Scores a chart using the same 10 parameters
  • Shows how many conditions are met (e.g., “8/10”)

Example shown: Power Grid meets more conditions than others; the indicator reflects which items are missing/tight/volume-related.

Use Case 3: Strategy creation + automated refinement + backtesting results

The creator asks TradingView (via Cloud) to:

  1. Create a strategy (EMA/ATR-based concept)
  2. Validate it on a chart
  3. Run it across a long period (example: Apr 2015 → Apr 2026), then refine further

Reported metrics (example flow)

  • Initial version: poor long-term fit by the creator’s standards
  • Cloud iteratively refines/enhances the strategy by testing improvements
  • Improvements reported include higher returns and changes in drawdown/win rate

Final example outcome (for a refined strategy)

  • Around 238 trades
  • Max drawdown ~ 24–29% (varies by stage shown)
  • Win rate ~ 43%
  • Profit factor described as “very high”
  • CAGR shown around 28% over the tested period (example: last ~11 years)

Key takeaway claimed: end-to-end crunching (years of backtesting, trade stats, refinement) happens in minutes.

Claimed benefits vs manual workflow

  • 30x faster strategy iteration (creator’s claim)
  • Automated end-to-end analysis runs in under ~1–2 minutes once connected
  • Deep strategy debugging without code-writing:
    • Connect strategy + multiple indicators
    • Cloud tests different mechanisms and updates dashboards
  • Includes troubleshooting guidance for common integration issues

“Morning brief” automation (daily routine)

Cloud can run a scheduled/typed prompt like a “morning brief” to:

  • Scan Nifty 50
  • Produce:
    • NSE A+ watchlist
    • Nifty pulse / market health read (example note: pulse wasn’t correct until the Nifty chart was included in the watchlist)
    • Sector breakdown (example sectors: healthcare, capital goods, chemical, financial)
    • For each company: grades and checks (liquidity, inside bar yes/no, volume dry-up, etc.)
    • Top picks
    • Optional news summary (global + Indian)

Output can be sent to email / phone / text, per the creator’s plan.

Main speakers / sources

  • Speaker: Gaurav (channel host and demonstrator)
  • Referenced source: Louis Jackson (credited for the underlying method; Gaurav adapts it for the Indian/Nifty context)

Original video