Video Summary
Claude acts as the central brain between TradingView (signals) and your exchange (orders).
You provide API key + secret + passphrase so Claude can place trades (example uses BitGet).
Strategy rules live in rules.json (VWAP, EMA, RSI, etc.); a safety layer blocks any trade if any condition fails.
Deployment to Railway makes the bot run 24/7; trade activity is logged to a spreadsheet formatted for accounting.
Everything is driven by a one‑shot prompt/onboarding agent in Claude Code — minimal coding required; start in paper trading mode.
Claude is the central brain that reads TradingView signals, holds strategy context (rules.json), runs a safety filter, and—when all conditions pass—sends orders to your exchange; it also logs successes and failures for accounting.
You must create an API key on your exchange and provide three pieces of information to Claude: the API key, the secret key, and any required passphrase.
The safety layer uses the strategy definition (rules.json) — e.g., VWAP, EMA trend, RSI thresholds — and only allows execution if every single criterion matches; any failed check blocks the trade and is logged with the reason.
You deploy the system to a cloud host like Railway; the video demonstrates setting up a cron schedule for periodic checks (example: every 4 hours) so the agent runs 24/7 and records activity to a spreadsheet.
No — the presenter recommends starting in paper trading mode, backtesting and observing results for days before switching to live funds; the demo uses paper trading and warns the example strategy is for tutorial purposes only.
"What you're looking at on screen right now is Claude reading a live TradingView chart, checking a strategy, running a safety filter, and placing real trades on an exchange."
Claude can now not only visualize strategies and backtest them, but it can also execute trades autonomously on your behalf.
By the end of the video, viewers will learn how to connect Claude to their exchange, allowing it to perform real trades based on pre-defined strategies.
The setup involves creating a comprehensive safety layer that adheres to the strategy conditions, ensuring trades occur only when all criteria are satisfied.
"You're going to walk away with a system that logs for accounting purposes every single trade and every transaction."
Claude will run continuously in the cloud, which means users do not need to keep their laptops powered on.
It is designed to log every executed trade and also those that did not meet the conditions for trade execution.
Users will have access to this logged information effortlessly, making accounting and tracking much simpler for tax purposes.
"There are only three tools involved in all of this and they each do one thing."
The primary tools needed are TradingView for market data, Claude for executing strategies, and the exchange where the trades will occur.
Connecting Claude to the exchange can seem intimidating, but it can be done simply by extracting the API key, secret key, and a passphrase, forming a secure connection for trade execution.
Rather than directly connecting TradingView to the exchange, Claude acts as an intermediary, monitoring signal activity on the TradingView chart and executing trades accordingly.
"All we need to do is give Claude the permission to talk to the exchange."
Users will download the app for the selected exchange and create a new API key with read and write permissions, which is essential for Claude to interact with the exchange.
It's recommended to keep the API key, secret key, and passphrase secure, as these pieces of information grant access to execute trades.
The video details specific steps to set this up using BitGet, an exchange that provides a bonus when funds are deposited, although similar steps can apply to other exchanges.
"Now that we've got AI, specifically with this onboarding agent, we don't need any of that complexity."
The process of working with AI has been made significantly simpler, eliminating the need for complicated configurations that used to be a barrier for users.
The introduction of the onboarding agent allows users to interact with the system by typing simple commands, reducing the cognitive load associated with complex settings.
The onboarding agent not only executes complicated tasks in the background but also engages with users, guiding them through the necessary steps, such as opening files and websites to streamline the experience.
"We're going to click exactly where we need to click on here. No extra steps."
The video instructs viewers to navigate to GitHub to access the necessary files for setting up the AI agent.
Despite initial appearances, the GitHub interface is manageable, and users are encouraged to follow straightforward instructions for a successful setup.
A prerequisite video is available for viewers to watch, providing essential context and information about setting up the bot and strategy.
"You're not having to do this individually; it's just going to take you through the process."
The onboarding process consists of a sequence of steps that the AI navigates automatically, ensuring a smooth user experience.
Users simply need to input commands as prompted, allowing them to focus on the key aspects of configuration without getting bogged down in technical details.
The structured approach is not only efficient but also user-friendly, making it accessible to individuals with varying levels of technical expertise.
"Now, let's get your API key. Here's the exact process for doing that."
As part of the onboarding, users are prompted to choose their exchange and obtain their API key, a crucial step for linking their account with the AI agent.
The system actively assists users by opening the exchange platform for them, which simplifies the process of account creation or login.
Following successful account integration, users are guided through securing their API key, secret key, and other necessary credentials, ensuring they understand each step.
"The onboarding process opens the ENV file for you, so you don’t have to go and find this confusing file."
Managing environment files, which store sensitive information like API keys, is typically a complex task for users.
The AI onboarding agent takes on the burden by automatically opening the ENV file, allowing users to input key data without navigating through file directories.
Any changes to the configuration can be made directly within the environment file through simple dialogue with the AI, enhancing usability and reducing frustration.
"If you haven't done the connection to Trade View and their MPC, you want to go to the previous video to set that up."
To successfully connect to TradeView, viewers must first ensure they set up the connection in a prior video, as this is essential for the process.
Upon establishing the connection, a health check is performed to confirm that everything is functioning correctly.
Users are prompted to build or select a trading strategy. There is an option to either create a new strategy from scratch or use an existing one.
The speaker mentions they have created a new scalping strategy, emphasizing the importance of communicating naturally with the AI during this interaction.
"The rules.json file is essentially a whole text document of the entire strategy."
The details of the trading strategy are contained in a "rules.json" file, which includes all necessary conditions for executing trades.
The strategy utilizes several indicators, such as VWAP, EMA, and RSI, to determine the optimal points to buy and sell.
To execute a buy, three criteria must be met: the price must be above the VWAP, above the 8 EMA in an uptrend, and the RSI must drop below 30.
The speaker warns viewers that the example strategy is not reliable and advises them to develop their own or source a better strategy from elsewhere.
"Railway is going to allow us to have this on the cloud, so it's running 24/7."
The next stage involves deploying the trading strategy to "Railway," ensuring the bot operates continuously regardless of whether the user's computer is on or off.
After logging into Railway, users enter specific commands to facilitate this process.
The frequency of trade checks can be customized; the speaker suggests checking for trades every four hours but acknowledges the impact of high-frequency checks on costs.
There is a mention of a "cron job," which is a scheduled task that executes at specified intervals.
"If any single one fails, do not trade."
The trading bot conducts checks before executing trades based on the defined strategy. If any condition is not satisfied, it will refrain from making a trade.
The speaker demonstrates the bot's functionality by highlighting a situation where the trade does not execute due to the RSI value being above the ideal threshold for buying.
A trade log is maintained for record-keeping, which is beneficial for tax purposes and tracking overall performance.
If a trade fails, detailed information, including the reasons for failure and insights about market conditions, is documented in the trades log.
"The bot is completely live now."
Despite the bot being live and operational, the speaker advises caution regarding reliance on example strategies without personal validation or development.
Throughout the video, viewers learn about setting up, deploying, and monitoring an automated trading strategy using advanced tools and programming.
The process not only aims to automate trading decisions but also encourages users to engage critically with the strategies they employ.
"Before we do a full demo of this to show you it working, I want to make sure that you hear what I'm about to say. Nothing in here is financial advice."
The speaker introduces a trading agent capable of executing trades every four hours, emphasizing that it is currently in paper trading mode. Users can switch to real money trading by simply instructing the agent to do so.
The speaker stresses the importance of using one's own trading strategy with the agent, stating that the effectiveness of the trades will depend on the user's strategy. A well-constructed strategy will yield positive results, while a poor strategy will lead to losses.
It is recommended to start in paper trading mode to gain experience and confidence. Users should backtest their strategies over several days before transitioning to real money trading. Once they are ready, they can prompt the trading agent to begin real trade execution.
"This is actively trading for me in the background... and all of it is being saved directly to a spreadsheet."
The trading agent displays real-time trading capabilities by assessing market conditions and executing trades that meet predefined criteria. Successful trades are marked with green check marks, while unsuccessful ones are blocked.
The agent operates continuously, trading around the clock within user-defined limits, and logs all activities in a spreadsheet for transparency and performance tracking.
Users are advised to watch the tutorial on a computer for a complete experience, as mobile viewing may limit their ability to replicate the demonstrated setup successfully.
"I'm going to teach you how to scrape the transcripts of YouTube channels that you kind of listen to... you think that maybe their strategy is going to be superior to yours."
The speaker outlines a method to scrape transcripts from YouTube channels, which can help users analyze other traders' strategies. A link to the relevant tool is provided in the video description.
Instructions are given to navigate to the Appify platform, create an API token, and integrate it with the user's trading agent. This enables the agent to access and scrape video transcripts efficiently.
Once the user identifies a YouTube channel of interest, they can instruct the agent to fetch transcripts for the latest videos and synthesize this information into a document. This insights-based approach allows users to refine their own trading strategies based on the content from successful traders.
