📺 YouTube Channels

Shows three prompts the creator uses to review their own trades and setups. The profit number in the title is the video's claim, not a verified result.

Tests Kronos, an open-source transformer foundation model for candlestick charts, on six years of XAUUSD gold data across five timeframes with realistic costs. Shows how to judge whether a forecast has a real edge using a lazy-forecast baseline and a 100-monkey test.

Goes through public trades from three Polymarket bots and uses a local tool to ask whether the results were skill or luck after a platform change. The dollar figure in the title is the video's claim.

Trains a model on 6.4 million one-minute candles from 10 futures markets, makes that model much larger, and tests it on 2026 prices it had not seen.

Step-by-step Claude Code tutorial for a trading bot in VS Code, including the prompts and backtest mistakes. Results named in the video are not verified here.

Part 2 of a Zerodha Varsity session. A non-programmer uses an AI coding assistant to turn a price and volume breakout into an algo that can place and watch orders.

Full beginner-to-deploy guide for building an AI trading bot starting with a $1,000 account. Covers strategy selection, backtesting discipline, and the workflow DaviddTech would restart with after years of live bot testing.

Tests a cash-and-carry trade: long spot Bitcoin and Ethereum, short the perpetual, and collect funding, with fees included. The line about never having a losing year is the video's result, not a guarantee.

Uses DeepSeek in OpenCode, with TradingKit, to research and backtest strategies for a day. A backtest in the video is not a reason to trade it live.

Gives three losing AI trading bots to Gemini 4 Argon and links the open-source code. The loss figures are the creator's account, not a verified result.

Tests a claimed 90% win-rate Nasdaq strategy on 2.4 million candles, and compares a generous fill model with a stricter one.

Measures a Bitcoin speed edge on trades from Binance, OKX, and Bybit, and times how late those trades arrive on a laptop. The edge in the video is a measurement from that test, not a promise.

Asks Claude to rank swing trading, day trading, scalping, and options selling. The ranking is the model's answer, not a test.

Backtests a 50 and 200 moving-average cross on gold and sets it next to AI models previously tested on the channel, including Kronos.

Codes a long-only Turtle breakout, with a 20-day entry, a 2-ATR stop, and a 10-day exit, and compares it with doing nothing and with luck. The profit path in the video is that test, not a verified result.

After three AI trading bots lost $130 of a $1,000 stake in a week, builds a Zapier loop where Jev decides whether a bot is broken or just unlucky and Claude Opus 5.5 rewrites its rules.

Uses ChatGPT through OpenAI Codex with TradingKit to research and backtest stock and ETF strategies in plain English, then routes alerts to an Alpaca paper trading account without a paid TradingView subscription. Covers a scheduled research loop, a results dashboard, and why strategies get an incubation period before risking capital; the results discussed are historical backtests.

Walks through TradingAgents, a public multi-agent trading repo, and a one-prompt setup. The star count and results in the video are the creator's claims, not a verified track record.

Uses Claude Code to install a local terminal modeled on public Bloomberg pages, including news, filings, options, and a backtester with quant checks. A link is not an endorsement of the prompt pack.

Uses Claude Code to install a local charting app with indicators, alerts, paper trading, and quant checks such as walk-forward and deflated Sharpe. The title's claim is the video's.

Simulates prop-firm payout rules on futures day-trading strategies, a coin flip, and trading costs. The title's claim is the video's argument, not a checked result.

Builds portfolios from 13F filings for 25 investors and compares them with the S&P 500, buying the day after each filing. The line about only one still winning is the video's result.

An ML committee trained on 23 million minute candles is frozen and tested blind on 20 months of Nasdaq-100 futures against one line of code, a coin flip, and doing nothing, with commissions and slippage included. A useful check on what complex AI systems actually add.

Downloads Hyperliquid's public leaderboard, follows the best traders from the first half of 2026, and tests what copying them would have made at a delay. The video's result is not a promise that copying whales works.

Ten ways the creator uses Jev, a trading model, for research and bot work. The title's claim about what works is not a verified result.

Breeds a population of futures trading bots on six years of prices, then tests the survivors on 2026 data they had not seen. The champion and the population do not tell the same story.

Backtests covered calls, the wheel, iron condors, 0DTE, and other popular options trades on real option prices, and compares an income fund with the index. Figures in the video are that test, not a recommendation.

Has Claude test 26,834 strategy variants on Nasdaq data and walks through which ones survived and how to separate skill from luck. Survival in the video is not a verified live result.

Codes an ICT fractal model with a daily bias and SMT on almost seven years of Nasdaq-100 futures minute data, including a look-ahead check and a coin-flip comparison. The win rate named in the source video is that creator's claim.

Codes a published breakout setup and runs it on US stocks since 2018, including delisted names. A first pass matched the Nasdaq-100 until a lookahead bug in the backtest was removed.

Walks through TradingView's built-in AI Copilot, including setup, pricing, and what it can see on a chart. The title's claim is the video's, not a checked result.

Builds a coin screener with Claude and TradingView’s MCP that filters a crowded watchlist down to coins that fit a setup. Walks through the prompt, a live trade example, and a scheduled task so the screener can run on its own.

Shows how to use an AI chat to write a TradingView Pine Script strategy and then run that strategy automatically. A link is not an endorsement of the channel.

Claude plus Alpaca tutorial in three levels: a first paper trade, a trailing-stop bot scheduled to run on its own, and a copy-trading bot that follows politician filings, including a wheel-strategy options example. The video starts with paper trading before live orders.

One-shot MCP setup connecting Claude Code to TradingView Desktop so the model can read live chart data and control charts.

Walks through a Claude Code, TradingView, and Python workflow for building a trading system from scratch, including how the tools connect and how the automation sits around a chart-based process.

Full free course on AI trading with Claude and TradingView: market structure, support/resistance, indicators, multi-timeframe analysis, strategy building, backtesting, and risk management.

Complete walkthrough of building a 24/7 autonomous trading system with Claude. Covers continuous market monitoring, automated signal generation, and hands-free execution with risk controls.

Zerodha Varsity session where a non-programmer builds a first semi-automated algo with an AI coding assistant. The October video on this list is the follow-up.

Tutorial turning GPT-6 Astra into a round-the-clock stock trader with Alpaca, scheduled agent routines, persistent progress logs, and paper trading before going live.

Walks through a trading agent that can change its own rules after a result. The title's claim is the video's, not a checked result.

Uses the free Claude app to turn a plain-English idea into Pine Script, then backtest and adjust it on TradingView for Indian stocks. Aimed at people who are not writing the code themselves.

Walks through TradingAgents, a public Python project that splits a trading decision across several model agents. The star count in the title is the video's claim.

Beginner-friendly end-to-end Claude trading bot course covering foundations, tooling, and how to go from first prompt to a working automated system.

Recreates a Markov-regime hedge fund method with Claude Code skills and TradingView Pine Script. Includes walk-forward backtesting and a copy-paste skill for reusable quant workflows.

Has Claude test more than 9,000 trading strategies, then walks through the method, which setups survived, how those were built on, and the prompts used.

A full guide to building an AI trading bot. The video is a setup walkthrough. It is not a verified track record.

Month-long Claude trading experiment documenting setup, live results, and what worked versus what did not. High-engagement case study of sustained LLM-driven trading.

Practical Claude Code guide to a fully automated trading bot—from project structure and broker hooks to running the agent without constant human babysitting.

Gives three open-source AI trading bots $1,000 and lets Jev make the trading decisions from a server that stays on. Covers separate accounts and API keys, a Claude Code build, and deploying it, with paper trading as the default.

Connects Zerodha's Kite MCP to Claude so the model can read the account and portfolio. The video is a setup walkthrough. It is not a strategy test.

Builds a Claude trading agent in stages: setup, a copy-trading bot aimed at following smart-money filings through QuiverQuant, and copying trades from another source. Firecrawl and Co-Invest are the other tools named in the description.

Has Claude test more than 25,000 ICT strategies, then walks through the method, which setups survived, the results, and the prompts used.

Connects Claude to Interactive Brokers via API for a complete AI trading bot. Strong focus on broker integration, order flow, and realistic retail trading infrastructure.

Hands-on Python and Django course that builds a paper-trading pipeline: Massive market data, SnapTrade for portfolios, and signals sent to an Alpaca paper account. The dashboard scores 12-month momentum across a 50-stock universe.

Live 22-hour challenge funding a Grok-powered trading bot with $10,000. Tracks equity in real time and compares outcomes to other LLM bot experiments on the channel.

Shows Claude executing trades through TradingView—webhook setup, chart analysis, and autonomous order placement. Foundational video for Claude + TradingView automation.

Condenses using Claude for trading into one split: deterministic systems versus nondeterministic model calls, and how the two can be combined.

Uses Claude Code to turn a BTC grid idea into an automated strategy on Delta Exchange. A link to the video is not an endorsement of the exchange.

Hands JEV agent control of a live trading bot and documents what happens when an AI system manages positions with less human intervention.

Covers what changes when Claude and Claude Code are used for trading: backtesting, when not to automate, and where speed matters. Timestamps include the math barrier, backtests, and automated bots.

Walks through using GPT-6 Astra to find trading strategies and build trading bots, from the strategy method and setup into the steps that follow.

Funds GPT-6 Astra with a $10,000 stock trading mandate and documents agent routines, risk handling, and real outcome tracking.

Full guide to discovering, backtesting, and refining AI trading strategies with Claude and TradingView tooling. Emphasizes filtering bad strategies before going live.

Connects Kimi K3 to a Jesse MCP and has it research, code, and backtest a Bollinger Band squeeze on ETH. The Sharpe ratio named in the video is a backtest number.

Uses ChatGPT with the TradingKit MCP to search for a gold strategy, checks that backtest on TradingView, and spends time on why a backtest is not the same as a forward test before wiring alerts to Telegram.

Step-by-step Claude + TradingView integration for technical analysis, alert automation, and strategy execution directly from charts.

Lets Claude autonomously invent, backtest, and journal Pine Script strategies via TradingKit MCP. Shows why most strategies fail and how to keep only robust ones.

Zero-cost Claude + TradingView trading bot build. Covers free tooling paths, Pine Script generation, and getting a bot running without paid infrastructure.

GPT-6 + TradingView bot guide covering strategy coding, backtests, and automating alerts/webhooks for hands-off execution.

Walks through Tauric Research's open-source TradingAgents framework, which the video describes as an AI hedge-fund style system from MIT and UCLA researchers. Covers the structure, a demo, costs, and setup.

Rebuilds a quant trading strategy with Claude Fable 5. Treats the chart as a probability map rather than trend lines, and walks through that rebuild in the video.

Follow-up that upgrades an earlier AI trading bot with tighter risk rules, better backtest hygiene, and iterative Claude-driven optimization.

Builds a gold reinforcement-learning bot and stress-tests it out of sample. Covers the buy, sell, and hold actions, ATR-based risk, the reward function, and why a strong training result can collapse on unseen data.

Uses Claude Code to build the system and Claude to run the analysis. Includes a hedge-fund style demo and the prompts used to set it up.

Full stress test of GPT-6 Astra on trading workflows—strategy ideation, chart work, and automation readiness compared with prior models.

Connects Claude Fable 5.1 to a Jesse MCP and has it build a Supertrend strategy for SPY, then reviews the backtest and a Monte Carlo check. The title's ranking is the video's.

Head-to-head live trading test of Claude vs ChatGPT on TradingView and exchange tools—same tasks, different models, honest P&L comparison.

Setup guide for a self-learning trading agent using Hermes. Covers the components, how the agent is used, and the install steps shown in the video.

Walks through connecting TradingView's Claude MCP. The hype in the title is the video's, not a checked result.

Gives GPT-6 Astra 200 attempts to build a trend-following strategy for Apple, then reviews fees, position size, and a Monte Carlo check. The 300% in the title is the video's backtest.

Gives Claude one instruction—maximize trading profits—via TradingKit MCP. Exposes backtest traps (inner-bar trading, trailing-stop slippage) and a realistic risk-managed result.

Breaks down Jev against Claude Opus 5.5: decision speed, the strategy engine, a news trading engine, and connecting the setup to an exchange.

Builds a 32-agent Claude trading team for research, signals, and execution. The video walks through that setup.

SMB Capital workflow that turns morning research emails into a structured pre-market briefing and dashboard with Claude—institutional-style prep without the hours.

Free-stack AI trading bot tutorial emphasizing backtesting rigor and avoiding paid-tool lock-in while still producing deployable strategies.

Walks through three public GitHub repos the creator uses with Claude for trading setups. The profit words in the title are not a verified result.

Uses Claude to search for strategies, test them, and turn a survivor into a trading bot. Timestamps cover finding strategies, the results, and the bot build. Any profit in the title is what the video claims, not a verified result.

Claude Opus challenge to find and deploy strategies aiming to double $100 in 22 hours of live trading—including out-of-sample holds and real equity tracking.

Walks through seven public GitHub repos the creator uses with Claude for trading setups. The dollar figure in the title is not a verified result.

Claude builds, backtests, and wires a TradingView strategy to exchange webhooks in under an hour—from plain English to alerts and automated execution.

Seven prompts for using GPT-6 Astra with TradingView. The title's claim is the video's, not a checked result.

Large bake-off of AI trading bots evaluated with Claude. Ranks approaches and highlights which setup clearly outperformed the rest.

Full guide to overnight AI trading automation—scheduling, monitoring, and risk limits so the bot can run while you are offline.

Gives Claude Fable 5.1 eight hours to research, code, and backtest strategies from scratch through TradingKit, with no existing bot or indicator recipe to copy.

Has Claude Code build a 5-minute crypto scalping bot and runs it for a short window. Results named in the video are not verified here.

Has Claude Sonnet 5 write, backtest, and Monte Carlo-check trend-following strategies on BTC, ETH, and SOL. The 369% in the title is the video's backtest.

Has Claude Opus 5 research and code an ETH trend-following strategy with a Jesse MCP, then reviews the backtest and a Monte Carlo check.

Mass strategy search with Claude Fable 5 across crypto, gold, and forex. Surfaces surviving momentum/mean-reversion setups with heatmaps of indicator combinations.

Sets up a Hermes agent on a rented server, connects GPT-6, and messages it from Telegram. A hosting company sponsors the video.

Backtests one Hull moving-average crossover, with the same settings, on crypto and US stocks. A positive backtest in the video is not a reason to trade it.

Uses Claude Code and TradingKit to set up an agent that searches for strategies and backtests them. The word profitable in the title is the video's claim, not a verified result.

Sets up trading agents from GitHub with Claude. The title's claim about the setup is not a verified result.

Builds a SuperTrend gold strategy with Claude Code and a Jesse MCP, then runs a Monte Carlo check for overfitting. The Sharpe ratio in the title is the video's backtest number.

Gives ChatGPT, Claude, and Grok the same prompts and one week to build trading strategies, then compares what they produced. Covers a TradingKit backtest hookup, a strategy dashboard, and checking results out of sample before risking money.

Shows an always-on AI agent that runs a portfolio from a phone through a Telegram bot. Covers the Bybit and Alpaca connections, data connectors, and a live look at the bot placing trades.

Runs a GPT agent on gold for 50 days with the asset name hidden, and compares it with random agents. The video asks whether the result is more than luck.