Claude Code trading bots: 10 months looking for an edge
Polymarket, Binance and Interactive Brokers: what Claude Code built, the seven ideas I tested, and why none of them became an edge. Real money lost: about $54.
The 11 slides
Click the slides or use the arrow keys. Open full screen.
What is on each slide
- I went looking for an edge
Claude Code trading bots · 10 months Polymarket, Binance, Interactive Brokers. Real money lost: about $54 . Found an edge: not really . - Building the bot was the easy part
What Claude Code built Polymarket An insider detector and an AI agent that trades Binance A key that can trade, never withdraw Interactive Brokers A connection that logs in by itself + Around it Telegram alerts, backtests, a news bot every hour - "We spotted the Maduro attack before it happened"
Where it started The dream Insiders bet on Polymarket, and every bet is public. Spot them, and bet along. Why I believed it It worked for them: an insider turned $35k into $442k, hours before the raid. Source: @DidiTrading on X, 3 January 2026 - Paris: someone tampered with a thermometer
Polymarket · insiders The dream My bot watches every Polymarket bet and flags the insiders. Why I didn't win April 2026: someone tampered with the temperature sensor at Paris airport and bet $65 on the exact temperature. It paid $27,500. My bot missed it: it only looked at big bets. Source: Financial Times, 22 April 2026: French weather service alerts police after suspicious Polymarket bets · Bloomberg, 23 April 2026 - Follow the wallets that know
Polymarket · insiders The dream A newsletter found wallets that bet big only on companies audited by KPMG, and keep winning. Follow them. Why I didn't win My bot never looked at those wallets. The signal was real: they kept winning. I just didn't catch it. Source: Matt Lamers, EventWaves, 24 Feb 2026 (the author alleges no wrongdoing) · my own check of the wallets - Buy oil the moment an attack hits the news
Oil The dream Hear about an attack on oil supply first, and buy before the price jumps. Why it didn't work It did in 2026: oil went from $61 to $118 and stayed high for months. But that happens once every year or two, and on day one you can't tell if it's the real one. Source: Brent crude prices 2025–2026 · my own study of oil spikes since 2015 - Buy Bitcoin when the US prints money
Bitcoin The dream When the government pumps money in, Bitcoin goes up. Buy as soon as it happens. Why it didn't work It did, twice: when the Fed rescued the banks in 2023 and when the US bought back its bonds in 2026. Twice in three years, and on the day itself you have to guess the cause. Source: idea after Reuters, 25 August 2026 · my own backtest, Binance 2023–2026 - Buy after great results
Stocks The dream When a big tech stock jumps after great results, it keeps rising for days. Buy at the close, and ride it. Why it didn't work It only worked in the AI hype years. Sell the same day, the next day or ten days later: none of it held up in other years, or live. Source: idea after De Tijd, 27 August 2026 (Nvidia +$350 billion in a day) · my own backtest and live alerts - Copy the politicians
Stocks The dream Members of Congress trade on what they know. Their trades are public. Copy them. Why it didn't work They report their trades weeks later. In my test, buying on the day you find out no longer beat the market. Source: congressional trade disclosures via Capitol Trades · my own backtest - "Will Trump insult someone today?"
Polymarket The dream He does it almost every day. Bet yes, every day, and collect. The catch A yes already costs about 90¢. You win 10¢ a day, and lose 90¢ on the day he doesn't. That leaves a few cents a day. Source: @CarOnPolymarket on X ("infinite money glitch"), April 2026 - The bot is easy. The edge is not.
Found a bot that really makes money? Tell me in the comments.
Related
Want this set up with your team in one day? Training: Vibe-code together with GitHub and Claude, $1,490.