AI Options Trading: How I Use Claude to Run My Options Book

AI options trading without a bot: three Claude agents that read the news, check my book and test trade structures every morning. AI prepares, I decide.

Share
How I Use AI to Run My Options Book cover

AI options trading does not mean letting a bot place trades. In my book it means three AI agents that read the news, check every position and test trade structures before I decide anything. Every morning they do about three hours of my work before I finish my coffee. This post shows how each one works, where it fails, and the one rule I never break.

Watch: the three agents on my real positions (9 min)

Disclosure: the positions shown were in my book when I recorded the video in October 2026. They are examples of the process, not recommendations.

Key takeaways

  • Three agents cover the daily work: a market digest, a morning book run and a trade structuring check.
  • Each agent is a skill: written instructions Claude follows the same way every time, connected to my own data.
  • The rule: AI prepares, I decide. It never places a trade and never sizes a position.
  • Every number that drives a trade gets checked at the source.

How the setup works

I run a long/short book built with options. I am not a programmer. Everything here I built by talking to Claude.

My positions, my Excel dashboard and my research notes are connected to it. Each job is a skill: a set of written instructions Claude follows the same way every time. What makes the output trustworthy is not the model. It is the dozen data connectors I added to each skill, so the agent works from real data instead of memory.

Agent 1: Market digest

I run it every morning before the open. It has five sections:

  1. Macro. Data released this morning, and what is due the rest of the week.
  2. Earnings. Who reported last night and this morning: revenue and EPS against consensus, guidance, and how each stock reacted. Plus who reports next.
  3. S&P 500 drivers. Which stocks actually move the index. Most people see only the mega caps and miss the rotation underneath.
  4. Top movers. The industries and stocks leading and lagging.
  5. Sentiment. The headlines moving the market, and the news that touches my own positions.

This used to take me about four hours. Now it is a few minutes of reading.

Agent 2: Morning book run

The second agent checks the book itself. It calculates the Greeks for the whole book and for each idea, shows how every idea is performing, and asks the questions I would ask: keep it, close it, roll it, or swap it for something better. Vega matters most around earnings, as I show in IV Crush: Why My Puts Lost Money When the Stock Fell 13%.

It also runs three scenarios, bull, bear and sideways, and shows my P&L in each. Then it builds a book snapshot in Excel: every holding at a glance, with cost basis, unrealized P&L, exposure by sector and theme, and a scenario sensitivity table.

Example from the video: Microsoft, long the $500 call and short the $550 call. $550 is my target price, so the short strike sits right on it. That vertical spread was up 78% when I recorded. I explain when a vertical fits in Vertical vs Calendar Spreads.

Agent 3: Trade structuring

The third agent runs after the research is done, when I am ready to turn a thesis into a trade. I give it three things: the direction, my target price and the expiry I want.

Then it works in steps:

  1. Tests the target. It checks how realistic the target is against how far the stock has actually moved in the past, based on realized volatility.
  2. Prices the structures I use. A long call or put, a vertical spread and a diagonal spread, each with its debit and the number of contracts that fit my position size of $10,000.
  3. Calculates P&L in a base case, a best case and an expected value case, and what each structure makes if the stock reaches the target.
  4. Marks a pass or fail for each structure against my rules.
  5. Asks for the catalyst, so the expiry matches the event that should move the stock.

In the Pinterest example it passed three structures: two call verticals and one diagonal. I have that trade in my book now. You can run a simple version of the same check on your own trade with the free options profit calculator. It shows your value at the target and tells you if the target is realistic. How I match a diagonal to a catalyst: How to Match an Options Structure to a Catalyst.

The one rule: AI prepares, I decide

AI gets things wrong, and quite often. It misreads a number. It finds a contradiction that is not there. A data feed goes stale.

So every number that drives a trade, I check at the source. The agents never place a trade and never size a position. They help me make better decisions faster. The decision stays mine.

How to start your own

  1. Pick one task you repeat every day. A morning news scan or a position check is a good first one.
  2. Write it as instructions. What to read, what to calculate, what the output looks like. That is your first skill.
  3. Connect your own data. Your positions and your data sources. Without them the output is a guess.
  4. Check what drives a trade. Use the output to prepare. Verify the numbers before you act.

What is next

This was part one. Part two covers the rest: the research agent that catches management contradicting itself on earnings calls, how I check if a move is stretched before I chase it, and how much risk sits quietly in what expires in my book. Subscribe on YouTube to catch it.

The process these agents support is the Method, and every month of results is on the Track Record page, losing months included.

FAQ

Can AI trade options for me?

It can, but I do not let it. AI is good at reading, calculating and summarizing. It still makes mistakes with numbers and context, so I use it to prepare and I make every trade decision myself.

Do I need to code to build AI trading agents?

No. I built mine by describing each task to Claude in plain language. The work is in writing clear instructions and connecting the right data.

Which AI do you use for options trading?

Claude, with skills for each job and connectors to my positions, spreadsheets and market data.

Educational content only. Not investment advice. Options involve risk and are not suitable for everyone. Past performance does not predict future results. See the Disclaimer.