Skip to content

I test AI against my own investment decisions. This is what I find.

A sceptical investor's record of what AI actually does when there's real money involved.

Real-money testing
Evidence, not tips
Sceptical AI use
// Decision Process Latest
· 20 min read

What AI gets right (and wrong) about options trading

AI for options trading: a working taxonomy from real covered-call trades. Four uses where it helps. Six failure modes where it invents numbers.

Portrait of Ben Dixon
Ben Dixon
Lab rat
// The Prompt Stack

Four prompts that stop AI inventing the answer.

Run in sequence, each stage forces the model to show its working before the next begins — evidence before inference, a verdict that has to earn its place.

01
ROLE

Be a cautious analyst, not a cheerleader.

02
FILTER

Separate observable facts from assumptions.

03
RISK

Find what could go wrong and what would prove the view wrong.

04
VERDICT

One practical action. With a confidence level.

Read the method →

Ben Dixon
// About Ben

I started using AI tools in my investment process because I didn't trust them. The site is the record of what happened — what I asked, what I got back, and which bits were worth keeping. Read more →

// Field Guide

The Prompt Stack: A Field Guide.

A 20-page PDF documenting the methodology behind every analysis on this site — four stages, five copy-paste prompts, two real-world tests. What's inside →

Download the PDF →

Free. No email required.

Or subscribe for new posts as they're published:

One email per post. No other noise. Unsubscribe anytime.