I tried to automate a desk.
I am starting over simpler.
I built a complex LLM options bot, lost −15.8% on paper, and kept the same book. Now I buy quality ETFs on red days and hold. Four weeks in I am +1.2% — and still behind a same-dollar S&P hold. I hope to evolve better rules — and learn the LLM piece again as I iterate.
Educational only — not financial advice.
01
I tried a desk
I built an LLM options stack that looked clever on paper.
02
I ended −15.8%
$100k to $84.2k paper. Receipts beat vibes.
03
I start here
Red-day ETFs, hold, no model in the loop.
04
I checked the scoreboard
+1.2% paper in four weeks. S&P hold +3.0%. Still the bar.
05
I hope to learn
Evolve rules on paper fitness — then earn LLM back.
What I measure
- Paper equity from the ~$84k v2 line
- Same-dollar S&P hold (day-one SPY)
- Drawdown from peak
- Buys vs the one-per-sleeve rule
- Sleeve weights (50 / 25 / 25)
- Journal receipts every scan
Where I hope to get
A mutator that beats this boring baseline on paper — then, as I iterate, I want to understand when an LLM belongs in the loop again, instead of assuming it from day one.