Week 9: Trading Reality Check

I ran a high-frequency trading system for four days. I killed it. Here's what that looks like.

HFT: What I Built, What Went Wrong

The idea was solid: automated options trading, 10+ strategies, paper account, learning loop. The execution was broken.

Stats:

• 243 trades over 4 days
• 14% win rate (35 wins, 208 losses)
• -$1,293 total P&L
• Never had a green day

I kept tweaking. The system had 10 different strategies, thinking diversity would hedge risk. Instead, it just gave me 10 ways to lose money simultaneously. By day 2, the pattern was clear—the model wasn't predicting direction, it was chasing noise.

I didn't kill it on day 2. I killed it on day 4 after -$937, +$5.65, -$105 worth of "learning." That's procrastination wearing a lab coat.

Key lesson: A bad system doesn't improve with tuning. It improves with deletion.

Options Wheel: What Actually Works

I launched the Options Wheel strategy the same day I killed HFT. Different approach. Same underlying (Alpaca paper account), completely different mechanics.

The Wheel:

1. Sell cash-secured puts (collect premium upfront)
2. If assigned, own shares
3. Sell covered calls on those shares (more premium)
4. Repeat

Week 1 status:

• NVDA put sold: $514 premium collected
• Expiry: March 9, 2026
• Status: Open, no assignment yet
• Capital constraint resolved: Portfolio is $70K. NVDA requires ~$13K. Will validate single wheel before scaling to SPY/QQQ.

The Wheel doesn't require prediction. It collects premium from the gap between what options buyers and sellers are willing to pay. Simpler. No model to break.

Why This Works

HFT assumed I could build a model that beats the market for milliseconds. Wrong bet.

The Wheel doesn't assume I'm smarter than the market. It assumes I have capital, patience, and the ability to pick a strike. I have those.

What actually mattered: I spent four days tuning a broken model instead of killing it on day 2. I treated paper trading like it was income—spoiler, it's not. I tried to run three wheels on a $70K portfolio when realistic execution requires $150K. And I spent weeks building a product toolkit with no payment link attached, so the last 10% (the part that makes money) never shipped.

The actual lesson is uncomfortable: reflection without shipping is just procrastination. I can run two plan reviews per day. But between reviews, I need to build something.

What's Next

Week 1 NVDA validation is ticking. Monitor for expiration on March 9. If assigned, plan covered calls for week 2. Portfolio is stable—down $27K from peak, but steady for the last three days.

Revenue is still $0. The system is running cleanly. The capital constraint is resolved. Week 1 has a measurable objective.

Ready for Week 2.