The Check-In
I walked away from Tauntaun for a week. Didn't look at it. Didn't touch it. The pipeline ran every 30 minutes like clockwork, scanning nine signal sources, fusing them into trade decisions, executing on Alpaca paper. Meanwhile the Shadow Book we built last week quietly logged 357 entries tracking where predictive and reactive signals disagreed.
Came back today and checked the numbers.
$280 on a $100K account. That's... not exciting. But then I looked closer.
The Real Number
The open positions are up $1,504 gross. Thirteen winners, seven losers. The winners are running:
| Symbol | P&L | Return |
|---|---|---|
| URA (uranium) | +$553 | +14.8% |
| QQQ (Nasdaq) | +$79 | +14.0% |
| IWM (small caps) | +$93 | +12.7% |
| HACK (cyber) | +$299 | +12.3% |
| XAR (aerospace) | +$124 | +7.2% |
Five positions returning 7-15% in three weeks. The system is picking winners. So why is the account only up 0.28%?
Because 62% of the money was sitting in cash doing absolutely nothing.
The Cage I Built
When we launched Tauntaun on Day 1, I set conservative guardrails. No single position over $3,000. Maximum 20 positions. Never invest more than 50% of the portfolio. These made sense at the time. We had zero data, zero track record, and I wanted to watch the system make mistakes before giving it real room to operate.
But here's what those guardrails actually did: with 20 positions maxed at $3K each and a 50% cap, the system physically could not deploy more than $38K of a $100K account. It was making double-digit returns on the money it was allowed to touch, and I had it handcuffed.
The realized losses from earlier stop-loss exits ($1,200 in closed trades) ate into the gross gains. That's fine. Stops did their job. But the drag from idle cash turned what should have been a noticeable gain into spare change.
The whole point of paper trading is to collect information about how the system performs under realistic conditions. Constraining it to 38% deployment doesn't tell you what happens at 70%. You're paying the cost of running the experiment without getting the data you need. Save the caution for real money.
What We Cranked Up
| Parameter | Old | New |
|---|---|---|
| Max position size | $3,000 | $5,000 |
| Max positions | 20 | 25 |
| Max invested | 50% | 70% |
| Sector concentration cap | 30% | 35% |
| Per-position cap | 5% | 7% |
That unlocks roughly $31,000 in new deployable capital. The system can now put up to $70K to work instead of $38K. Positions can be bigger when conviction is high. And there's room for 5 more concurrent trades.
The profit ratchet and trailing stops are untouched. Risk management stays tight. We're not removing the brakes. We're pressing the gas.
What's Coming Next
The parameter bump is the quick win. But there are two bigger changes in the pipeline.
The Barometer reweight. For the past week, the Shadow Book has been tracking where our predictive sources (Kalshi, ORALE, the Professor) disagree with reactive sources (RSS, GDELT, GPR). We have 357 data points now. The pattern from Entry #27 holds: predictive sources are producing better trades. The next step is giving them a weighted premium in the fusion engine so their signals carry more authority. That's Phase 2 of the Barometer plan, and the data is ready.
New predictive sources. Options flow and SEC insider filings are next in line. When a CEO quietly buys $5M of their own stock, that's a signal that arrives days before any headline. When unusual put volume spikes 3x on a ticker, something is about to move. These are the kinds of signals that the current system can't see.
The Real Test
URA at +14.8% and HACK at +12.3% are great. But they're great on small positions. The question that matters is: do those returns hold up when the positions are bigger and there are more of them? Does the system scale, or does it break?
We're about to find out.
The system was right about the trades. I was wrong about the size. Fixed it.