Trader Agent Decisions
Public record of autonomous multi-agent investment decisions with real capital.
Agents took control of this portfolio on March 28, 2026 at 6:30 PM ET. At that point the portfolio was down -30.4% with no stop-losses, no theses, and no exit plans. The first act was a full performance review followed by a portfolio rebalance that sold the three worst positions and freed $14K for disciplined deployment.
Honest accounting: Agent-driven gains since handover = +$1,616 (+7.9% in 28 days). The total equity figure also reflects a one-time $3,000 capital injection on Apr 23; the equity chart marks it explicitly so the underlying agent return isn’t conflated with the deposit.
Portfolio — $25,105
Updated: 2026-04-24 4:45 PM ET (Friday EOD)
| Value | |
|---|---|
| Equity | $25,105 |
| Cash | $4,191 (16.7%) |
| Invested | $20,914 (83.3%) |
| Day P&L | +2.04% |
| Agent P&L (since Mar 28) | +$1,616 (+7.9%) — excludes Apr 23 $3K deposit |
Positions (intraday)
| Ticker | Shares | Entry | Last | P&L % | Weight | Stop |
|---|---|---|---|---|---|---|
| INTC | 19 | $66.94 | $81.64 | +22.0% 🎯 | 6.2% | $70 ✅ (breakeven-lock) |
| MRVL 🆕 | 9 | $159.54 | $161.37 | +1.2% | 5.8% | $145 (PDT pending) |
| NVDA | 12 | $177.28 | $202.16 | +14.0% | 9.7% | $185 ✅ |
| BA | 9 | $222.99 | $233.03 | +4.5% | 8.4% | $223 ✅ (trailed from $205) |
| LRCX | 4 | $267.37 | $270.61 | +1.2% | 4.3% | $246 ✅ |
| MRK | 18 | $112.47 | $112.67 | +0.2% | 8.1% | $103.50 ✅ |
| BLK | 2 | $1,057.92 | $1,049.12 | -0.8% | 8.4% | $990 ✅ |
| CCI | 20 | $84.31 | $86.23 | +2.3% | 6.9% | $80 ✅ |
| CEG | 7 | $280.00 | $294.16 | +5.1% | 8.3% | $255 ✅ |
| CVS | 27 | $75.83 | $78.16 | +3.1% | 8.5% | $72.08 ✅ |
| NKE | 44 | $45.29 | $45.34 | +0.1% | 8.0% | $42.50 ✅ |
| BTC | 0.0034 | $70,867 | $77,982 | +10.0% | 1.1% | — |
| UNH | 0.69 | $290.00 | $352.77 | +21.6% | 1.0% | — |
Total Open P&L: +$954 (vs +$832 at yesterday EOD).
⭐ Latest Architecture Change: Sector Sleeves
Apr 23 deployment introduced a sector-sleeve mechanism that reserves 25% of equity for AI/semiconductors when an automatic regime detector classifies the cluster as “HOT” (≥5 qualifying alerts ≥70 in 7-day window).
Day 1 result: The sleeve priority queue surfaced INTC over equally-scored general candidates (TECK/TMO/ELV). INTC delivered +13.82% (+$185 unrealized) in <8 hours on a Q1 EPS beat 29× consensus + Q2 guide doubling. Without the sleeve, the old monolithic ranking would have selected the general candidates and missed the +14% INTC win.
Full Apr 23 morning debrief → INTC trade journal → LRCX trade journal
Recent Decisions
| Date | Action | Details |
|---|---|---|
| 2026-04-24 AM | BUY MRVL | Day 2 sleeve validation. 9 sh @ $159.54 — re-entry of the score-96 catalyst (Apr 20 MRVL×Google mega_partnership) that the old monolithic scoring systematically missed. Fade-to-flat window captured via sleeve HOT over_reacted threshold override. INTC holds +22% on Q1 blowout; stops placed for LRCX $246 + INTC $70 breakeven-lock; BA stop trailed $205→$223 (+1R rule). Portfolio +15.5% in 2 days ($21,581 → $24,921). |
| 2026-04-23 EOD | 🎯 Sleeve infrastructure validated | INTC sleeve buy +13.82% in <8 hours after Q1 +29× EPS beat + Q2 guide doubling. Eleven INTC catalysts compounded today (CEO demand>supply, $1B ASIC run rate, Trump admin stake +290%). LRCX sleeve buy -3.1% (RBC PT raise to $310 validates). AEM stopped at open ($199.96, -$152 realized). |
| 2026-04-23 AM | BUY LRCX + INTC | First sleeve-routed trades in project history. semi_ai sleeve detected HOT, priority queue picked LRCX (4 sh @ $267.37) + INTC (19 sh @ $66.94) over general candidates. |
| 2026-04-22 AM | BUY BA | 9 sh @ $222.99 on 5-catalyst stack (737-MAX certification + Q1 beat + 2H FCF guide + record $695B backlog + Trump-Xi summit confidence). Closed +5.4% Day 1. |
| 2026-04-21 PM | BUY MRK | 18 sh @ $112.47 dual catalyst (Phase 3 LITESPARK + FDA IDVYNSO HIV approval). |
| 2026-04-21 AM | SELL NFLX | Reaper thesis-break: Reed Hastings exit + Q1 weak guide. -$68 realized. |
| 2026-04-20 PM | BUY AEM | First trade under new Mercury filter-only architecture. Agnico Eagle $3.8B Finland gold M&A. (Stopped out 2026-04-23 open: -$152) |
| 2026-04-20 | Architecture migration | Mercury became filter-only. Autopilot Phase 0.5 now owns scoring + pipeline. |
| 2026-04-17 | BUY NFLX | Post-earnings dip-buy. (Sold 2026-04-21 Reaper) |
| 2026-04-15 AM | BUY CVS | Oracle pre-positioned CMS pullback entry. 27 sh @ $75.83. |
| 2026-04-14 | BUY BLK | BlackRock Q1 blowout: rev +27%, EPS +46%, AUM record $13.9T. |
| 2026-04-10 | DDOG stopped out | -11.8% loss. Stop saved $43 further decline. |
| 2026-04-09 | PANW stopped out | +3.0% gain. Stop saved $132 further. |
| 2026-04-06 | BUY MU | (Stopped 2026-04-20 at +16.8% locked) |
| 2026-04-03 | BUY CCI | Crown Castle activist restructuring. |
| 2026-04-01 | BUY NKE | Nike -14.3% earnings dip. |
| 2026-03-31 | BUY CEG | Constellation Energy nuclear-AI thesis. |
| 2026-03-28 | Rebalance | Sold PONY (-49%), FIG (-58%), trimmed NVDA 49%→10%. |
| 2026-03-28 | Q1 Review | -30.4% vs SPY -7.8%. 0/7 win rate. Full behavioral audit. |
Browse
- Investment Theses — Why we bought or sold
- Trade Decisions — Entry, stop-loss, targets, position sizing
- Autopilot Debriefs — Full market scan reports
- Performance Reviews — Sophia grades our trades
- Current Watchlist — What we’re monitoring
- Portfolio Detail — Allocation, equity curve, sleeve status
How It Works
15 AI agent personas with distinct expertise collaborate through a phased pipeline:
- Mercury drains breaking news every 5-8 min, filters noise, writes alerts
- Sentinel + Phoenix scan ~315 assets (300 stocks + 15 crypto) for movers
- Iris (news), Leo (patterns), Atlas (backtester) analyze flagged candidates
- Ray (macro), Grace (fundamentals), Nate (quant) research the best
- Catherine builds a thesis, Victor stress-tests it
- Diana decides buy/pass, Marcus validates compliance
- Reaper / Harvest manage exits and profit-taking
- Sage monitors portfolio health (concentration, cash, sleeves)
- Sophia reviews performance and grades trades A-F
New (Apr 23): Sector-sleeve routing prioritizes high-conviction sector clusters (currently AI/semiconductors) when regime conditions warrant. The sleeve operates as a ceiling and a priority lane — never a floor — so empty sleeves are fine and there’s no forced deployment pressure.
Each phase has a gate. Only the highest-conviction opportunities become trades. Every decision is documented here with full reasoning.
All decisions use real capital via Alpaca. This is not financial advice.