/decisions · 2026-05-18 · investing · power-cycle-v1.1
v1.1 kills all mechanical price-based exits (the -8% per-trade stop was the alpha-killer: 6 of 10 stop-fires in analog cycles hit names that became long-term winners). Expands basket from TLN+CEG+CCJ to TLN+CEG+CCJ+VST+GEV+KTOS — the 3 adds came from the 2026-05-18 pressure-test (VST + GEV have 13F smart-money corroboration; KTOS is Boom Superpower conviction-overlay). Backtest-v2 on live AI-power cycle: V2-expanded returned +276% vs BH-expanded +258% (+18pp), vs BH-original +169% (+107pp). R-weighted by signal strength: GEV 1R + TLN 0.75R + VST 0.75R + KTOS 0.75R + CEG 0.5R + CCJ 0.25R = 4R bucket cap.
Power's diligence pass landed materially differently from memory's. Memory v1.1 was archived because smart-money-mirror already covered the exposure. Power v1.1 has a tradeable path because:
Two diligence reports converged on the GREENLIGHT-with-revision recommendation:
The bull case for v1.1: the structural anchor is real (ERCOT queue +400% growth, hyperscaler PPA velocity accelerating, GEV Q1 electrification orders exceeded all of 2025), smart-money corroboration on 2 of 3 adds, Boom Superpower verified concrete (not just a press release — $1.25B order book, $300M Series E closed). The mechanical-rule layer was the entire source of underperformance; remove it and the strategy outperforms across the only cycle where the thesis actually applies. R-weighting tilts toward names with multiple convergent signals (GEV: backlog + Q1 orders + smart-money) and underweights names with thinner signals (CCJ: fuel-cycle pull-through only).
The revision risk: N=1 for the actual thesis test. The 2024-AI-power cycle is the ONLY cycle where this specific thesis applies — the 4 analog cycles (1999-dotcom, 2007-commodity, 2010-shale, 2017-Texas-energy) are robustness checks, not primary evidence. Leave-one-out (drop 2024-live) collapses V2 to +23% — same AI-supercycle dependence as memory v1.1. GEV's +730% single-name return on the live cycle could be largely backlog-momentum that doesn't repeat. 1999-dotcom analog cycle excludes Enron/Calpine/Mirant (bankrupt by 2003), so analog returns are survivorship-biased UP. Smart-money signal on VST + GEV is <12mo old; could reverse next quarter. The 4-anchor founder-review gate replaces 5 mechanical exit rules — manual review IS slower than autonomous exit.
The picks: APPROVE locks in the V2-no-mechanical + expanded-basket shape at R-weighted sizing. REVISE walks back (e.g. drop KTOS for thinner signal, equal-weight instead of R-weighted, smaller initial deploy). ARCHIVE concedes the thesis itself is wrong (not just the strategy). DEFER waits for Q2 13F (mid-Aug) to confirm VST + GEV holdings.
| Change | v1 | v1.1 |
|---|---|---|
| Mechanical -8% per-trade stop | Active | KILLED (6 of 10 stops in analogs hit winners) |
| Phase-marker mechanical exit | PPA cadence reversal exit | KILLED; replaced by founder-review gate |
| Tranche-entry logic | 0.5R / 0.5R / 1R at -0/-5/-10% | Unchanged (value-additive) |
| Anchor-based position ADD | None | NEW — drawdowns >20% with bullish anchors = +0.25R add |
| Max per name | 1R ($5k) | R-weighted — see deploy table below |
| Bucket cap | 4R ($20k) | 4R ($20k) unchanged |
| Ticker universe | TLN + CEG + CCJ | TLN + CEG + CCJ + VST + GEV + KTOS |
| Profit trims | +50% trim 0.5R; +100% trim another | RETIRED — let winners run |
| Exit trigger | 2+ phase markers flip bearish 2Q | Single anchor flip = founder review via /decisions/; 2+ HIGH anchors confirmed = bucket close |
| Name | R | $ | Convergent signals |
|---|---|---|---|
| GEV (GE Vernova) | 1.0R | $5,000 | $163B backlog +67% YoY; Q1 orders > all of 2025; only Western SMR moving (BWRX-300 2029); Tiger holds. Strongest single-name signal. |
| TLN (Talen Energy) | 0.75R | $3,750 | First-mover Amazon nuclear-data-center deal (Susquehanna 960MW PPA). Already re-rated. |
| VST (Vistra) | 0.75R | $3,750 | 16.4x fwd P/E vs CEG 24.7x. New 2.2 GW Meta PPA. Tepper + Druckenmiller hold (only 2-manager smart-money name in basket). |
| KTOS (Kratos) | 0.75R | $3,750 | Owns Florida Turbine Tech (Symphony engine designer for Boom Superpower). $1.25B order book. Priced as defense-drone; AI-power unpriced. No smart-money — sized accordingly. |
| CEG (Constellation) | 0.5R | $2,500 | Flagship nuclear pure-play but premium-priced 24.7x. Half-weight of VST given cheaper alternative for similar exposure. |
| CCJ (Cameco) | 0.25R | $1,250 | Uranium upstream pull-through. Zero smart-money corroboration; weakest fundamental link. |
| Total | 4.0R | $20,000 | Fits bucket cap exactly. Paper only, Alpaca sandbox. |
| Param | Value |
|---|---|
| Mode | Paper only (Alpaca paper sandbox) |
| 1R unit | $5,000 |
| Initial deploy | 4.0R total = $20,000 split per R-weighted table above |
| Bucket cap (power total) | 4R ($20k) — full deploy at gate |
| Per-trade stop | None (thesis-stop, not price-stop) |
| Profit trims | None — let winners run (Druckenmiller doctrine) |
| Exit trigger (single signal) | Founder review via /decisions/ page within 7 days |
| Exit trigger (confirmation) | 2+ HIGH severity anchors flip 1-2Q = close ENTIRE bucket |
| Kill switch | Founder channel "halt power" or "pause investing" = immediate close |
| Review cadence | Quarterly (Ray runs 4-anchor pull, surfaces to founder only on anchor flip) |
Ray fires 4R deploy ($20k paper) into Alpaca per the R-weighted table — GEV 1R + TLN 0.75R + VST 0.75R + KTOS 0.75R + CEG 0.5R + CCJ 0.25R. Logs to positions/. Sets up quarterly 4-anchor watch. v1 deploy gate auto-marks superseded. Use the field for any deploy notes.
Approve + sendv1.1 is close but a parameter needs adjusting (e.g. drop KTOS for thin signal, equal-weight instead of R-weighted, smaller initial deploy, keep a wider stop). Name what; Ray re-files v1.2.
Revise + sendIf the thesis itself doesn't pencil — not just the strategy. Ray files as status: archived-pre-deploy with reason. No paper trade. Power bucket closes.
Hold for next quarterly smart-money 13F (re-confirm VST + GEV positions held/added, possibly catch a 2+ manager signal on GEV) before deploying. Q2 13F filings land mid-August.
Defer + send