Four candidates tested under the house protocol (full grids reported, cal/val halves, cost stress, controls). Scripts in validation/, per-study reports alongside.

Verdict table

#	Candidate	Result	Evidence
P0	Multi-asset daily trend (28 ETFs, dual SMA200+ROC60)	PASS — strongest breadth result in repo history	`etf_trend_scan.md`: dual 200/60 positive on 28/28 assets (median +7,189 bps / 20y), 20/28 both-halves-positive; dual 200/120 → 24/28 both halves; taker-resilient. 340/448 grid cells both-halves-positive
P1	Classic 12−1 cross-sectional momentum, long-only	CONDITIONAL — promising but survivorship-inflated	`ts_momentum_121.md`: +20.0% CAGR net, Sharpe 0.85 vs SPY +14.4%/0.82 — but universe = today's S&P members (momentum loves past winners → biased UP). L/S version negative gross (short leg poisoned by same bias). Needs clean point-in-time universe before any belief
P2	Overnight premium + turn-of-month	Real effect, NOT tradable	`overnight_tom.md`: overnight +8.9%/yr Sharpe 0.76 gross → −34%/yr after daily round trips; TOM +3.4%/yr gross → +0.9% net
P3	Portfolio construction over P0	PASS as diversifier	`etf_trend_portfolio.md`: combined 1x CAGR 4.2%/Sharpe 0.66/MaxDD −12.3%; crash-gate variant Sharpe 0.82/DD −8.1%; vs SPY DD −53.9%. Managed-futures risk profile

What round 1 actually found

The repo's one robust edge — slow-memory trend — generalizes across asset classes, exactly as time-series-momentum literature predicts. This is no longer "a crypto quirk": same rule family, 28 instruments, three asset classes, fee-resilient.
Portfolio-level product exists: 28-sleeve trend book delivers equity-class Sharpe with 1/4 the drawdown of buy-and-hold, plus 2008-style crash protection. As a sleeve beside gate60/flagship it improves the aggregate.
Momentum (P1) is the one lead worth pursuing further only after fixing survivorship (point-in-time index membership). Do not trust +20% until then.
Calendar/microstructure premia (P2) are confirmed present and confirmed untradable at our costs — consistent with every intraday finding here.

Recommended next steps

Deep-dive P0→P3 chain: per-sleeve vol targeting, correlation-aware weights, rebalance-band tuning, and a walk-forward robustness check of the portfolio (not per-cell) metrics; MC permutation null on the combined curve.
P1 survivorship-clean rerun (Nasdaq/NYSE point-in-time lists via data/universe.py pool machinery).
Paper-trade a small ETF trend sleeve beside the live bots if the deep-dive holds.

Round 1 Deep-Dive (2026-08-23, later same day)

P3 portfolio robustness (reports/etf_trend_portfolio_robust.md)

test	result
NULL-R permutation (positions fixed, returns shifted, 1000 reps)	null mean Sharpe 0.57 vs realized 0.66, p = 0.208
Rule neighbors (dual200/120, dual100/60, ma200, roc60)	Sharpe 0.63-0.78 - robust, not cell-specific
Leave-one-out (28 drops)	Sharpe range [0.63, 0.68]
Inverse-vol weighting	Sharpe 0.75, MaxDD -8.4% (best risk-adjusted variant)

Honest reading: NULL-R says most of the portfolio Sharpe is secular-bull drift through trend exposure, not timing skill (p=0.21). Same conclusion as phase-13 MC-1 for ETH: beta capture with a filter. The sleeve's value is drawdown control (-8 to -12% vs SPY -54%) and crash convexity, not alpha. Deployable as a diversifier; do not market it as market-neutral alpha.

P1 momentum, point-in-time eligibility (reports/ts_momentum_121_pit.md)

universe	net CAGR	Sharpe	MaxDD
biased (today's members)	+20.0%	0.85	-55%
PIT-inclusion-fixed	+12.3%	0.61	-51%

Inclusion look-ahead alone was ~8pp/yr of the headline. After the fix, momentum no longer beats SPY B&H (+14.4%) - and delisting survivorship still inflates the remaining number (no keyless delisted-price source). Verdict: P1 downgraded to NOT VALIDATED on available data.

Round-1 final standings

Multi-asset daily trend portfolio: VALIDATED AS DIVERSIFIER (beta-harvest with drawdown shield; p=0.21 vs timing null).
12-1 equity momentum: NOT VALIDATED (universe artifacts).
Overnight/TOM premia: real gross effects, untradable after costs.
Equity stat-arb L/S: rejected round 0.

Only actionable product change from round 1: consider a small multi-asset ETF trend sleeve (inverse-vol weighted) beside gate60/flagship after a forward-paper period. No new live deployments without that paper phase.