The regime gate fired at its lowest reading on record. No new long ideas today — that is the output, not a gap in it. The re-pricing still ran, and it forced a retraction: the hedge-fund signal this page has led with for three sessions does not survive removing the tails.
The threshold for any new long is 55. Today reads 32 — the lowest since tracking began, and a 45-point collapse from Tuesday’s 77. Under the method that means no new long ideas, full stop. What would have cleared the funnel is listed below for the record; it is not a shortlist and is not ranked.
Standing down was authorised on 2026-09-01. This is the first time Gate 1 has fired.
partial: false.| Sector | Momentum | Accel | 1M | Read |
|---|---|---|---|---|
| Shipping · BOAT | +12.38 | −16.82 | +16.80% | Clear leader, sharply decelerating |
| Commodities · DBC | +7.71 | −9.57 | +9.61% | Second, also fading |
| Energy · XLE | +6.67 | −10.87 | +8.26% | The ledger’s best sector, decelerating |
| Crypto · BLOK | +6.14 | −4.88 | +8.78% | Softening, −1.9% today |
| Nuclear · URA | +4.84 | +4.04 | +1.70% | Only accelerating leader · flagged WATCH |
| Clean Energy · ICLN | +0.04 | +8.11 | −1.15% | Tagged TURN — inflecting from a low base |
Gate 2 held only because URA is still accelerating. Breadth is 16 positive sectors of 39. Airlines (JETS) is tagged OUT.
| Ticker | Sector | Stage | SATA W/D | RS | Wks in S2 | Lists |
|---|---|---|---|---|---|---|
| KGEI | Energy | 2 | 10 / 9 | 26.5 | 4 | Stage 2 only |
| EQNR | Energy | 2 | 10 / 9 | 29.9 | 37 | All three · buy |
| TNK | Energy | 2B | 10 / 9 | 28.4 | 70 | All three · buy |
| TEN | Energy | 2B | 10 / 9 | 25.3 | 66 | All three · buy |
| CVX | Energy | 2B | 10 / 9 | 12.5 | 62 | All three · buy |
| MEOH | Materials | 2A | 9 / 9 | 15.1 | 1 | All three · buy · earliest |
| CHYM | Financials | 2 | 9 / 8 | 40.4 | 7 | Stage 2 + buy |
| SKHY | InfoTech | 2 | 9 / 10 | 20.6 | 2 | All three · hold |
| TBBB | Staples | 2B | 10 / 8 | 30.6 | 47 | All three · buy |
Five of nine are Energy, which is where the
ledger says the funnel has worked. Not one has a confirming EMA crossover today, and
dark_pool_score is 0 for every name — the known coverage gap, not evidence of absence.
spread_screener returns zero rows, so there are no bull put spreads — the defined-risk
vehicle Gate 1 would normally point toward. The CSP table is fresh but holds only 12 tickers
(AAPL, MU, IWM, META, GOOGL, IBIT, SPCX, TSLA, USO, AMZN, VRT, NVDA) and every single row is
rated hold. Zero overlap with the Stage 2 names, as usual — two systems sharing a dashboard.
An all-hold CSP table in a risk_off tape is consistent with the regime read rather than a contradiction of it.
| Cohort | Avg alpha | Beat SPY | Best | Worst |
|---|---|---|---|---|
| 2026-08-31 · 7 | +0.64% | 3/7 | CVI +17.57% | ESTC −12.66% |
| 2026-09-01 · 7 | −2.96% | 2/7 | VLO +7.18% | PRGS −11.59% |
| 2026-09-02 · 3 | −2.15% | 1/3 | VET +1.67% | HP −6.00% |
| 2026-09-03 · 3 | −2.49% | 1/3 | CRGY +3.14% | ZYME −8.05% |
| 2026-09-08 · 3 | +0.75% | 2/3 | MTDR +1.63% | ELPC −0.00% |
| All 23 | −1.21% | 9/23 | median −1.19% · two worst carry 87% of the drag | |
Conviction rank is still inverted. Ranks 1–3 average −2.64%; ranks 5+ average −0.99%. Six sessions in, the ranking has never once pointed the right way. Cohorts after 09-01 were left unranked, which was the right call.
On the headline number it still looks strong: names with hedgeFundsScore > 0.4
average +0.10% alpha against −3.68% for the rest — a +3.78 spread,
wider than the +2.53 reported Tuesday.
But this desk’s own tail-concentration rule says to check whether a few names are carrying it. They are. Drop the two biggest winners and the two biggest losers — CVI, VLO, ESTC, PRGS — and the spread inverts to −0.92. Removing CVI alone takes it from +3.78 to +2.53, which is exactly the number reported on Tuesday and suggests that figure was already one name deep.
Prior runs also never wrote TipRanks fields for the 08-31 and 09-01 cohorts, so the claim was
untestable from the ledger until today’s back-fill. Those 14 rows are stored separately in
stage2_tipranks_backfill_2026-09-10.csv and carry today’s values, not flag-time values —
a real weakness, since hedge-fund scores drift. Treat the institutional-positioning thread
as unproven, not established.
| Feature | Mean spread | Median spread | n | Verdict |
|---|---|---|---|---|
| Sector = Energy | +6.40 | +4.89 | 11/12 | Holds on both. Strongest yet. |
| Analyst upside > 15% | −5.97 | −4.36 | 9/14 | Strongly anti-predictive |
| On Rotation Gems | −4.23 | −3.26 | 15/8 | Anti-predictive |
| StrongBuy consensus | −3.78 | −2.86 | 10/13 | Anti-predictive in every sample |
| Converged on 3+ lists | −3.13 | −2.86 | 8/15 | Anti-predictive |
| Mansfield RS > 20 | +2.32 | +3.21 | 10/13 | Holds, modest |
| Regime ≥ 71 at flag | +2.67 | +1.66 | 13/10 | Supports the regime gate |
| Hedge-fund score > 0.4 | +3.78 | +2.06 | 15/8 | Fails tail removal — retracted |
| Smart Score ≥ 8 | +1.87 | +2.88 | 12/11 | Has flipped sign twice — noise |
“YTD gain > 50%” shows a +6.11 spread and looks like a second strong signal. It is not: 8 of those 9 names are Energy, and the one that is not (CRWD) was the worst non-Energy pick in the book at −8.6%. The two findings are the same finding.
The split is stark. 8 of 11 Energy picks beat SPY; 1 of 12 non-Energy picks did. That is the cleanest result in the ledger, and it argues the funnel adds value when it is pointed at the leading sector and destroys value when it is not. It is also a warning: Energy has led for six weeks and is now decelerating at −10.87. This finding may be measuring one sector’s run rather than anything durable about the screen.
as_of and conviction_rank columns.