Two names today, and both arrive with the same warning. The broad market reading is a strong 82, but all five leading groups are slowing and only ten of thirty-nine are rising. Both names are short-term overbought — that is recorded as a caveat, not a veto. Underneath a calm index, money is visibly leaving semiconductors and artificial intelligence for energy, materials, insurance and staples: ten fresh buy signals today and not one sell. The record: sixty-two measured picks across twenty-one sessions, and the four most recent groups were re-priced this morning.
The headline reading is 82 with volatility at 15.3, which on its own allows a full list. Underneath, all five strongest groups are decelerating — shipping −27, energy −14, cloud −40, software −28, commodities −14 — and only 10 of 39 groups are rising, 26%. The list is capped at two.
The clearest signal today is not in the finalists but in the timing tape: ten fresh buy signals and no sells, and every one is insurance, energy, midstream, utilities, property, staples or steel. The rotation engine independently picked the same three destinations. Meanwhile semiconductors are −9.2% on the week and the largest names in the income screens fell hard this session.
Options flow agrees and the headline reading cannot see it. Net call premium peaked at +$34.1M at 10:05 ET and had fallen to +$10.2M by 11:20, while net put premium flipped from −$21.5M to +$11.8M and net volume went from +140k to −305k.
Shipping (+9.61), energy (+6.10) and cloud (+5.66) top the momentum table, with software (+4.75) and commodities (+4.12) next. Every one of them is decelerating, cloud most steeply at −39.7. Semiconductors sit at −2.57 momentum with −9.2% on the week, and the names that dominate the income screens — memory, accelerators, AI infrastructure — were among the weakest in the session.
Three independent reads point the same way. The rotation engine names Basic Materials (80), Energy (70) and Consumer Defensive (70), all early rotation, with institutions 83% buy-side in materials. The timing tape fired ten buy signals today, all in insurance, energy and midstream, utilities, property, staples and steel, and zero sells. And the two names that clear the one selection rule the record actually supports sit in materials and energy. Where the finalists are weak is breadth of confirmation, not direction.
The highest relative strength in the whole universe, and the thinnest name on this page. Relative strength 103.98 — more than double the next finalist — stage 2, 9/7 stage scores, fifteen weeks since entering Stage 2, screen score 64.7. Its sector is one of the three the rotation engine names, and it is accelerating (+2.45). A partner announced battery packs at 420 Wh/kg with design wins in commercial drones this morning, which is what the move is. Read the caveats before the strength. This is a $197M company that traded under 25,000 shares today; it is up 7.95% on the session, five hundredths of a point below the intraday veto; RSI is 79.4, the most overbought reading on this page; it carries no analyst coverage and no price target; its earnings date is unpublished, so the earnings gate could not run for it; and the trailing multiple is 99.6. A relative-strength figure of 104 on a company this small is mostly a record of a 154% year, not evidence about next week.
The more defensible of the two, and short-term stretched. Relative strength 42.19, stage 2B, a perfect 10/10 on stage scores, screen score 64.5, and a rotation gem at 55.5. Tankers, which ties it to shipping — the single strongest momentum group. The business is cheap on its own numbers: 5.5× earnings and a 3.7% dividend, up 129% this year, with earnings a clear 42 days out. All twelve moving averages read Buy. It was last flagged on 2026-09-11 at $46.87 and is +18.1% since, now outside the ten-session re-flag window. Caveats: RSI 71.4 with the entire oscillator cluster on Sell — stochastics 93.4, Williams %R −0.7, CCI 136 — so it is extended in the short term. Energy momentum is positive but decelerating at −14.4. The average analyst target is 0.8% above the current price, which is to say the coverage sees no room. Seventy-one weeks since the Stage 2 breakout makes this a mature trend, not a fresh one. Not in the crossover universe, so there is no timing confirmation either way.
No structure is available for HLP or TEN, which is the normal result and not a failure: the income screens want high-volatility large caps and the Stage 2 funnel produces small and mid-cap momentum names. The screens themselves are rich today, so they are worth reading on their own terms.
On the cash-secured put side the cleanest setups by cushion, liquidity and freedom from an earnings event before expiry were META 735 put (4.1% cushion, 1,830 open interest, earnings safely after expiry, and the strongest technical structure on the board — sixteen buy readings and no sells, all twelve moving averages positive), NVDA 217.5 put (8,421 open interest, 43 days to earnings) and SPCX 150 put (26,027 open interest on $15k of collateral). The fattest premium on the board was MU at 99% annualised, and it is the one to be careful with: MU fell 5.0% this session, sits below its 5-, 10-, 20- and 50-day averages, and semiconductors are −9.2% on the week.
Only three of the eleven spreads came back on an uptrending underlying, and all three are rated poor on risk-to-reward: META 680/665 at 15.7:1, SPCX 140/135 at 19:1, SPCX 141/135 at 17.8:1. Held to expiry they need a 94% to 95% win rate merely to break even. Their positive edge exists only under managed exits at half credit.
Worse, the variance risk premium on all three is negative — META's realised volatility is 52.3% against 44.4% implied — so the premium being collected is less than the underlying is actually moving. The screen's own filter excluded five names for exactly this; these three slipped through with negative readings.
The two spreads with genuinely sound structure are not uptrending: IREN 30/29 at 4.6:1 with a strongly positive variance premium, the only good risk-to-payoff rating on the board, and SNDK 1290/1280 at 6.7:1. IREN trades below all twelve of its moving averages and SNDK's short strike carries 183 open interest. Good structure, bad chart; good chart, bad structure. There was no spread today that was both.
| Cohort | Mean alpha | Median | Beat SPY | Note |
|---|---|---|---|---|
| 2026-08-31 · 7 | +4.56% | +2.16% | 4/7 | CVI +40.1 carries it |
| 2026-09-01 · 7 | −1.53% | +2.94% | 4/7 | mean and median disagree |
| 2026-09-02 · 3 | −12.83% | −11.55% | 0/3 | all energy |
| 2026-09-03 · 3 | −9.14% | −8.44% | 0/3 | CRGY, ZYME, PFE |
| 2026-09-08 · 3 | −2.96% | −7.40% | 1/3 | ELPC the only winner |
| 2026-09-11 · 3 | −1.78% | −3.07% | 1/3 | all energy · TEN flagged here |
| 2026-09-15 · 3 | −9.65% | −9.63% | 0/3 | all energy |
| 2026-09-16 · 2 | +2.50% | +2.50% | 1/2 | MPC flagged here |
| 2026-09-17 · 3 | +10.58% | +6.52% | 3/3 | best cohort |
| 2026-09-18 · 2 | −2.50% | −2.50% | 0/2 | ABCL, CMBT |
| 2026-09-21 · 3 | +8.06% | +7.91% | 3/3 | RBRK, ALAB, NBIS |
| 2026-09-23 · 3 | −3.32% | −2.51% | 1/3 | RNG, MRNA, OMER |
| 2026-09-25 · 2 | −0.64% | −0.64% | 1/2 | SN, LCUT |
| 2026-09-28 · 3 | −7.09% | −4.87% | 0/3 | DGII, HUM, RVMD |
| 2026-09-30 · 1 | −1.96% | −1.96% | 0/1 | AMN |
| 2026-10-02 · 2 | −0.38% | −0.38% | 1/2 | SMTC, NAT |
| 2026-10-05 · 3 | +3.12% | +2.99% | 2/3 | IMOS, NTAP, DAC |
| 2026-10-06 · 3 | +1.39% | +1.34% | 2/3 | re-priced today · OKTA +4.95 |
| 2026-10-07 · 1 | −8.11% | −8.11% | 0/1 | re-priced today · AMD |
| 2026-10-08 · 2 | +2.34% | +2.34% | 2/2 | first measure · PAYS, MPC |
| 2026-10-12 · 3 | +0.61% | −0.49% | 1/3 | first measure · HNGE +2.83 |
| Book · 57 of 62 | −1.04% | −1.87% | 25/57 | the 2026-10-08 re-price, unchanged |
| 2026-10-13 · 2 | — | — | — | HLP, TEN · flagged today |
The nine open picks re-priced this morning average +0.28% with a +1.34% median and five of nine ahead of the index. Drop AMD alone, at −8.11%, and the other eight average +1.33%. That is the same tail concentration this record has shown since the first week, and it is why the mean is never reported here without the median beside it.
Conviction rank is inverted again on this sample. Rank 1 averages −0.32%, rank 2 averages +2.16%, rank 3 averages −1.32% — not monotonic, and the top rank is the worst. AMD was a rank 1, so the inversion is tail-driven rather than structural, but the test has now moved in both directions across sessions and the list should not be read as an ordering.
Two of the four worst cohorts on this page were entirely energy. That is why thirteen energy gems were not turned into a list today, even with energy named as an early-rotation sector by three separate reads.