Options AI Screener Desk

Session 2026-09-30 · Wed rebuilt 11:20 ET

One name today, and the reason is worth more than the name. The market checks allowed up to three; the desk's own rules cut it to one. The strongest candidate on every measure was excluded because this desk already flagged it eight sessions ago and a second entry on the same move is not new evidence. A second was dropped on a downgrade, a third because the one confirmation channel available today could not reach it. The record improved sharply: across forty-seven tracked picks the book is now 2.31% behind the index on the mean with nineteen beating it, up from 3.25% behind and sixteen yesterday.

Today's call

One name. Three gates allowed three; the desk's own rules cut it to one.

Why only one

The market reading sits in the middle band, so the list was capped at four before anything else, then cut to three because all five leading groups are decelerating and only eight of thirty-nine are advancing at all. Five names then cleared the strength test. Four were removed by rules, not by preference.

The one that hurts is the best of them. It leads on every measure the desk tracks — and this desk already flagged it eight sessions ago. The standing rule is that a name is not re-flagged inside ten sessions, because a second entry on the same move is a duplicate observation, not new evidence. It stays out. A rule that only bites when it is convenient is not a rule, and this one is the reason the performance record below can be trusted at all.

Gates

Two fired. Breadth is the narrowest this desk has recorded.
1 · Regime62 · middle band 11:06 ETVolatility 15.8 and stable, but only 1 of 3 indices above its 20-day average. 55–69 means half a list and a mandatory confirming signal per name. Ceiling 4 before the other gates.
2 · LeadershipFIRED 11:04 ETAll five highest-momentum groups decelerating: semis +4.69/−29.66, shipping +4.49/−31.31, commodities +3.53/−14.29, cloud +2.38/−40.17, AI +0.84/−23.44. Cap 3. Cloud is shedding acceleration at −40, the steepest reading in weeks.
3 · Intraday heatPassed 11:09 ETNothing near the +8% veto. AMN +1.60%, NTAP +0.95%, ABCL +0.27%, all against SPY +0.52%.
3b · EarningsPassed 11:10 ETNearest print is AMN at 29 days. NTAP 62, S 69, UMC 28. Threshold is 3 sessions — nothing close.
3c · RSIFlag only 11:11 ETAMN 55.8 neutral, oscillators 4 buy / 4 neutral / 2 sell, all 12 moving averages Buy. NTAP at 66.2 carries a net-Sell oscillator block (stochastic 92.5, Williams %R −3.1, CCI 157.1) — the same short-term stretch DGII showed on Monday.
4 · BreadthFIRED · 8/39 11:04 ET20.5% — the narrowest this desk has recorded, against 10/39 Monday and 14/39 a week before. Caps at 3 alongside Gate 2.

Where money is moving

Leadership intact but fading fast; the screen finally points at it.

Semiconductors still lead on momentum (+4.69) but the whole top of the table is losing ground quickly — cloud at −40.2 acceleration, shipping −31.3, semis −29.7. Energy has flattened to +0.27 momentum, tenth of thirty-nine and negative on relative strength. Below the leaders the market is uniformly negative, with solar at −10.4 and uranium −8.2.

One genuine improvement over Monday: the rotation screen named Technology and Healthcare as its target sectors and produced five Technology names against four Healthcare. On Monday nine of eleven were Healthcare while Healthcare was negative on momentum. The screen is now pointing where the tape is, which removes the concentration worry that shaped the last list.

Finalist

One. Four others cleared the strength test and were removed by rule.
AMN$35.51 · AMN Healthcare Services · Healthcare staffing

The only candidate that cleared every rule. Relative strength 32.95, comfortably above the one threshold this desk selects on, and it appears on both source screens — the rotation list at a gem score of 50.1 and the buy screen. Stage 2, daily strength above weekly (9 vs 6), technicals read Strong Buy overall with all twelve moving averages positive, and it is outperforming the index intraday at +1.60% against +0.52%. Earnings are 29 days out, well clear.

Rel str32.95
Stage2
SATA6 / 9
RSI55.8
vs SPY+1.08
Earnings29d
Both screensWeak confirmationNo catalyst

Options structures

None for the finalist. The normal result.

Both income screens were rebuilt by this run — 652 cash-secured puts and 389 bull put spreads — and the finalist appears in neither. Those screens want large, high-volatility index names; this funnel produces mid-cap trend names.

Dropped, and why

Four cleared the strength test. Each was removed by a stated rule.
ABCL already carried — 8 sessions The best name on the board, excluded by our own rule. Relative strength 142.37 and rising 28.71, the top composite score at 72.4, a perfect 10/10 stage score, smart score 10, Strong Buy from both analyst panels, hedge-fund 0.66, insider 0.86, the only breakout flag and the only non-zero smart-money reading. Fresh JPMorgan initiation at Overweight. It was flagged on 2026-09-18, eight sessions ago — inside the ten-session re-flag window. It is up 17.5% alpha since then, which is precisely why a second entry would be double-counting one move rather than finding a new one.
S negative catalyst Cleared relative strength at 29.86 and carried a valid confirming signal (hedge-fund 0.59). But Bernstein SocGen downgraded it from Outperform to Market Perform on stretched valuation, amid a cybersecurity-wide unwind and disclosed insider selling. News sentiment 0.0. That combination is the one explicit veto in the method — and it applies even though the stock is up 2.6% today, because the downgrade landed yesterday evening.
NTAP no confirming signal available Arguably the second-best name: relative strength 44.92 and rising, the highest stage score at 63.6, sitting exactly at its 52-week high, with a BofA target raise this morning. Its hedge-fund score is 0.3663 against a 0.4 threshold, dark pool is zero and the crossover scan never covered it — so no confirmation channel was open. Two independent caveats support the exclusion anyway: it trades 5.3% ABOVE its average analyst target, and its oscillator block is net Sell.
UMC quality fails Relative strength 63.42 looks strong until you read the rest: RS is falling (−1.19 on the week), smart score 4, analyst consensus Sell on both panels, and an average price target 44.7% BELOW the current price. Hedge-fund 0.24 also failed the confirmation test. Cleared one filter and failed everything else.
IART · ODTX · ADI relative strength All three on the rotation list, all three under the 20 threshold — 10.18, 9.24 and 9.57. ADI is the notable one: it carries a dark-pool score of 59, the only meaningful reading in the whole candidate set, and would have had a valid confirming signal. The strength rule cut it anyway.

Forty-seven picks, fourteen flagging sessions

Alpha versus SPY from each flag-date close, re-priced at 11:10 ET. All 47 priced.
CohortMean alphaMedianBeat SPYBestWorst
2026-08-31 · 7+1.28%+0.54%4/7CVI +24.8%GEN −28.8%
2026-09-01 · 7−2.66%+1.65%4/7CRWD +14.9%SLB −18.9%
2026-09-02 · 3−16.95%−15.38%0/3VET −14.0%HP −21.5%
2026-09-03 · 3−4.77%−6.37%1/3PFE +0.2%ZYME −8.1%
2026-09-08 · 3−7.98%−6.31%0/3ELPC −2.3%MTDR −15.3%
2026-09-11 · 3−4.30%−7.21%1/3TEN +1.7%UROY −7.4%
2026-09-15 · 3−11.21%−9.30%0/3CVE −7.3%SM −17.1%
2026-09-16 · 2−5.88%−5.88%0/2MPC −4.2%LPG −7.5%
2026-09-17 · 3+12.76%+2.51%2/3IOVA +38.1%SMTC −2.3%
2026-09-18 · 2+5.52%+5.52%1/2ABCL +17.5%CMBT −6.5%
2026-09-21 · 3+6.14%+4.96%3/3ALAB +9.3%RBRK +4.1%
2026-09-23 · 3−3.90%−5.05%1/3MRNA +0.9%OMER −7.6%
2026-09-25 · 2−2.96%−2.96%1/2SN +1.8%LCUT −7.7%
2026-09-28 · 3−0.50%−0.29%1/3RVMD +1.4%HUM −2.6%
All 47−2.31%−3.97%19/47trimmed −2.81% · ex top two −3.81% · −108.4 pts
2026-09-30 · 1———flagged today · first measurable reading next session

The best single-session improvement this record has had

Yesterday the book was 3.25% behind the index with 16 of 47 beating it. Today it is 2.31% behind with 19 of 47, and cumulative drag has fallen from −189.9 points on Monday to −108.4. Four cohorts are now positive, and the 09-17 cohort has gone to +12.76% mean on the back of IOVA at +38.1% alpha.

Apply the same trimming to a good day as to a bad one. Removing the two best names leaves −3.81%, worse than the headline, and the trimmed mean is −2.81%. So this is a genuine improvement in a book that still sits below its benchmark, not a turn. The 09-02 cohort remains at −16.95% and three cohorts are still worse than −7%.