The oversold band turned over completely this scan: $WMT and $XLU both climbed out (33.2→45.7 and 35.8→43.9), while $MCD, $HD, and $COST all fell in for the first time, and $TSN crashed further to become the cut's new deepest print at RSI 26.4 — worse than $HRL's 25.7 last scan. That's not an improving cohort, just a reshuffled one: the same twelve of fifteen names still show a flat-to-negative year-over-year return with weak alpha, and only $SBUX, $VZ, and $TGT clear the long-trend test cleanly, unchanged for three straight scans. $T's RSI (60.0) again looks like one of the cut's healthier prints even though its year-over-year return (-13.2%) and alpha (-31.8) are both worse than most of the names sitting in the oversold band.

Data as-of the 2026-09-04 close (Friday's settled session) — every name in this cut carries a fresh bar this scan.

Source: defensive.json. Prior scan: 2026-08-28-defensive-scan.

$TSN was this scan's clearest casualty: RSI fell from 36.9 to 26.4 — a 10.5-point drop — on a 7.2% weekly decline, putting it just above the capitulation threshold and making it the cut's new deepest print, surpassing $HRL. Its year-over-year return is negative (-8.5%) against a badly negative alpha (-27.1), so this isn't a clean buy-the-dip setup; the long-trend numbers say the name is structurally weak, not just having a rough week.

The oversold band turned over almost entirely this scan. $WMT and $XLU, the two names that crashed into the band two scans ago, both climbed fully out this time — $WMT to 45.7 (+12.5) and $XLU to 43.9 (+8.1) — while $MCD (34.3, down from 44.2), $HD (37.8, down from 42.4), and $COST (38.1, down from 48.1) all fell newly into the band. Net effect: the band grew from four names to five, but only $HRL and $TSN carried over from last scan's membership. That's a rotation inside the same weak cohort, not a broad recovery — $WMT's improvement still leaves it 9.7% below its 200-day average even with a positive year-over-year return, and $MCD, $HD, and $COST all fail the long-trend test outright (all below their 200-day with negative-to-flat year-over-year returns).

The weakness extends well beyond the oversold band. $MCD, $HD, and $LOW are all below their 200-day average with negative year-over-year returns (-19.3%, -22.0%, and -22.7% respectively) and alpha readings between -38 and -41 — badly lagging the tape over the past year. $COST, now also below its 200-day (-4.6%), carries a near-flat year-over-year return (-4.2%) with weak alpha (-22.9) — a clear deterioration from its prior "flat but near its 200-day" read. $XLU and $XLP remain flat-to-slightly-positive on the year (+2.9% and +4.9%) with weak alpha, and $ECL (+1.6% on the year) sits in the same bucket. $T's RSI (60.0) still looks like one of the cut's healthier prints, yet its one-year return (-13.2%) and alpha (-31.8) remain worse than most of the names sitting in the oversold band — the same misleading-RSI pattern flagged in this cohort for multiple scans running.

The exceptions remain the same three names, unchanged for three straight scans. $TGT is still up 77.4% over the year and 34.2% above its 200-day — the cohort's cleanest long trend — and it stabilized this scan (RSI 63.5 to 63.2) after last scan's sharp cooldown from an extended level. $SBUX (RSI 44.9, up 20.1% on the year, 7.0% above its 200-day) and $VZ (RSI 62.7, up 13.5% on the year, 9.7% above its 200-day) both continue to clear the test cleanly, though $SBUX's RSI cooled meaningfully (55.4 to 44.9) even as its long-trend numbers held. $PPC's rally continues to fade (RSI 62.0 to 53.3), while the trend underneath remains down 31.2% over the past year and below its 200-day — still a bounce, not a reversal.

• $TSN crashed further and became the cut's new deepest print. RSI 36.9 → 26.4 (-10.5), price down 7.2% for the week — now the cohort's worst RSI, surpassing $HRL.
• $WMT and $XLU both climbed fully out of the oversold band. WMT: 33.2 → 45.7 (+12.5). XLU: 35.8 → 43.9 (+8.1) — both remain below their 200-day.
• $MCD, $HD, and $COST all fell newly into the oversold band. MCD: 44.2 → 34.3 (-9.9). HD: 42.4 → 37.8 (-4.6). COST: 48.1 → 38.1 (-10.0) — $COST also crossed below its 200-day for the first time this scan has tracked (vs-200-day -1.5% → -4.6%).
• $HRL partially recovered but remains deep. RSI 25.7 → 28.9 (+3.2), still 8.1% below its 200-day with a -14.7% year-over-year return.
• The oversold band grew from four names to five, with near-complete turnover. Only $HRL and $TSN carried over; $WMT and $XLU exited, $MCD/$HD/$COST entered.
• $SBUX cooled sharply. RSI 55.4 → 44.9 (-10.5), though its long-trend numbers (1Y +20.1%, vs-200-day +7.0%) held up.
• $TGT stabilized after last scan's cooldown. RSI 63.5 → 63.2 (-0.3) — still the cut's cleanest long trend.
• No stale bars this scan — every name carries a fresh 2026-09-04 close.
• Universe unchanged. Same 15 tickers as the prior scan.

RSI measures how stretched a stock is over the past 14 sessions — under 30 signals oversold, over 70 signals overbought. vs SMA20 is the price gap to the 20-day average. 1Y% and vs SMA200 are the numbers that separate a genuine breakdown from a rough week: a name below its 200-day with a flat-to-negative year and a badly negative alpha — like $TSN, $HRL, $MCD, $HD, $LOW, $COST, and even the higher-RSI $T — is structurally weak regardless of how this week's RSI reads. This scan's oversold-band turnover is the clearest example yet of why RSI level alone doesn't tell you which names are "recovering" versus "deteriorating" — $WMT and $XLU climbing out and $MCD/$HD/$COST falling in are opposite directions inside the same structurally weak cohort.

None — $TSN at 26.4 is just above the line, the deepest print in the cut.

$TSN (26.4), $HRL (28.9), $MCD (34.3), $HD (37.8), $COST (38.1) — five names, up from four last scan, with near-total membership turnover: only $HRL and $TSN carried over, while $WMT and $XLU exited and $MCD/$HD/$COST entered. All five are below their 200-day with a negative-to-flat year-over-year return, so none reads as a clean buy-the-dip setup on the RSI alone.

• Genuine breakdown (below 200-day, negative 1-year): $LOW — unchanged from the prior scan; $MCD and $HD carried the same signature but are now in the oversold band above.
• Flat/negative 1-year despite decent RSI (weak alpha, watch closely): $XLU, $XLP, $WMT, $ECL, $PPC, $T — six names where the year-over-year number and alpha undercut a more constructive-looking RSI.
• Genuinely healthy long trend: $SBUX, $TGT, $VZ — the same three names that clear both the 200-day and the 1-year alpha test cleanly, unchanged for three straight scans.

• $TSN just became the cut's new deepest print — does it find a floor, or does it cross into capitulation territory (RSI < 25) next?
• $WMT and $XLU both climbed out of the oversold band while $MCD, $HD, and $COST fell in — is this a genuine rotation toward strength for the exiting names, or noise inside the same weak cohort?
• $TGT stabilized this scan after a sharp cooldown — does the cleanest long trend in the cut hold here, or does the pullback resume?
• Twelve of fifteen names in this cut still show a flat-to-negative year with weak alpha regardless of RSI level — does the defensive complex broadly stabilize, or does the deterioration keep rotating through new names?

• Tape data (price/RSI/SMA/returns): full-scan pre-computed summaries as of the 2026-09-04 close — defensive.
• Prior scan for comparison: 2026-08-28-defensive-scan.