Recommendations
1 to acquire · 3 to dispose
| Ticker | Name | Close | Action | Action Price Where to act — a buy-limit on a pullback or a buy-stop above resistance, not the current price. Sanity-check vs Close: on a held name, an action price stranded above the close means the queued level was set when the stock was higher — stale, awaiting reprice. | Stop The risk-defined exit, set below the entry for a long (~1.2–1.8×ATR). A held long whose Stop sits at/above the Close is already below its stop — a stale box awaiting reprice. | 1wk Model estimate of the likely price at this horizon — a forward projection, not a guarantee, and separate from the trade's Action Price or Stop. Where the ML estimator is enabled these are per-ticker calibrated forecasts (~80% coverage band) that are refreshed periodically and can drift between refreshes; otherwise they're interpolated from the signal's swing target. Confirm against the settled close. | 4wk Model estimate of the likely price at this horizon — a forward projection, not a guarantee, and separate from the trade's Action Price or Stop. Where the ML estimator is enabled these are per-ticker calibrated forecasts (~80% coverage band) that are refreshed periodically and can drift between refreshes; otherwise they're interpolated from the signal's swing target. Confirm against the settled close. | 12wk Model estimate of the likely price at this horizon — a forward projection, not a guarantee, and separate from the trade's Action Price or Stop. Where the ML estimator is enabled these are per-ticker calibrated forecasts (~80% coverage band) that are refreshed periodically and can drift between refreshes; otherwise they're interpolated from the signal's swing target. Confirm against the settled close. | Rationale |
|---|---|---|---|---|---|---|---|---|---|
| CVX | Chevron Corp | 199.77 | ADD — QUEUED @197.97 | 197.97 | 190.82 | 200.46 (+1.3%) | 206.26 (+4.2%) | 222.54 (+12.4%) | Chevron is the first name this book has rated a buy in five sessions, and it earns it on the setup we most trust: a pullback toward its volume-weighted average price inside an intact uptrend, not a chase into strength. The business behind it is firing - a $12bn adjusted-earnings quarter, revenue of 1bn against a 3bn consensus, a record 4bn of debt repaid, 5bn returned to shareholders in the quarter, and a fresh oil and condensate discovery offshore Angola that carried the shares near an all-time high. Fair value reads roughly full rather than cheap, so we buy on the dip to the model's limit rather than at the market, and we keep the add small. OPERATIONAL: BUY - ADD cv73 PullbackBuy_VWAP, Buy_Rank 1, 29.97 -> of the held 1,104 =. Entry_Distance_ATR -0.50 (close sits 0.50 ATR ABOVE the limit) so this is a GTC LIMIT, not a market fill - queued, expires 2026-09-01 (anchor + 3 sessions). L1 satisfied (PullbackBuy_VWAP is the favoured family). val_adj -4 = clamp(round(-0.7441 x 5)); scorecard PullbackBuy_VWAP|buy -1 (n=7, hit_1w.429, hit_4w.000 - a weak 4wk record, recorded against the card); adj conviction 73 -4 -1 = 68, clear of the 65 floor. Size modulator 1.0x (MoS -0.744 is inside the -1 tier line). Quality_Pass True and irrelevant here - the gate blocks new entries only, never ADDs. STOP CONFLICT, stated openly: the ADD tranche carries the model stop, while the existing keep the trail (0.27% cushion, the tightest in the book). A fill at is below that trail - these are not contradictory because the trail is CLOSE-based and the limit is INTRADAY: an intraday dip to that closes back above fills the add without firing the trail. If CVX settles below the trail fires first and this order must be withdrawn at the next EOD. L8: executes at the 08-28 open, pending. No CVX-specific catalyst in the 10-day horizon window. · news |
| CB | Chubb Ltd | 338.31 | SELL 88.52% | 338.31 | 327.02 | 338.31 (+0.0%) | n/a | 321.56 (-5.0%) | We are stepping out of most of our Chubb position because the price has broken decisively through the risk line we set for it and the model's three-month view has turned negative. The insurer itself is doing nothing wrong - it beat on its last quarter, premiums written grew 3.6%, the dividend was lifted to and Wall Street still carries a Moderate Buy with a target about 9% above today's price. But we do not hold positions through a broken stop on the strength of a fundamental story: the stock has now closed below its stop by two-thirds of a daily trading range while the forward projection points 5% lower, and that combination is our exit rule. We keep a small residual rather than escalating to a full exit, because the valuation case here is neutral and gives us no licence to sell more than the model asks. OPERATIONAL: ORIGINATION IS L2 (decisive close below stop + negative 12wk), NOT the label - close is 1.07% / -0.65 ATR below the stop with 12wk = -4.95%. This is CLAUDE.md S9's named contrast case (a decisive stop-loss break), so it is floor-independent: the live STRONG SELL - REDUCE cv60 / Sell_Rank 1 / 88.52 CORROBORATES but cannot carry the card on its own, because RallySell_BearTrend|sell is this book's WORST actionable cell at -8 (n=11, hit_1w.091, hit_4w.125) and 60 - 8 = 52 sits below the 65 floor. Stated explicitly rather than argued away. Flagged one session in advance: the 08-26 entry named CB 'the tightest cushion in the book - re-arms on a fresh label OR a close below '; BOTH arrived together. Trend context: close is below EMA50 but still +4.17% above EMA200, so the EMA200 exit form does not apply - the stop break does. PORTION 88.52% NOT ESCALATED: MoS_FV -0.006 is nowhere near the -3 line, so S9(b) grants no tier bump; the 1,880-sh / residual is left standing on discipline rather than rounded away (contrast DVA on 08-26, where MoS -3.376 did entitle the escalation). Trim-discipline gates G1-G3 do not apply (they protect uptrending profit-trims; this is a bearish stop break on a losing position). RESIDUAL STOP RE-BASED -> (the scan level): the old watermark has already fired AND been acted on with an 89% cut, which is exactly the post-reduction re-base ADR-0020 contemplates - this is not walking a stop down under an uncut position. Realised loss on the tranche ~-vs the cost. L8: executes at the 08-28 open, pending. · news |
| KO | Coca-Cola Co | 89.06 | SELL 30% | 89.06 | 89.06⚠ | 89.55 (+0.6%) | n/a | 100.17 (+12.5%) | We are banking roughly a third of a large, profitable Coca-Cola position that has gone quiet. The position is up 5.9% and sits more than three daily ranges above our cost, which arms our profit-taking rule; the trigger is that the model has now failed to find a tradeable signal in the name for three sessions running while the price has drifted from to and is now sitting exactly on its stop. The company is not the problem - it raised 2026 guidance to about 5% organic revenue growth and 9-10% comparable EPS growth, has beaten five quarters in a row, and lifted its dividend for a 64th consecutive year - which is precisely why we take a third rather than the whole thing. At 26x trailing earnings on mid-single-digit growth, letting a silent winner give back a 27% year-to-date run is the risk we are declining to carry. OPERATIONAL: ADR-0012 armed + fired, tick (d). ARMED at Gain_ATR 3.42 (+5.85% on an cost; unrealised on the book's 2nd-largest line). FIRED: NO TRADE cv0 on 08-25, 08-26 AND 08-27 = three consecutive prints against a requirement of two. Note the row prints PROFIT_ARMED but so do all 27 held lines including DVA at Gain_ATR -10.8 - the flag is degenerate (systemic finding 1), so armed status is derived by hand from Gain_ATR >= 3.0, this book's standing convention. Other ticks dead: (a) no secondary Exit_Warning code, (b) conviction flat, (c) 12wk vs the threshold, alive by 10.3%. ONE tick -> profit_take_1 = 30%. Never a full exit from a profit-take. G1 cannot soften it (NoTrade cv0 is not a BreakoutUp_*/MomentumContinuation_* family at cv>=80, and ticks (b)/(c)/(d) fire through G1 regardless). Valuation GROWTH_PRICED_IN MoS -2.250 is short of the -3 tier line -> up-ranks only, no bump. Scorecard NoTrade|sell +0 (n=43, hit_1w.486, hit_4w.364) - a weak 4wk record, recorded against the card, but ADR-0012 is a risk rule about banking a fired winner, not a directional forecast. RE-ISSUE of the 08-26 card, repriced -> per the action-price provenance rule (the 08-26 card never executed - executor blocked); the one-session delay COST RETAINED keep the stop at - now ZERO cushion, the tightest in the book; the ADR-0012 chandelier (close - 2.0 x ATR =) would LOOSEN it by 3.2% and is refused, consistent with the 08-26 entry which rejected the same. A further close below fires the remainder. Realised gain on the tranche ~+Ex-dividend on 09-15, outside the 10-day window. L8: executes at the 08-28 open, pending. · news |
| NUE | Nucor Corp | 252.37 | SELL 30% | 252.37 | 248.53 | 251.97 (-0.2%) | n/a | 257.54 (+2.0%) | We are trimming a third of a small Nucor position that is up 14% but has stopped generating a signal - five sessions in a row with no tradeable read while the model's three-month projection has flattened to roughly +2%. Nucor's own results are strong: second-quarter earnings of per share against a year earlier, sales up 23%, management guiding to a better third quarter, and steel imports still running about 30% below last year under Section 232. The offsetting risk is the one the model is reacting to - reports that the US and Canada will halve steel tariffs from 50% to 25% knocked the shares down almost 6% in a session, and this is a cyclical whose entire margin structure is a policy variable. We bank a third and keep the rest. OPERATIONAL: ADR-0012 armed + fired, tick (d) ONLY - the portion steps DOWN from last night's 50% to 30% because tick (c) has un-fired. ARMED at Gain_ATR 3.66 (+14.32% on a cost). FIRED: five consecutive NO TRADE cv0 prints (08-21, 08-24, 08-25, 08-26, 08-27). TICK (c) UN-FIRED: 12wk vs the dead-view threshold (1.02 x) - alive by. On 08-26 the same test fired by and the card was written at 50%; tonight it fails by and the card drops to 30%. That swing driving a 20-point portion change is exactly the rounding-noise fragility flagged against the 08-26 card, and it is honoured mechanically in BOTH directions - the same discipline that cut against the book on 08-26 now cuts in its favour. ONE tick -> profit_take_1 = 30%. Valuation GROWTH_PRICED_IN MoS -2.278, short of the -3 line -> no bump. Quality_Pass False (gates new entries only; never blocks a trim). Scorecard NoTrade|sell +0 (n=43, hit_4w.364). RE-ISSUE of the 08-26 card, repriced -> and resized ->; delay cost on the originally-called was -Immaterial in dollars either way - = on a book - and written because the rule fired, not because the size matters. RETAINED keep the stop at (1.55% cushion); peak stop stands and was never chased down. Realised gain on the tranche ~+L8: executes at the 08-28 open, pending. · news |
| BAC | Bank of America Corp | 61.17 | HOLD | — | 61.28⚠ | n/a | n/a | 65.12 | Bank of America printed a sell label today but we are not acting on it. The signal is very weak, the price is only a whisker below our stop rather than decisively through it, and the model's own three-month projection still points 6.5% higher - so the two halves of our exit test disagree and we hold. The bank continues to trade below most intrinsic-value estimates after a 139% three-year run, goes ex-dividend on 09-04, and its own August fund-manager survey shows a record 56% of investors expecting no landing. We keep the stop as the live line and let the price decide. OPERATIONAL: STRONG SELL - REDUCE cv22, Sell_Rank 2, 88.52 - REJECTED on THREE independent counts. (1) L2 fails: the close is only -0.18% / -0.11 ATR below the stop, a graze not a decisive break, AND 12wk is POSITIVE +6.46% - L2 requires both. (2) Floor fails badly: cv22 on the RallySell_BearTrend|sell cell at -8 (n=11, hit_1w.091 - the worst actionable cell in the book) gives adj conviction 14, 51 points below the 65 bar. (3) ADR-0012 unavailable: Gain_ATR 1.01 is not armed, so the 27-point conviction collapse cannot fire tick (b). Contrast CB, where the same label and the same -8 cell DID produce a card - because there the stop broke by 0.65 ATR with a negative 12wk and L2 carried it. Same label, same bad cell, opposite verdict, on the stop-break test alone. line, +1.73%, close +10.35% above EMA200. Stop HELD (scan would loosen). Re-arms on a decisive close below or a negative 12wk. |
| AXP | American Express Co | 334.16 | HOLD | — | 330.57 | n/a | n/a | 315.19 | American Express is the largest position in the book at and the weakest on forward view, but it still has not given us a reason to sell. The model rates it avoid rather than sell, the price is holding above the stop, and an avoid is not an exit instruction. We keep it on a short leash. OPERATIONAL: AVOID cv57 -> cv52, not an actionable sell label. No L2: close is +1.09% / +0.66 ATR ABOVE the stop. Not armed (Gain_ATR -3.27). 12wk = -5.68%, the second-worst forward view among material holdings. Now -5.06% vs cost and 11.6% of NAV. Close sits below EMA50 and only +0.43% above EMA200 - a decisive EMA200 loss WITH the negative 12wk already in place would originate an exit under S9's second form, so this is the name closest to firing without a label. Scan stop printed ABOVE the close (short-side level on a long-only book) and REJECTED - 4th consecutive session, systemic finding 4. Re-arms on any actionable sell label, a close below, or a decisive close below EMA200. |
| VRSN | VeriSign Inc | 294.73 | HOLD - trail raised | — | 280.81 | n/a | n/a | 311.94 | VeriSign is our best-behaved winner - four straight sessions rated hold, up 15%, with the three-month view still 6% above today's price. It has run far enough above our cost to arm profit-taking, but nothing has deteriorated, so the correct response is to raise the stop rather than sell any of it. OPERATIONAL: ADR-0012 ARMED (Gain_ATR 5.50, +14.94%) but ZERO ticks fired -> mandatory trail-raise, not a trim. (a) no secondary Exit_Warning code; (b) conviction, UP; (c) 12wk vs the threshold, alive by 3.8% for a third session; (d) 4th consecutive HOLD, no NO TRADE chain. STOP RAISED -> (4.96% cushion) - the scan level is a genuine tightening and is taken. The FailedBreakdown_Up_20D|sell cell sits at -5 (n=5), so this recovering HOLD earns no extra credit. Immaterial at Watch tick (b): a slip to cv <= 48 next session fires. |
| OXY | Occidental Petroleum Corp | 59.17 | HOLD | — | 57.90 | n/a | n/a | 65.02 | Occidental is the best position in the book, up 21%, and it is behaving - a second straight hold rating with the three-month view about 10% above today's price. Armed for profit-taking but with nothing deteriorating, we raise the risk line and keep the whole position. OPERATIONAL: ADR-0012 ARMED (Gain_ATR 6.96, +20.98% - the book's best) but ZERO ticks -> trail-raise. (a) no code; (b) conviction flat; (c) 12wk vs the threshold, alive by 7.7%; (d) chain broken two sessions ago. STOP HELD (2.19% cushion); the scan's would loosen by 2.9% and is refused. Dividend raised 8% to, ex-date 09-10, outside the window. line. |
| CVX | Chevron Corp | 199.77 | HOLD - trail | — | 199.24 | n/a | n/a | 222.54 | Separate from the queued add above, the Chevron shares we already own stay put on a tightened stop. The position is up 18% and the signal has turned outright bullish, so there is nothing to bank - but the risk line is now uncomfortably close to the price and we are not moving it further away. OPERATIONAL: this is the EXISTING tranche, distinct from the QUEUED 331-sh add. ADR-0012 ARMED (Gain_ATR 8.55, +18.07% - the highest in the book) with ZERO ticks: (a) no code; (b) conviction, UP, and the label improved HOLD -> BUY - ADD; (c) 12wk vs the threshold, alive by 9.2%; (d) chain broken. STOP HELD - a 0.27% cushion, the TIGHTEST in the book and stated as uncomfortable; the scan's would loosen it by 4.2% and is refused. The queued add carries its own tranche stop; see the CVX buy card for why a close-based trail and an intraday limit can coexist. |
| DAL | Delta Air Lines Inc | 81.14 | HOLD | — | 81.36⚠ | n/a | n/a | 73.23 | Delta is our worst performer at -12.5% and its stop has been grazed, but not decisively enough to sell. The price is a tenth of a daily range below the line - inside normal noise - so we hold and let one more session decide. A genuine break takes us out. OPERATIONAL: L2 CANDIDATE, DECLINED ON DECISIVENESS. The negative half is emphatic - 12wk = -9.75%, and the position is -12.52%, the worst in the book. The break half is not: close is only -0.27% / -0.10 ATR below the stop, against CB's -0.65 ATR which did qualify. Judgement stated openly - a tenth of an ATR is a graze inside daily noise, not a decisive break, and this book does not exit on a graze. NO TRADE cv0, not armed (Gain_ATR -5.18), close still +7.77% above EMA200. line. Stop HELD. RE-ARMS on a close below (0.5 ATR through the stop), which would make this the largest L2 exit available in the book. |
| MCO | Moody's Corp | 509.02 | HOLD | — | 493.86 | n/a | n/a | 561.31 | Moody's has gone quiet again after a strong run. The position is up 4% with the three-month view 10% higher, and there is no signal in either direction, so we hold. OPERATIONAL: HOLD cv73 -> NO TRADE cv0 - the anticipation setup that printed a Buy Stop at on 08-26 has lapsed, so the ANTICIPATE slot stays empty for a 27th session. Not armed (Gain_ATR 1.86, below the 3.0 line), so ADR-0012 cannot fire on the NO TRADE print. 12wk = +10.27%, the best forward view among material holdings. line, stop HELD (3.07% cushion). Re-arms if the Action ever prints BUY - ANTICIPATE in snapshots - MCO remains the name most likely to take that slot. |
Outcome & track record
Accuracy at the time of this record.
141 Graded decisions Past recommendations old enough to score against what actually happened (target hit / stopped out / closed). This is the evidence base — the bigger the number, the more history behind the calls. Click to inspect every one. See the decisions →
-0.086 1-week calibration How much of the eventual 4-week move typically shows up in the first week (0 = almost none, 1 = most of it). Use it to read the 1-week estimates: a low number means prices drift toward the target slowly, so don't over-read a soft week-1.
46 Published records How many dated end-of-day verdicts exist for this book — the length of the public track record you can browse in the Archive below.