This website is for educational purposes only and does not constitute investment advice. Always do your own research and assess your own risk tolerance before making investment decisions.

US End-of-Day Verdict 27 Aug 2026, 16:00:00 GMT-4

Berkshire Mirror — End-of-Day Verdict

EOD 2026-08-28

Recommendations

0 to acquire · 4 to dispose

TickerNameCloseActionAction 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
CBChubb Ltd338.60SELL 88.52%339.92327.02339.92 (+0.0%)n/a321.37 (-5.5%)Chubb is being cut back because it still has not climbed back above the risk line that triggered the exit a session ago, and the model now sees the shares roughly 5.5% lower over three months. The insurer itself is doing fine - the last quarter beat on stronger underwriting and investment income, premiums grew 3.6%, the buyback finished, and analysts still carry an average target near - but a position that has spent five sessions without a single constructive signal while trading under its stop is a risk problem, not a valuation debate. Valuation is neutral here (essentially at fair value) so it neither adds to nor softens the case; we take 89% off and let the small remainder run. OPERATIONAL: carried and repriced from the unexecuted 2026-08-27 card; L2 basis established on the 08-27 decisive -0.65 ATR break, tonight -0.39 ATR below the pre-cut line, so the exit stands under the no-flip-flop bar (contrast DVA 08-27, withdrawn because it was never under its stop). NoTrade|sell bias +0 (n=45). val_adj 0, size x1.0, no tier bump (MoS -0.006 far from -3). CANCEL CONDITION: if CB opens above on 08-31, cut the order to 50% or withdraw. EXECUTION: work the print via VWAP/participation - 08-31 is MSCI August review effective at the close plus month-end rebalance flows. Residual stop on. Book-wide executor blocker: input/usbhportfolio.json unchanged since 08-18, so qty_at_decision is the live 16,377. · news
DALDelta Air Lines Inc78.00SELL ALL80.0781.3680.07 (+0.0%)n/a70.33 (-12.2%)Delta is being sold outright because the shares broke decisively below the stop we set and the twelve-week outlook has turned firmly negative, around -12%. The stock has now gone four straight sessions without the model finding anything constructive to say about it, it is down almost 4% over the past month while the market rose the same amount, and recent coverage has centred on insider selling. The position is 13.7% under water. A cheap-looking discounted-cash-flow value does not rescue a name that keeps making lower lows - we respect the stop and step aside rather than average into weakness. OPERATIONAL: L2/L3/US-1 - close vs stop = -1.59% / -0.57 ATR, versus the -0.10 ATR graze correctly declined on 08-27, so the break is now 5.7x deeper; four consecutive NO TRADE prints; RSI 35.4; unrealised -vs cost. Full exit is doctrinal for a decisive stop break. NoTrade|sell bias +0. DCF sits 46% above the close - valuation is explicitly barred (CLAUDE.md section 9) from vetoing a risk exit. ADR-0020 checked and NOT fired: stop vs peak stop = 9.74% erosion, under the 15% bar - this is signal-based, not risk-discipline. No earnings inside the 10-day horizon. Stop retired on close-out. · news
AAPLApple Inc316.85SELL 30%319.70307.93331.77 (+3.8%)359.94 (+12.6%)319.55 (-0.0%)We are banking a third of the Apple position because it has already captured the entire twelve-week gain the model expected of it - the target and the current price are now the same number, meaning the model sees no further upside from here over that horizon. The business news is good: Apple just raised Apple TV and Apple One subscription prices, a clean addition to services revenue, and iPhone sales are up 22% year on year on strong iPhone 17 demand. But John Ternus becomes chief executive on September 1, the first leadership change in fifteen years, landing in the exact week we would otherwise be holding the full stake. Taking 30% off and raising the stop on the remaining 70% keeps most of the upside while removing the part of the position the model no longer justifies. OPERATIONAL: ADR-0012 profit-take, armed + 1 tick. ARMED genuinely - T12_Progress 100.05 on +3.59% / +1.7 ATR (cost, T12, close = 101% of the modelled move). FIRED on tick (c) only: Target_12Week < 1.02 x Close. Ticks (a)/(b)/(d) did not fire (PROFIT_ARMED is the sole warning code; conviction, i.e. rising; one NO TRADE, not two). G1 inapplicable (AnticipationUp_VCP, not a fresh BreakoutUp_*/MomentumContinuation_*) and tick (c) fires regardless of G1 per ADR-0012; G2/G3 govern only the valuation-elevate path, and G3 is satisfied anyway (forward view flat, not solidly positive). MoS -2.788 amplifies but MISSES the <=-3 tier bump by 0.21, so portion stays 30%, not 50%. AnticipationUp_VCP|sell n=3 = report-only, no bias applied. Counterfactual evidence: optional_trim n=125, median_wh_4w +2.54 [cost] - untaken trims have been costly, which is why this is a mandatory card and not prose. WARNING 12wk exceeds the DCF anchor - projection assumes growth beyond priced-in. Estimates use f=0.30 (calib_1w -0.086, degenerate, L5/US-3). Entry range - (+/-0.25 ATR). STOP RAISED -> on the retained. · news
NUENucor Corp249.64SELL 50%250.50248.53250.50 (+0.0%)n/a254.11 (+1.4%)Half the Nucor holding is being sold because the position has run 13.5% above cost and the model's three-month target has now collapsed to essentially where the shares already trade - there is no expected return left to hold for. The name has also gone four straight sessions without a constructive signal. This doubles the size of the trim we flagged a session ago, when the forward-view test missed firing by twelve cents; it has now firmly failed. OPERATIONAL: re-issue and reprice of the still-unexecuted 08-27 card (30% @) - NOT a fourth new decision; Step 7 queue-reconcile. Case strengthened: tick (c) now fires (Target_12Week < 1.02 x Close) alongside tick (d) (four consecutive NO TRADE) = 2 ticks -> profit_take_2 = 50%. Armed genuinely (Gain_ATR 3.50, +13.48% on cost). MoS -2.278 amplifies but misses the <=-3 tier, so no bump to 100%. Stop held on the 139-share remainder.
KHCKraft Heinz Co25.67NO ACTION25.7024.1525.70 (+0.0%)n/a25.72 (+0.1%)Kraft Heinz cleared the bar for a half-position trim - the model's three-month target sits essentially at today's price, so there is no expected return left, and it has gone four sessions without a constructive signal. It did not make tonight's three-action limit because the 'profit' being banked is only 1.3%, economically trivial next to the larger positions ahead of it. It is first in the queue for the next session. OPERATIONAL: ADR-0012 ticks (c) T12 < 1.02 x close AND (d) four consecutive NO TRADE = 2 ticks -> profit_take_2 = 50%; @ = Armed only via T12_Progress 99.92 on +1.30% / +0.48 ATR - valid but thin. Lost the slot to AAPL on banked-dollar terms. Ledgered optional=true so the counterfactual grader prices the deferral; NOT prose-only advice. Stop held.
MCOMoody's Corp504.64NO ACTION514.95493.86521.60 (+1.3%)537.11 (+4.3%)571.37 (+11.0%)Moody's triggered a 30% profit-trim on a momentum-divergence warning - the price is holding up while the underlying momentum reading weakens, a classic early caution on a position already 5.4% ahead. It lost tonight's slot because it has the healthiest forward view of any flagged name, roughly +11% over three months, making it the least urgent of the exits. OPERATIONAL: ADR-0012 tick (a) - RSI_DIV printed alongside PROFIT_ARMED = 1 tick -> profit_take_1 = 30%; @ = G1 does not shelter it (signal is FailedBreakdown_Up_20D cv47, not a fresh BreakoutUp/MomentumContinuation at cv>=80). G2 DOES suppress valuation amplification: Target_12Week vastly exceeds DCF_per_share, so the DCF anchor is unreliable and the MoS -1.148 reading must not amplify. Ledgered optional=true for counterfactual grading. Stop held (scan is lower). 1wk uses f=0.30.
COFCapital One Financial Corp214.51NO ACTION215.67208.23215.67 (+0.0%)n/a231.81 (+7.5%)Capital One qualified for a small 30% trim on the back of two consecutive quiet sessions from the model on a position already 5.1% ahead. It is the weakest of tonight's flagged exits - the three-month view is still a constructive +7.5% - so it did not make the three-action cut. OPERATIONAL: ADR-0012 tick (d) only (two consecutive NO TRADE) = 1 tick -> profit_take_1 = 30%; @ = Armed via T12_Progress 93.04 on +5.14%. Ledgered optional=true for counterfactual grading. Stop held.
MMacy's Inc21.96NO ACTION22.0120.6022.01 (+0.0%)n/a19.26 (-12.5%)Macy's broke below its stop by the widest margin of any holding tonight and its three-month outlook is around -12.5%, which is a clear exit case. It did not make the three-action limit purely on size - the entire position is worth against a book - so spending a slot here rather than on a decision would be poor allocation. It should be cleared before the September 10 results. OPERATIONAL: L2 - close vs stop = -2.13% / -0.68 ATR, the deepest ATR breach in the book, plus 12wk -12.49% and a STRONG SELL - REDUCE cv61 at Sell_Rank 2. NOTE the label originates on RallySell_BearTrend, tonight's worst cell (bias -8, hit_1w 0.091, n=12) - adj conviction 61-8 = 53; the exit survives on the INDEPENDENT stop-break basis, not on the label. ADR-0020 not fired (stop vs peak = 3.68% erosion). Ledgered optional=true for counterfactual grading. Stop held. Berkshire has been adding to M - noted as context for a mirror book, not a signal input. · news
STZConstellation Brands Cl A130.62NO ACTION130.69128.36130.69 (+0.0%)n/a114.77 (-12.2%)Constellation Brands printed the book's strongest sell label tonight with a three-month view around -12%, but the entire holding is worth and the pattern behind the label has been right once in twelve tries. It is the weakest case on the board and does not warrant one of three actions. OPERATIONAL: STRONG SELL - REDUCE cv63 at Sell_Rank 1, portion 88.52% = @ = Originates purely on RallySell_BearTrend (bias -8, hit_1w 0.091, n=12) with NO independent stop break - close is above the trail. Adjusted conviction 63-8 = 55, under the floor. Quality_Pass False. Ledgered optional=true for counterfactual grading. Stop held (scan is lower).
No acquisitionNO ACTIONn/an/an/aNo acquisition tonight. Not one of the twenty-nine names in this book produced a buy signal of any kind on the close - the universe is entirely holds, cautions and sells. With no candidate clearing the bar, the correct answer is to buy nothing and let cash build. OPERATIONAL: zero STRONG BUY / BUY / BUY - ADD / BUY - ANTICIPATE rows in the 08-28 snapshot (11 NO TRADE, 6 HOLD, 5 AVOID, 2 STRONG SELL - REDUCE, plus unheld LPX/JEF both AVOID). No ANTICIPATE slot either - no name printed BUY - ANTICIPATE in >=2 of the last 3 snapshots. Highest held conviction is AAPL cv83, which is a DISPOSE tonight; next are CVX 62 and KO 62, both under the 65 floor. (49.6% of NAV) untouched; cash rises to ~56.8% after tonight's sells. Fifth consecutive session with no buy label - see systemic finding 6.
AXPAmerican Express Co330.17HOLD330.57n/an/a312.27Holding American Express. It is the book's largest position and its softest - down 5.3% against cost with a negative three-month view - but the shares are still above their stop and the model is not calling for action. We watch rather than sell. OPERATIONAL: dormant sell case, 4 consecutive AVOID prints, 12wk = -6.3%. NOT armed (PROFIT_ARMED print is spurious - see systemic finding 1; position is -5.33% / -3.68 ATR). Scan Stop_Price prints ABOVE the close and is REJECTED (6th consecutive session); trail held at.
BACBank of America Corp61.94HOLD61.28n/an/a66.20Holding Bank of America. It was the strongest contributor on the session, up 1.9%, the model's confidence has recovered sharply from a brief scare, and the three-month view is a constructive +6%. Nothing to do. OPERATIONAL: conviction, recovering; ARMED via T12_Progress 94.14 on +3.64% but NO tick fired -> trail-raise only. Scan Stop_Price prints ABOVE the close and is REJECTED; trail held at.
GOOGLAlphabet Inc Cl A339.35HOLD333.68n/an/a310.41Holding Alphabet. The model has nothing to say about it either way and the shares are comfortably above their stop; the position is 3.7% under cost. Goes ex-dividend on September 4. OPERATIONAL: 4 consecutive NO TRADE, 'Consensus conflict + low conviction'. The stale news overlay (WEEK 2026-07-05, 55 days old) carries a standing override invalidated below - close is above the line, so technically live but NOT acted on given staleness. NOT armed (spurious PROFIT_ARMED, position -3.70%). Stop held.
OXYOccidental Petroleum Corp60.18HOLD - trail57.90n/a61.7864.60Holding Occidental with the stop kept tight. It is the book's best performer at 20.8% above cost and the three-month view is still positive, so there is no reason to bank it yet. OPERATIONAL: ARMED (Gain_ATR 7.60) but NOT fired - conviction drop is only 6 pts, under the 20-pt bar; 12wk = +9.3%, well clear of tick (c) -> trail-raise only. Trail held at (scan is lower).
VRSNVeriSign Inc288.28HOLD - trail280.81n/an/a309.35Holding VeriSign, 13.8% ahead, with the trailing stop unchanged. The model went quiet on it after three constructive sessions, which on its own is not a reason to sell a winner with a positive three-month view. OPERATIONAL: interpretation recorded - the HOLD cv68 -> NO TRADE cv0 flip is NOT treated as ADR-0012 tick (b). NoTrade rows print cv 0.0 by construction; firing (b) on them would make tick (d) (which deliberately requires TWO consecutive NO TRADE prints) dead letter and would trim on the first quiet session. NO TRADE deterioration is governed by tick (d) only - consistent with the 08-26/08-27 runs, where KO fired on its THIRD consecutive NO TRADE, not its first. Applied identically to COF/KHC/NUE tonight. ARMED (Gain_ATR 4.77), not fired; 12wk = +6.0%. Trail held at.
DVADaVita Inc176.01HOLD169.18n/an/a113.62Holding DaVita on its restored stop. It is the book's worst position at 23% under cost and the model's three-month projection is severe, but the shares are still well above the stop level and the sell label that appeared on 08-27 has since evaporated - we do not chase a signal that un-fired. OPERATIONAL: STRONG SELL cv79 (08-27) un-fired to NO TRADE cv0 and stayed there; card was withdrawn 08-27 and the stop restored. NOT armed despite the PROFIT_ARMED print - position is -23.09% / -11.07 ATR, T12_Progress 159.02 is the degenerate-ratio artifact described in systemic finding 1. MoS -3.376 is the most extreme in the book but valuation cannot originate a sell (CLAUDE.md section 9) and there is no Exit_Warning code to elevate. Stop held.
SIRISirius XM Holdings Inc28.17HOLD27.81n/an/a26.08Holding Sirius XM. Quiet from the model for four sessions and 6.5% under cost, but still above its stop, so no action. OPERATIONAL: 4 consecutive NO TRADE. NOT armed (spurious PROFIT_ARMED; -6.46% / -3.55 ATR). Quality_Pass False. 12wk = -8.6% - worth watching for an L2 setup if the stop goes. Stop held.
GOOGAlphabet Inc Cl C335.41HOLD320.20n/an/a303.76Holding the small Alphabet C-share stake alongside the larger A-share position; same read, no action. OPERATIONAL: 4 consecutive NO TRADE; mirrors GOOGL. Ex-dividend 09-04. Stop held.
LLYVKLiberty Live Group Ser C103.52HOLD103.21n/a108.56105.28Holding Liberty Live C. Mildly constructive but below the confidence level needed to add, and 3.4% under cost. No action. OPERATIONAL: conviction 59, under the 65 floor. Valuation NA (tracking stub) -> val_adj 0, quality gate off. Stop held (scan is lower).
ALLYAlly Financial Inc42.21HOLD41.56n/an/a36.71Holding Ally Financial. It screens as the cheapest name in the book on intrinsic value, but the model has been silent on it for four sessions and it sits 6.8% under cost - cheap is not a buy signal without a trigger. Above its stop, so we wait. OPERATIONAL: 4 consecutive NO TRADE; UNDERVALUED MoS +0.075 but Quality_Pass FALSE, so a new entry would be blocked by the quality gate anyway (ADDs to held names are never blocked). NOT armed (spurious PROFIT_ARMED; -6.83%). 12wk = -13.0%, the weakest forward view among the quiet names - watch for an L2 setup below. Stop held.
NYTNew York Times Co Cl A67.27HOLD64.15n/an/a54.53Holding the New York Times. A dormant sell case - four straight cautionary readings and a weak three-month view - but the shares remain above the stop and the position is small. OPERATIONAL: 4 consecutive AVOID; 12wk = -20.0%, the worst in the book. Scan Stop_Price prints ABOVE the close and is REJECTED; trail held at. Becomes an L2 candidate the moment breaks.

Outcome & track record

Accuracy at the time of this record.

145 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.

This website is for educational purposes only and does not constitute investment advice. Always do your own research and assess your own risk tolerance before making investment decisions.