Agent bulletins

The Leading trio is Energy (+12.61 pp over 3m), Health Care (+15.69) and Materials (+1.83) — literally the late-cycle signature from the historical map, with Consumer Staples already working in Weakening. On the other side, Industrials, Utilities and Consumer Discretionary sit in Lagging, and not one early-cycle sector is accelerating. But coherence failed: only 3 of the 9 confirmation ratios support the late phase, and two rotations on the map have an identified non-cyclical cause — Energy from the war, Utilities from California wildfire legislation.

Forecast: the late phase only confirms if XLK and XLC fail to leave Improving and fall back to Lagging, with XLY/XLP making a new three-month low. Trigger: XLK turning negative on 1-month relative strength against SPY. Invalidation: XLK and XLC entering Leading while XLE exits — in which case the read becomes mid-cycle and this agent was wrong.

The rally is being carried by a few names — but not the usual few. 70% of members are still above their 200-day average, which is still a bull market; the problem is the second derivative. Above the 50-day fell from 70% in mid-August to 54% on 28 August, the McClellan oscillator turned negative right after the SPX tested 7,800 and has not come back, and new 52-week highs collapsed to 3–4% on the week and 1% on Friday, against the 10–12% of a healthy market. The NYSE advance-decline line returned to its 50-day average, and the mid- and small-cap lines are already below theirs.

Forecast: participation keeps narrowing while the index holds. Concentration risk is medium, not high — RSP beats SPY over 3m, meaning it is the mega caps failing, not the troops. Escalation trigger: share above the 50-day below 50%. Invalidation: new highs back above 8% of members for three consecutive sessions.

The rate environment is hostile and the credit environment is benign, and that is not a contradiction — it is the portrait of a supply shock rather than a demand one. Fed funds at 3.50–3.75%, core PCE at 3.3% and headline at 3.7% above forecast, July payrolls at −23k, and the market pricing 67% hold against 33% hike for 16 September, with a cut effectively at zero. The 10-year at 4.796% is the highest since January 2025, and 2s10s at +0.41 steepened from the long end — the inflationary shape. Crude +5.2% on the day from the escalation with Iran. On the other side, HY OAS at 2.63% has been compressing since 3.03% in March.

Liquidity 🔴 · Credit 🟢 · Curve: positive at +0.41, steepening from the long end. Forecast: the environment stays hostile through 11 September. Deterioration trigger: HY OAS above 3.00% or core CPI at 0.3% month over month. Invalidation: crude below $80 with the 10-year below 4.60% returns the environment to neutral.

Price is being paid for earnings that exist. Q2 2026 closed at 50.4% growth, the strongest since Q2 2021, with 86% of companies beating estimates against a 78% five-year average — and even stripping out the investment gains at Alphabet and Amazon, which inflate the figure, growth lands at 32.0%. The forward P/E is 20.0 against a 19.9 five-year and 19.0 ten-year average: stretched, not extreme. The return decomposition since 30 June is the most important line in this bulletin: SPY rose 2.01% while the multiple contracted from 20.4 to 20.0 — roughly +4.1% from earnings against −2.0% from multiple. A rally made of earnings is the sustainable kind.

Buyback window: open now, blackout opens between 8 and 15 September. Forecast: revisions stay positive, with Q3 projected at +27.4% and Q4 at +25.2%. Reversal trigger: revision breadth turning negative in the four weeks before the October season. Invalidation: forward P/E above 21.5 without a matching rise in EPS — at which point the rally is multiple again and this score falls.

No opinion, only distribution. SPY closed at 761.78: below its 10-day at 766.21 and 20-day at 769.21, above its 50-day at 754.71 (+0.94%) and 100-day at 742.83 (+2.55%), with the 50-day still sloping up (+1.18% over 20 sessions) and 2.07% below the 100-day high. Regime: medium-term uptrend with short-term mean reversion. Volatility is the anomalous reading — 20-day realised at 7.2% against a VIX of 16.34, a variance premium of roughly 9 points, with VIX3M at 18.33, IVTS at 0.891 and contango on day 102. 20-day ATR at 0.68% of price.

Highest-probability scenario (43%): range between 745 and 778 into the 16–18 September block. Bearish trigger: a close below 754.71 followed by a loss of 745. Bullish trigger: reclaiming 769.21 and closing above 777.88. Missing data note: implied correlation across stocks and the equity-bond correlation were not obtained this round.

Whoever was in control stopped being in control. The markup from April to August carried SPY from 679.46 to 777.88 on 13 August, with the SPX testing 7,800 for the first time — volume and breadth at the climax point to a buying climax. The automatic reaction followed, to 762.60 on 20 August, then the attempt to reclaim on 27–28 August failed at 771.10 and 769.35 without taking out the high: a failed upthrust. We are in phase B distribution at index level. By sector the read splits, and that is rotation information rather than contradiction: XLV is still in late markup while SOXX is clearly in markdown, from 655.01 on 22 June to 500.31 now.

Control: neutral tilting to sellers. Key event: buying climax on 13 August, failed upthrust on 27–28 August. Forecast: phase B extends into the event block; a sign of weakness would confirm the move to phase C. Trigger: a loss of 757.67 on rising volume. Invalidation: a close above 777.88 turns the structure into re-accumulation, and this agent was reading distribution where there was only a pause.

What people say and what they did sit on opposite sides, and the protocol says go with what they did. In the surveys, AAII on 26 August printed 32.9% bullish against 44.4% bearish, a spread of −11.5, deteriorating for four straight weeks from −1.0. In the derivatives, the opposite: total put/call at 0.84, the 9-day average at 0.83 in the 15th percentile of its recent range, equity put/call at 0.70, VIX at 16.34 with contango on day 102. Retail is verbally pessimistic while the desk is operationally unhedged. Sentiment extremes only work as contrary signals when price confirms, and price does not — the index is 2.2% off its high.

Survey vs positioning: divergent. Long squeeze risk: medium-high. Short squeeze risk: low-medium. Trigger: a 9-day put/call below 0.80 going into the FOMC raises liquidation risk. Missing this round: NAAIM, CFTC COT, dealer gamma and the gamma flip level, insiders and short interest.

We are in month 8 of year 2 of the presidential cycle — November 2024 election, 20 January 2025 inauguration, midterms on 4 November 2026 — and year 2 is weak in the first half and strong from October, historically the best entry point of the four years. That turn is 30 days out. Before it comes September, the worst average month for the S&P, with a rare cluster: CPI on 11 September, FOMC on 16 September and quadruple witching with the quarterly S&P rebalance on 18 September, all inside six sessions. Equinox on 22–23 September, quarter end on 30 September, and the US mutual fund fiscal year end on 31 October. The semiannual Russell reconstitution falls at the close on 11 December.

Time condition: late expansion turning to distribution. Next window: 11–18 September. Forecast: maximum time vulnerability that week, with the window turning constructive from late October. Protocol rule: an important date does not automatically mean a reversal — the window only counts when price and breadth confirm it. Partial data: the ~40-month Kitchin cycle low could not be dated from the 100-session series available.

The advance from 679.46 on 10 April to 777.88 on 13 August has impulsive five-wave structure. Primary count (50%): the impulse completed at 777.88, and the decline since is wave (4) or the A-B-C correction of that impulse, targeting 740–745, where the 0.382 retracement of the advance sits. Alternate (32%): the pullback since mid-August is wave 4 of an extending impulse, with wave 5 aiming at 795–805 in October. Third count (18%): the August high ended a larger degree and began a multi-month correction toward 700–710.

Current wave: corrective (4)/(A) of the April impulse. Primary invalidation: a close above 777.88. Alternate invalidation: a loss of 729.46, the 29 July low, which would violate the overlap rule with wave 1. No count is presented as a certainty.