Mon-T Weekly Review — w/e 25 Sept 2026

Something shifted this week. Not in the markets, which continued their volatile dance between AI euphoria and yield terror, but in the desk itself. Five directional calls, four correct, an 80% accuracy rate that matches the best performances of the year. The Nasdaq, this week's Market of the Week, was called BULLISH at 7/10 and surged 4.33% as the S&P Global Flash PMI landed at 58.4, the strongest composite reading in five years and a blowout that left the 55.2 consensus estimate looking like a bad guess at a pub quiz. CNBC confirmed the S&P 500 climbed 0.51% to close at 7,743.41 on Friday, while Yahoo Finance reported the 10-year yield touching 5.18%, its highest since the global financial crisis.

The desk committed where it had conviction and delivered. Copper BULLISH at 7/10 gained 2.39%. Treasury bonds BEARISH at 6/10 fell 2.59% as the yield surge continued its relentless march. The Russell 2000 BEARISH at 6/10 caught the right side with a modest 0.43% decline. The sole directional miss was soybeans, BEARISH at 6/10, which rallied 1.27% as Chinese buying overwhelmed the desk's positioning-unwind thesis. One miss from five is the kind of week that rebuilds trust after the 33.3% debacle I was grading a fortnight ago.

The ten NO CALL markets tell the familiar story I have been writing since March, though this week the paragraph writes itself with particular force. Gold dropped 2.21%. Silver fell 2.71%. Crude oil slid 2.89%. EUR/USD slipped 0.56%. Sterling dropped 1.04%. The Aussie crashed 1.47%. The desk said nothing about any of them. I wrote last week that central bank weeks are where conviction goes to die. This week, the post-FOMC clarity should have given the agents something to work with. On equities and bonds, it did. On everything else, the silence continued.

Five directional calls this week, with four landing on the right side. The other ten markets got the NO CALL treatment. An 80% directional accuracy rate is the desk's best since the w/e 14 Aug week that delivered 83.3%, and the average confidence of 6.4 across those five calls represents genuine conviction rather than the whispered 5.3 of the summer lull.

The calibration was clean in a way that the last few weeks were decidedly not. The two highest-conviction calls at 7/10, NQ BULLISH and HG BULLISH, both delivered emphatically with 4.33% and 2.39% respectively. The 6/10 calls split two correct (ZB, RTY) and one missed (ZS). When your strongest conviction produces your best results and your only miss comes from moderate conviction, the system is telling you something useful about where its genuine edge lives. After the 33.3% horror show two weeks ago and the 42.9% the week before that, four from five feels less like normal service and more like rehabilitation.

The rolling twelve-week figure sits at 53.7% across 67 directional calls, with 110 no-call abstentions. That engagement split means the desk calls direction on roughly 38% of market-weeks, a rate that has been declining since February's 70% pace and is now at a level where the NO CALL column is doing nearly three times the work of the directional one. This week's 80% on five calls provides a modest lift, but with only five calls entering the denominator, the needle barely moves. The persistent mountain of abstentions means the rolling number is hostage to bad weeks that happened months ago. To crack meaningfully above 57%, the desk needs to increase its directional volume while maintaining accuracy above 65%. At the current pace of five calls per week, that remains aspirational.

The Sentiment agent had its best week in months, and the timing could not be more appropriate. Its identification of the AAII -24.5% bull-bear spread on the Nasdaq as a textbook contrarian buying signal was the analytical foundation of the week's best call. After years of this column documenting the Sentiment agent being 'too cautious' or 'offering no useful signal,' it earned its weighting this week by reading retail capitulation correctly while institutional money was doing the opposite.

The Economic agent also performed well, correctly supporting the BULLISH equity thesis through its RISK-ON regime classification and providing the macro backdrop for the bearish bond call that delivered 2.59% of downside. The Institutional agent earned genuine credit on both NQ, where it read the 61% COT surge as trend-following conviction rather than crowding, and on ZB, where its identification of extreme speculative shorts at the 3.2nd percentile correctly flagged the prevailing bearish pressure. The Fundamental agent was the week's mixed performer, driving the correct BULLISH call on copper through the El Nino supply disruption thesis while simultaneously supporting the soybeans BEARISH call that missed. Same agent, same structural supply-demand framework, opposite outcomes depending on which demand catalyst the market cared about.

The calendar thins into early October, with the next FOMC not until late October and Q3 earnings season the dominant narrative. The PMI data that powered this week's NQ rally has shifted the conversation from 'will the economy survive higher rates' to 'is the economy accelerating despite them,' which is a fundamentally different question for every asset class. Gold at $4,327, having fallen 2.21% while the desk watched from behind the signal threshold, faces continued headwinds from the yield surge that pushed the 10-year to 5.18% on Friday. Crude oil's quiet -2.89% decline suggests the de-escalation thesis the desk has been tracking on Hormuz may be gaining traction again. And the Nasdaq at 30,896, sitting at new 52-week highs, must now contend with whether the PMI acceleration is already priced or whether Q3 earnings provide fresh fuel. The desk will have its Sunday views. Given this week's results, I expect the directional conviction on equities to persist.