Mon-T Weekly Review — w/e 18 Sept 2026

Kevin Warsh hiked rates. I know, I know, you already knew that. By Tuesday afternoon, CME FedWatch was pricing a 92% probability of the 25bp move to 3.75-4.00%, and by the time the announcement hit Wednesday at 2pm, the only people surprised were those who had been on a silent retreat since August. The unanimous vote and the dot plot showing 16 of 18 officials wanting more tightening were the real story. GoldSilver reported gold gave back its morning gains within 30 minutes of the announcement. But by Friday, gold had recovered to $4,417, because apparently the market decided a hawkish hike was good news if you squinted hard enough.

The desk made six directional calls into this catalyst inferno and got two right. Two. Out of six. A 33.3% accuracy rate, the worst on the books since that 12.5% nightmare back in April that still gives me involuntary twitches. The Japanese yen, called BULLISH at the desk's joint-highest conviction of 7/10 on a thesis about the BOJ September 17-18 hike, dropped 1.21% as the anticipated rate increase failed to deliver the yen strength the agents expected. Gold was called BEARISH at 5/10 and rose 1.16%, which is the market's way of saying your hawkish thesis was priced in three days before you published it. Wheat went the wrong way. Soybeans went the wrong way. The only markets that cooperated were Treasury bonds and copper, and one of those barely moved.

The nine NO CALL markets included crude oil dropping 4.64%, silver surging 3.6%, the Nasdaq rallying 1.91%, and sterling falling 0.91%. The desk said nothing about any of them. I have been writing some variation of this paragraph since March, and at this point the paragraph has developed its own career trajectory and is considering retirement.

Six directional calls this week, with two landing on the right side. The other nine markets got the NO CALL treatment. A 33.3% directional accuracy rate is the desk's worst since the April catastrophe, and it arrives during the most catalyst-dense week of September. The average confidence of 6.2 across those six calls tells you the desk was not whispering this time. It committed with genuine conviction and got demolished.

The calibration is particularly painful this week. The highest-conviction call, the yen at 7/10, missed by 1.21%. The two correct calls came from bonds at 7/10, which delivered a modest 0.32% decline, and copper at 6/10, which surged 3.89%. When your best result comes from a market the desk quietly tracks in the Extended tier rather than the MOTW showcase, the framework is telling you where its genuine edge lives. The FOMC binary outcome resolved against four of six directional positions, which is what happens when you position ahead of a central bank meeting that was already 92% priced by Tuesday.

The rolling twelve-week figure sits at 51.5% across 66 directional calls, with 109 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 approaching the level where the NO CALL column is doing roughly three times the work of the directional one. This week's 33.3% drags the rolling number below the 53% line that had served as a floor for months. The desk needs a string of clean weeks above 65% to recover, and the September-October calendar density should provide the opportunity if the agents can translate catalyst clarity into directional accuracy.

The Fundamental agent had a split week that continues its year-long identity crisis. On copper, its tariff-driven supply squeeze thesis and structural deficit framework produced the week's only impressive win at +3.89%. On gold, the same agent called BEARISH on overvaluation versus real yield models, and the market went up. On soybeans, its BEARISH thesis about the WASDE supply-side surprise met a market that rallied 1.78% anyway. Same agent, same week, different outcomes depending on which market the macro regime cooperated with.

The Economic agent had a poor week across the board. Its hawkish FOMC thesis drove bearish calls on gold and soybeans that both missed, and its bullish yen thesis about BOJ policy divergence compression failed to deliver the expected currency strength. When the most heavily weighted discipline (25-30% across multiple markets) drives the wrong direction on four of six calls, the synthesis framework needs to reconsider whether central bank meeting weeks demand a fundamentally different approach to conviction. The Institutional agent was the week's quiet casualty, with its extreme COT readings on both the yen (+103K swing) and soybeans (99.4th percentile) both failing to translate into correct directional calls.

The FOMC aftermath will dominate the coming week as markets digest the first rate hike since 2023, the hawkish dot plot showing 16 of 18 officials wanting further tightening, and whatever Chair Warsh communicated about the path forward. The BOJ delivered its September 17-18 decision within this week's grading window, and the yen's reaction suggests the hike was priced, raising the question of whether the desk's bullish thesis has run its course or simply needs a different catalyst. Gold at $4,417, having survived the FOMC hike with a rally, faces the question of whether the 'sell the rumour, buy the fact' bounce has legs or is a one-week respite before the next hawkish repricing. Crude oil's 4.64% decline on a NO CALL is worth watching as the Hormuz premium continues its seasonal ebb. The desk will have its Sunday views. Given what the FOMC just did to four of six directional calls, I expect some serious recalibration.