Goldman's post-Jackson Hole hike odds outrun its own forecast for a September hold

Two of Goldman's published numbers point in different directions. Take the midpoint of the 50-60% hike probability the firm attached to September after Jackson Hole, call it 55, and the implied odds of no change are 45, which makes the firm's own base case the less likely of the two branches by ten points on its own arithmetic ^[5][7][1]. At the bottom of that range it is a coin flip and the forecast is a tiebreak. The inflation figure doing the work is a core print running near 0.2% in late August, which compounds to roughly 2.4% annualised, and that is the kind of number a committee sits through without owing anyone an explanation ^[8][2].

The dot plot being demoted has been the bond market's reference for something over a decade, by the same account, which makes the instrument Warsh is retiring younger than the statement length he has gone back to ^[2][3][3]. Forward guidance functioned as a transfer of information, from the committee to anyone whose positioning was built on it, and withdrawing it does not change the path of the funds rate so much as it changes who eats the forecast error and who gets paid for shrinking it. Hatzius, who says the premium on reading raw releases has gone up and that models built on guidance now run on a thinner dataset, works at a firm that sells exactly that reading ^[1][10]. Goldman has left its rate call unmoved despite the volatility it describes, which amounts to telling clients to fade the probability the same house published ^[7].

This could run a few ways. Hatzius is right and asset prices whip on phrasing rather than fundamentals, which is his stated second risk, with the first being real-economy damage from assets priced off a misread committee ^[9]. Or, the more interesting version, surprise simply relocates from Fed days to CPI and payroll days, and total realised volatility barely moves; the source material carries no volatility measurements either way, so nothing here settles that. Or less guidance breaks the loop in which the Fed follows the curve its own dots created, which is close to Warsh's stated intent of making markets trade fundamentals instead of decoding the committee ^[4].

The test is cheap and arrives soon. If the Fed holds in September with core at 0.2% and long yields hand back the Jackson Hole move without disorder, the word "unproductive" is doing rhetorical work rather than describing anything, and Hatzius himself frames this as a lower signal-to-noise market rather than a disaster ^[5][8][12]. The thing to measure is the split: bigger moves on data releases, smaller moves on statement days, and a total that has not grown. Hatzius's forecast is not for a worse economy but for a market in which being right about the economy pays better than being right about the Fed, and that is a repricing of research, not of risk ^[12]. Cryptobriefing's extension of the argument to bitcoin rests on a correlation with rate expectations it asserts but does not quantify ^[11].