A plain-English guide to a Chair keynote flipping the distribution.

Today's Friday August 28 Warsh keynote at Jackson Hole did something the market treats as rare: a single Fed Chair communication flipped the probability distribution for the next FOMC from a 65 percent dovish-cut base case to a 45 percent hawkish-hike base case. That is a 110-percentage-point swing across the modal outcomes in one session. It happens roughly once every 12 to 18 months on a global basis, and when it does it produces a specific set of tape signatures, a specific set of second-order asset moves over the following one to four weeks, and a specific set of framework re-calibrations for the accounts that had built exposure against the prior distribution. This piece is the framework for reading a distribution-flip session and for updating positioning frameworks in its wake.

What "flipping the distribution" means

Two states get confused and are worth separating. A surprise where the market was priced for one modal outcome and got the modal outcome plus more (a hawkish hike where the market expected a hawkish hold) is a magnitude-surprise; it moves price and OIS but does not restructure the distribution. A surprise where the market was priced for one modal outcome and got a different modal outcome (a hawkish hold where the market expected a dovish cut) is a distribution-flip; it moves price and OIS more, restructures the strike-density of options positioning, and forces a re-calibration of the framework the market has been running for the last several weeks.

The Warsh keynote is a distribution-flip in the pure form. The market had been priced for a cut with roughly 65 percent probability, a hold at 25 percent, and a hike at under 5 percent. The keynote language moved the modal outcome from cut to hold-with-hawkish-language and put a hike at 45 percent. The modal outcome changed. That is the distribution-flip signature.

The tape signatures on a flip session

1. Cross-asset amplification well above the eve-gamma projection. The keynote-eve compression framework calls for the amplification case to run in 55 percent of keynote-day tapes and to move price 40-60 percent above the fundamental-flow projection. A distribution-flip runs at 100-150 percent above the projection. Today's DXY closed 55 pips higher on a keynote where a 25-30 pip amplification was the framework base case; that is amplification-plus-flip.
2. Front-end leads the belly and long-end. On a distribution-flip the 2-year yield moves 2 to 3 times as much as the 10-year, because the flip is a reaction-function change and the reaction function is priced in the front end. Today's 12.4bp 2-year move versus 5.5bp 10-year move is a 2.3-to-1 ratio, right in the middle of the distribution-flip band.
3. Cross-asset consistency across all pairs. A magnitude-surprise moves one or two asset classes and leaves others; a distribution-flip moves gold, rates, dollar, cross-JPY, and cross-EUR all in the same direction inside two hours. Today's tape moved all five in the hawkish-Fed direction inside the 10:00-12:00 PM ET window.
4. Positioning-driven overshoots in the first 60 minutes. The pre-flip crowd that has to unwind its dovish positioning creates an initial 60-minute overshoot in the direction of the flip. The overshoot typically retraces 20 to 40 percent inside the first 90 minutes as the second wave of accounts enters at more attractive levels. On today's tape, DXY hit 99.48 at 10:47 AM ET and closed the session at 99.34, a 14-pip retrace of the 74-pip amplified move, right in the middle of the 20-40 percent retrace band.
5. Options implied vol collapses even as realised extends. The one-day ATM implied vol on the crosses that had been priced for the keynote collapses inside 30 minutes of the speech opening as the event risk is removed. The 24-hour and 48-hour realised vol extend for several sessions as the flow follows through. This produces a specific footprint that can be measured: implied vol curves invert their normal upward slope for one to three sessions post-flip.

The four-week second-order pattern

A distribution-flip does not resolve on the flip session. The four-week pattern that follows is worth naming because it drives the calibration for the next several pieces of analysis.

1. Week 1: crowd unwind. The accounts that had built exposure against the prior distribution unwind their positions. This creates persistent directional flow in the direction of the flip. On today's flip, the persistent flow through the next five sessions is dollar-strength and gold-weakness.
2. Week 2: data-flow test. The market tests the flip against the incoming data. If the data corroborates the flip (in today's case, if NFP prints firm on Friday September 5), the flip consolidates and Week 3 opens with the new distribution as base case. If the data contradicts the flip (NFP prints soft), the market begins pricing the flip as an over-reaction and the retrace begins.
3. Week 3: distributional re-calibration. The accounts that were positioned against the flip either capitulate (extending the move) or rebuild positioning at more attractive levels (partial retrace). This is the week where the framework's new distribution firms up.
4. Week 4: next-catalyst focus. By four weeks in, the market's attention has moved to the next scheduled catalyst (in today's case, the September 17 FOMC). The flip is now baked in; what happens next is priced against the new distribution rather than the old.

The five re-calibration questions

A framework carrying probability weights that were wrong needs to fix the weights, not paper over the miss. Five questions to ask.

1. Was the modal outcome wrong or the probability weighting wrong? If the modal outcome was in the framework's distribution but weighted too low, the fix is a calibration change; if it was not in the distribution at all, the fix is a structural change to the framework's state space.
2. What input did the framework misread? On today's flip, the July PCE at consensus was read as removing the hawkish-tilt anchor. It did the opposite because the reaction function had changed under Warsh. The input was not wrong; the reaction function assumption was.
3. Were the cross-asset consistency checks confirming or discomfirming? The framework's cross-asset checks were confirming through Wednesday and Thursday. They were confirming a wrong distribution. That means the cross-asset checks work as designed and the input distribution is what needs the calibration change.
4. Is the flip a one-communication anomaly or a regime change? A Chair keynote that reads as a break from prior communication style is a regime change signal. Warsh's "quieter Fed" framing is regime-change language; the flip should be treated as durable rather than as a one-communication over-reaction.
5. What does the new distribution need to have priced correctly for the next catalyst? On today's tape, the new distribution needs to have priced correctly the September 17 FOMC. That means the hawkish-hike weight (45 percent) and the hawkish-hold weight (35 percent) both need to survive the next four weeks of data-flow. The data-flow calendar is stacked (JOLTS Tue, ADP Wed, ISM Services and Claims Thu, NFP Fri all next week), so the framework will be tested rapidly.

Base rates for a distribution-flip's follow-through

Historical base rates for how a distribution-flip tape resolves over the four weeks that follow:

1. Flip extends and consolidates (roughly 55 percent). The new distribution holds; the second-order asset moves continue for four to six weeks; the framework re-calibration is durable.
2. Flip fades over two to three weeks (roughly 30 percent). The next round of data flow contradicts the flip; the market partially retraces; the distribution settles somewhere between the pre-flip and flip states. This is the "over-reaction" resolution.
3. Flip fails and inverts (roughly 15 percent). The data flow strongly contradicts the flip; a subsequent Chair communication moderates the initial signal; the pre-flip distribution reasserts, sometimes with additional strength.

The base rate says the flip extends more often than not; the extensions are typically shallower than the initial move, and the fades that happen typically happen inside two weeks. The "flip fails and inverts" case is the one to watch specifically because it produces the largest single-week price moves when it happens, but it is also the rarest.

Related reading

• Today's analysis piece: the Warsh flip and the honest post-mortem on the prior thesis.
• Keynote-eve gamma compression: the Thursday-eve framework that fed into Friday's amplification-plus-flip signature.
• Reading a Jackson Hole speech: the six-dimension text framework the flip was priced against.
• End-of-week thesis missed post-mortem: the editorial model for writing through a thesis that did not survive contact with the data.
• Monetary policy frameworks: parent piece on the framework-switch dynamics a Warsh-era Fed is inside.