The Staircase Turn

The Federal Reserve spent late 2025 walking interest rates down. Inflation has spent 2026 walking back up. On Wednesday afternoon, the market says there is a real chance the Fed turns around on the stairs.

TON618 Capital Research · As of July 27, 2026 · Fed funds target 3.50–3.75% · Download PDF

Every figure in this note is computed from Federal Reserve Economic Data (FRED) vintages, US Treasury fiscal data, and live prediction-market prices as of July 27, 2026; series IDs and pull dates are listed in the appendix.

The strangest bet board of the year

On Wednesday, July 29, at 2:00 p.m. Eastern, the Federal Reserve's rate-setting committee — the FOMC — announces what it will do with the short-term interest rate that anchors the price of every mortgage, car loan, and corporate borrowing in America.

Most Fed meetings are foregone conclusions. The market usually knows the answer weeks in advance, and the announcement lands like a weather report confirming yesterday's forecast. This one is different, and you can see it in the betting.

On Polymarket, a prediction market where traders stake real money on real-world outcomes, five contracts cover Wednesday's decision. Together they traded roughly $9.5 million in the past 24 hours. A contract's price is, in effect, the crowd's probability: pay 71 cents for a contract that pays $1 if the Fed does nothing, and the crowd is telling you there's about a 71% chance the Fed does nothing.

Read the board from the bottom up, because that's where the story is. The two cut contracts — the direction the Fed was moving as recently as December — are priced at a fifth of a penny. Dead. And the second-most-likely outcome, at 28 cents on the dollar, is a rate hike.

The Federal Reserve has not raised interest rates since July 2023. For most of the past year, the argument in markets was about how fast the cuts would come. Now, real money — not pundits, not surveys, money — assigns nearly a three-in-ten chance that the Fed's next step is back up the staircase it just walked down.

This note is about why that bet exists, whether the data supports it, and what Wednesday actually decides. We wrote it before the meeting on purpose: the answer will grade our reasoning in public within 48 hours.

How the Fed got to this landing

Rewind three years. In 2022 and 2023 the Fed raised rates at the fastest clip in four decades to break the post-pandemic inflation wave, topping out at a 5.25–5.50% target range in July 2023. It worked, slowly. Inflation fell, and the Fed began easing: a first round of cuts in late 2024, a pause through mid-2025, then three quarter-point cuts in quick succession — September, October, and December of 2025 — bringing the target range to 3.50–3.75%, where it sits today.

A quarter point, by the way, is 25 basis points in market language — a basis point is a hundredth of a percentage point. The board above is priced in those units.

Then, starting in January 2026, the Fed stopped. Four consecutive meetings — January, March, April, June — no move. The official line was patience. The staircase down had reached a landing.

Here is the problem. To see it, you need the right speedometer.

The inflation number that makes headlines is the 12-month rate — "prices are up X% from a year ago." That number is honest but slow, like judging a car's speed by averaging the last hour of driving. The Fed's staff and every serious forecaster also watch a faster gauge: take the last three months of price changes and ask, if this pace continued for a year, what would inflation be? That's the three-month annualized rate. It tells you what inflation is doing now.

Run that gauge on the Fed's own preferred price index — the PCE deflator, a broader basket than the better-known CPI — and this is what the two staircases look like on one chart:

In November 2025 — the month between the final two cuts — the three-month pace stood at 2.7%, within sight of the Fed's 2% target. It was the last low it would see. The pace has risen every single month since: 2.7, then 3.0, 3.7, 4.4, 5.8, 6.1, and, as of the May data, 6.3% annualized. Six readings after that low, six increases, no interruptions. The 12-month rate has followed the fast gauge upward, reaching 4.1% — double the target.

One staircase walked down. The other turned and walked up. Wednesday is where they meet.

Three dials on the Fed's desk

The Fed's mandate is jobs and prices, refereed through a third thing it never quite admits to watching: how loose or tight financial markets themselves are. Three dials. Before every meeting, this is the read:

The wrinkle that ties them together: at a 3.6% policy rate against 4.1% headline inflation, the Fed’s rate is negative in real terms — while the 10-year Treasury yield (4.69%, near the top of its 15-year range) is tightening on its own.

Prices: hot, and accelerating. The numbers above speak for themselves, but two honest footnotes. First, the burst is strongest in the headline index; the core measure — which strips out volatile food and energy to reveal the underlying trend — is running cooler, at 3.5% annualized over three months and 3.4% over the year. Cooler, but still nearly a point and a half above target, and its own path this year (2.4% last November, peaking near 4.8% in February) hardly reads as mission accomplished. Second, the better-known CPI runs cooler than PCE right now — 3.5% headline over the past year. Whichever gauge you prefer, none of them is at 2%, and the fast gauges are pointing up.

Jobs: firm. The historical case for cutting rates is a weakening job market, and it simply isn't there. Unemployment is 4.2% — low by the standard of any decade you care to compare. New claims for unemployment benefits are running at 187,000 a week — the lowest weekly reading in 15 years. Job openings are rising, not falling. If the labor market were cracking, the cut contracts wouldn't be trading at a fifth of a penny. They're priced that way because it isn't.

Financial conditions: easy. This dial is the least familiar to non-professionals and the most striking to us. When the Fed worries its policy is too tight, it looks for stress: companies paying punishing premiums to borrow, markets seizing up. The opposite is true today. The extra yield that lenders demand from risky "high-yield" companies over safe Treasury bonds — the spread, markets' fear gauge for defaults — is 2.79 percentage points, near the bottom of its three-year range. Investment-grade spreads are similarly compressed. The Chicago Fed's overall financial-conditions index reads looser than average. Money, in short, is not hard to get. Policy that was billed as "restrictive" is not restraining much.

And there is a wrinkle that ties the three dials together: with headline inflation at 4.1% and the policy rate at 3.6%, the Fed's rate is now below inflation — negative in real terms, against the headline gauge. Meanwhile the bond market has stopped waiting: the 10-year Treasury yield, which the Fed does not control, has climbed to 4.69%, near the top of its 15-year range, and is still rising. Long-term borrowing costs are going up without the Fed lifting a finger. The market is, in a sense, already hiking.

Add it up: not one of the three dials argues for a cut. Two of the three argue the current setting is too easy for the inflation actually being printed. That is why a hike contract that would have been free money to sell a year ago costs 28 cents today.

The honest case for standing still

If the data leans hawkish, why is "do nothing" still the 71-cent favorite? Not because the committee is blind. There is a real case for holding, and a note written the day before the verdict owes you its strongest version.

First, the fast gauge is fast in both directions. Three-month momentum swings hard, and the current burst is concentrated in the volatile headline components. Core momentum — the series the Fed trusts most — has actually decelerated for three straight months, from 4.8% annualized in February to 3.5% in May. A committee that reacts to every headline swerve whipsaws the economy; that is precisely why core exists.

Second, the bond market's own inflation forecast is calm. The 5-year breakeven rate — the inflation forecast you can extract from the gap between ordinary Treasury bonds and inflation-protected ones — sits at 2.18% and has been falling since spring. Traders with money at stake believe this burst fades on its own. In the 1970s — the comparison we take up next — that anchor was gone entirely. Its presence is the single best argument that this time is genuinely different.

Third, a hike is a heavy tool with a long delay. By the standard central-banking rule of thumb, rate changes bite the economy a year or more after the fact. Tighten into a price burst that was going to fade anyway and the bill — in jobs — arrives just in time for the slowdown. The 10-year yield at 4.69% is already doing quiet tightening work on mortgages and corporate borrowing, unpaid and unannounced.

That is a serious case. It is also, word for word, the case a patient central bank makes right before the moments history remembers it for being wrong. Which brings us to the ghost.

The ghost on the staircase

Once before in the modern era, the Fed cut rates while inflation was structurally alive, watched it re-accelerate, and had to turn around on the stairs. The decade was the 1970s; the chairman was Arthur Burns; and the episode is the reason every central banker alive today uses the phrase "stop-go" as an insult.

The sequence, from the data: coming out of the 1974 oil-shock recession, the Fed cut the funds rate from a peak of 12.9% in mid-1974 to 4.6% by January 1977, as inflation fell from 12.2% to 5.0%. Declaring the fever broken, policy stayed easy at the landing. Inflation turned within months — back to 6.7% by mid-1977, 9% by the end of 1978, 13.3% by the end of 1979. The Fed spent the rest of the decade chasing what it had released, the funds rate ended up above 17%, and the episode was finally closed only by the Volcker recessions — the deepest downturn between the Depression and 2008.

We show this chart to be honest about the stakes, not to predict a rerun. The differences are real and they matter: inflation today is 4%, not 12%; expectations, as the breakeven shows, are anchored, not unmoored; and today's Fed carries the 1970s in its institutional memory precisely because it is determined not to repeat them. Rhyme is not prophecy.

But the shape — cut while the underlying trend is unbroken, pause at the landing, watch momentum turn against you within months — is not a loose analogy. It is the same staircase. The 28-cent hike contract is the market pricing the possibility that this Fed knows the shape too, and would rather take one hard step early than ten harder ones late.

The third staircase

There is one more staircase in this story, and it is the reason we do not treat Wednesday as a routine meeting. It appears nowhere in the Fed's statement, and nobody at the press conference will volunteer it. It belongs to the borrower.

The United States government owes $39.7 trillion as of last Friday's Treasury statement. The interest on that debt now runs at roughly $1.2 trillion a year gross — about $970 billion net, the figure budget scorekeepers use — and it climbs regardless of what the economy does, because of a quirk of arithmetic: the average interest rate on the government's marketable debt is just 3.41%, a legacy of the cheap-money years, while a newly issued 10-year Treasury pays 4.69%. Every maturing bond that rolls over refinances at today's higher rates, so the average ratchets upward mechanically, month after month. The Treasury's average maturity is about six years and deliberately skews shorter than that, so rate changes reach the interest bill unusually fast. One percentage point on $39.7 trillion is roughly $400 billion a year once fully rolled — real money even by Washington standards.

Economists have a name for what happens when a debt gets this large: fiscal dominance — the point at which interest rates start being set with one eye on the borrower's bill rather than both eyes on the price level. Last month we published a full note on exactly this — The Debt Endgame — and its argument bears directly on Wednesday. The newly seated chair, Kevin Warsh, is by reputation a hawk; he has also inherited an instrument that doubles as a fiscal weapon pointed at his own government. A hike that would have been routine at 2007's debt levels now lands on the Treasury's bill almost immediately, and the political pressure to move rates the other way — down, so that $39.7 trillion can be refinanced cheaply — is neither subtle nor secret. That is the endgame note's thesis in one line: the textbook says the data decides, and the debt says the textbook is out of date.

Hold that against everything above and the collision is clean. The economic staircase argues up: 6.3% momentum, 15-year-low claims, loose credit. The fiscal staircase argues down: every basis point of tightening invoices the Treasury. The odds board is the market's attempt to price both forces at once — and it explains what the economic data alone cannot: why hold trades at 71 cents rather than far lower against inflation this hot, and why the cut contracts, dead on the economics, will resurrect the moment the data gives an inch.

There is a check on all of this, and it is the bond market. If the Fed is ever seen easing for the borrower's sake while inflation momentum runs above 6%, the 10-year — already near the top of its 15-year range — does the tightening itself, and the refinancing problem gets worse, not better. That is the knife's edge the chair walks on Wednesday: hike, and invoice the Treasury; ease, and risk the bond market concluding the inflation fight is no longer the priority. The staircase turn is not just a policy choice. It is a solvency argument conducted in public, 25 basis points at a time.

What Wednesday actually decides

Strip the theater away and Wednesday resolves one question: how much does this Fed fear the ghost — and how much does it owe the borrower?

The favorite — hold, with harder language (about 71%). The committee stands still but rewrites the story around the stand. Watch the statement's characterization of inflation: for months the language has leaned on "elevated." If it shifts toward "has risen" or "progress has reversed," and the chair's press conference declines to rule a hike out, then the pause has quietly become a coiled spring, and the odds boards for September — the next meeting — become the most-watched numbers in finance. A hold with hawkish language is not "nothing happened." It is the Fed acknowledging the second staircase without stepping onto it yet.

The live outsider — a quarter-point hike (about 28%). The shock would not be the 25 basis points. It would be the regime change: confirmation, three years after the last one, that rates can still go up, priced at less than one-in-three. Every asset that spent 2026 assuming the next move was down — and most did — would have to re-underwrite that assumption at once.

The dead branch — a cut (well under 1%). The board says it plainly: on this data, with claims at 187,000 and momentum at 6.3%, there is no version of Wednesday where the Fed eases. The era of asking "when's the next cut" is, at least for this stretch of the staircase, over.

For readers keeping score at home, four numbers tell you the next chapter regardless of the announcement: the June PCE release (this Thursday, July 30 — it either extends or breaks the climb), weekly jobless claims (still under 200,000 means the hold-vs-hike debate stays alive), the September hike odds on the same betting boards the moment the chair stops talking — and the 10-year Treasury yield's reaction, the bond market's running referendum on whether policy is serving the price level or the borrower.

The landing

We opened by calling the bet board strange. The data says it is not strange at all — it is priced almost exactly where the evidence points. Inflation momentum has more than doubled since the Fed stopped cutting, and every month of 2026 has made it worse, not better. The job market gives no cover to ease. Credit markets say policy isn't restrictive now. Against that: a calm expectations anchor, a decelerating core, and a bond market already tightening on the Fed's behalf. That is a legitimate two-sided argument between hold and hike — and, on the economics, no argument at all for the direction everyone spent a year assuming. Unless the argument was never only about the economics: $39.7 trillion is leaning on the handrail, and it wants the staircase to go down.

Whatever the committee announces at 2 p.m. Wednesday, one thing has already happened, and it happened before the meeting: the market stopped assuming the staircase only goes down. The turn is priced. Now we find out if the Fed takes it.

This note will be graded in public on Wednesday afternoon. That is the point of writing it now.

Addendum — the grade, 2 August 2026

Added to the original note without altering anything above it. We publish the score on the same page as the call.

The decision: hold, and we were right for incomplete reasons. The committee left the target range at 3.50–3.75% on July 29, exactly as the 71-cent favorite implied, and not one member voted to ease — the dead branch stayed dead. But the hold was not the quiet kind. Three members dissented in favor of a quarter-point hike: Beth Hammack, Neel Kashkari, and Lorie Logan. A three-vote hawkish dissent is the loudest internal argument for higher rates in years, and it is the "coiled spring" this note described. What we got wrong was the instrument. We told you to watch the statement's characterization of inflation for a shift toward "has risen." It never shifted — the language held at "remains elevated relative to the Committee's 2 percent goal," with part of the blame assigned to supply shocks including energy. The hawkishness surfaced in the vote tally, not the prose. Right conclusion, wrong place to look.

The data: our central staircase broke the next morning. June PCE landed on July 30 and ended the streak that this entire note was built on. Headline three-month annualized momentum came in at 3.07%, down from 6.31% in May — six consecutive monthly increases, then a halving on the seventh reading. Headline year-over-year eased to 3.67% from 4.08%. The chart at the top of this note now shows a local peak, not a trend. Any reader who took the acceleration as a forecast rather than a description was, within twenty-four hours, holding the wrong picture.

The part that aged best was the part we argued against ourselves. The counter-case we carried honestly — core momentum decelerating and the expectations anchor holding — is the reading the data has since endorsed. Core three-month annualized fell to 2.89% from 3.56%, a fourth straight deceleration, and the five-year breakeven sat at 2.24% on decision day against 2.18% at publication: no repricing of long-run inflation in either direction. The remaining scoreboard numbers were quiet. Initial claims were 197,000 for the week ended July 25, still under the 200,000 line that keeps the hold-versus-hike debate alive — though we owe readers a sharper version of that dial than the one we published. The 187,000 print we called a 15-year low was real and does stand as the lowest weekly reading since 2011, but it was a single-week trough in a series that had printed 209,000 to 230,000 for the five weeks before it. The four-week average through July 25 is 202,750 — above our own stated threshold. The labor dial was firm at publication, but we described it with its best week rather than its trend, and the trend is closer to the line than the note implied. The 10-year closed at 4.67% on the decision and 4.68% the day after, against 4.69% at publication — the bond market's referendum returned a shrug.

Net. The directional call was correct and the market's pricing was, as we argued, not strange at all. The thesis underneath it was half-refuted inside of a day: the inflation re-acceleration that made the hike branch live is, on the June vintage, not currently re-accelerating. Our hedge outperformed our thesis. The fiscal argument in "The third staircase" is untouched by any of this and remains the reason we think a hold trades at 71 cents rather than far lower.

Appendix — data and method

Prediction-market prices. Polymarket July 2026 FOMC contracts, pulled July 27, 2026: no change $0.71; +25bp $0.28; +50bp or more $0.006; −25bp $0.002; −50bp or more $0.002; 24-hour volume across the five contracts ≈ $9.5M. Prices are lightly rounded and, as with all prediction markets, need not sum to exactly 100%. Kalshi's July FOMC market was checked first (house convention for date-bounded events) and carried no live quotes at pull time.

Rates and inflation. All series from FRED (St. Louis Fed), pulled July 27, 2026: fed funds target upper bound (DFEDTARU — cuts effective 2025-09-18, 2025-10-30, 2025-12-11); effective funds rate (FEDFUNDS, 3.63% June 2026); headline and core PCE deflators (PCEPI / PCEPILFE, May 2026 vintage); CPI (CPIAUCSL / CPILFESL, June 2026); unemployment (UNRATE), initial claims (ICSA), job openings (JTSJOL); financial conditions (NFCI), high-yield and investment-grade spreads (BAMLH0A0HYM2 / BAMLC0A0CM), 10-year yield (DGS10), 5-year breakeven (T5YIE). Three-month annualized momentum = ((P_t / P_{t−3})^4 − 1) × 100. Percentile framings use each series' own history: 15 years for the macro series, three years for the spread series (the window available on FRED for those tickers). 1970s panel: FEDFUNDS and CPIAUCSL, 1972–1982.

Addendum sources (2 August 2026). FOMC decision and dissents: Federal Reserve press release, July 29, 2026 statement. Post-publication data all from FRED, pulled August 2, 2026: target range (DFEDTARU / DFEDTARL, unchanged at 3.50–3.75%); headline and core PCE (PCEPI / PCEPILFE, June 2026 vintage released July 30); initial claims (ICSA, week ended July 25); 10-year yield (DGS10); 5-year breakeven (T5YIE). Momentum computed on the same three-month annualized formula used above.

Debt and interest. Total public debt outstanding $39.69T (US Treasury, Debt to the Penny, July 24, 2026); average interest rate on total marketable debt 3.411% (US Treasury, Average Interest Rates, June 30, 2026); gross federal interest payments $1.219T annualized (FRED A091RC1Q027SBEA, Q1 2026); net interest outlays $970B (FRED FYOINT, FY2025). The ~6-year weighted-average maturity, the short-bill skew, and the full fiscal-dominance argument are developed in our June 30 note, The Debt Endgame.

Method notes. This is a point-in-time read of re-estimated indicators, not a forecast; indicators sitting near their historical average can flip on a single print. The odds quoted are market prices, which move continuously — they are cited as of the pull time above and will differ by publication time.