Market expectations: what the yield curve, inflation-linked bonds and options reveal about the future

Q: How many rate cuts is the market pricing in?

You compare the forward rate for a future period with today's policy rate and divide the difference by 0.25. If the forward rate in one year is 0.50 percentage points below the policy rate, about two cuts are priced in, before term premia.

Q: What is breakeven inflation?

The yield spread between conventional and inflation-linked government bonds of the same maturity. It shows the inflation that is priced in, but contains an inflation risk premium and liquidity effects of the inflation-linked bonds.

Q: How reliably does the yield curve predict recessions?

In the US, every recession from 1968 to 2005 was preceded by an inversion in the monthly average, with no false signal. The long inversion from October 2022 to December 2024, by contrast, was not followed by a recession. The curve is a warning sign, not a certainty.

Q: How do you calculate the recession probability from the yield curve?

With a probit model: P = Φ(α + β · spread), where spread = 10-year yield minus 3-month T-bill, as a monthly average. With the New York Fed's 2006 parameters, a spread of zero gives 27.3% and a spread of −1.64 percentage points gives 72.7%.

Q: Are probabilities from option prices real probabilities?

No, they are risk-neutral: they contain the premium for protection. Priced-in probabilities of price declines are therefore usually higher than the actual frequency.

In depth12 min read · updated 27 October 2026 · 20 sources

Market expectations are embedded in many prices: the yield curve reveals what interest rates the market expects in a year's time, inflation-linked bonds show the inflation that is priced in, the slope of the curve can be translated into a recession probability, and options price in market swings. Here you will learn how to calculate these figures from yields and prices, why they are not pure forecasts (risk premia distort them) and what research knows about their accuracy.

What “pricing in” means

If you buy a two-year bond today instead of two one-year bonds in succession, you are betting on the interest rate in the second year. Because many investors run this calculation, the price contains an expectation. That is what “the market is pricing in” means. The same applies to inflation: a conventional bond and an inflation-linked bond deliver the same return if inflation exactly equals the gap between their yields. And options are more expensive the more strongly the market expects prices to fluctuate.

The catch: besides expectations, prices also contain risk premia. Anyone holding long-dated bonds demands compensation for the price risk (the term premium); anyone buying protection pays, on average, more than it pays out. You therefore cannot read an expectation off a market price without assuming something about the premium.

Question	Source	What else is in the price
Where are policy rates heading?	forward rates from the yield curve, interest rate futures	term and liquidity premia
What inflation is coming?	breakeven inflation, inflation swaps, surveys	inflation risk premium and liquidity premium
Is a recession coming?	slope of the yield curve in a probit model	model error, fitted with hindsight
How much will equities fluctuate?	implied volatility, e.g. the VIX	hedging premium, skew

Forward rates: the rate path in the yield curve

A forward rate is the interest rate for a future period that you can lock in today via the yield curve; the basic idea is explained in the article yield curve. For market expectations, calculations usually use continuously compounded rates, because time segments can then simply be added together. Yields on US Treasuries are quoted on a semi-annual basis and are converted first. The gap between the forward rate and the policy rate is called the priced-in rate change. Divided by 0.25, it gives the number of priced-in rate moves.

r = 2 · ln(1 + y / 2)
f(t₁, t₂) = (r₂ · t₂ − r₁ · t₁) / (t₂ − t₁),   as an annual rate e^f − 1
priced-in rate moves = (forward rate − policy rate) / 0.25

y: yield for maturity t, semi-annual basis, as a decimal
r₁, r₂: continuously compounded rates for maturities t₁ < t₂ in years
f(t₁, t₂): continuously compounded forward rate between t₁ and t₂

Because a forward rate concentrates the change between two yields into a short segment, it reacts more strongly than the yields themselves: if the curve slopes downward, the forward rates lie below the yields; if it slopes upward, above them. For the euro area, the ECB publishes instantaneous forward rates (forward rates for an infinitesimally short period) every day from its yield curve for AAA-rated government bonds, estimated with the Svensson model (ECB, yield curve methodology 2026).

Expectations hypothesis and term premium

If forward rates were pure expectations, the expectations hypothesis would hold: long-term rates are the average of expected short-term rates, and the forward rate for the second half of the year is exactly the rate the market then expects. In reality, the following is closer to the truth:

forward rate = expected short-term rate + term premium (+ liquidity and convention effects)

term premium: compensation for the risk that interest rates develop differently than expected; not observable, only estimated

The premium varies over time (see the research below). The same applies to monetary policy: many studies use futures on the Fed's overnight rate (fed funds futures) as a measure of the expected rate path, but their prices also contain a premium. As a rough assessment: the further in the future a forward rate lies, the more premium and the less pure expectation it is likely to contain.

Breakeven inflation: what inflation-linked bonds price in

Breakeven inflation is the yield spread between a conventional government bond and an inflation-linked one of the same maturity; in the US, between Treasuries and TIPS. The basics are covered in the article real return. As with interest rates, a forward value can be calculated from two breakevens: the 5y5y forward inflation, the inflation priced in for years six to ten.

breakeven = nominal yield − real yield (same maturity)
5y5y = ((1 + BE₁₀)¹⁰ / (1 + BE₅)⁵)^(1/5) − 1 ≈ 2 · BE₁₀ − BE₅

BE₅, BE₁₀: breakeven inflation for 5 and 10 years

Premia are at work here too, and in both directions. Investors demand a premium on nominal bonds for the risk of unexpected inflation, the inflation risk premium, which Bekaert and Wang (2010) estimate for several countries. It raises the breakeven. TIPS, on the other hand, are less liquid than conventional Treasuries. That raises their yield and pushes the breakeven down. How large both effects were is covered in the research section below.

Surveys instead of market prices

In the US, breakevens from TIPS provide a market price for inflation expectations. For the euro area, there are no freely available market prices such as inflation swaps. The most important alternative there is the ECB's Survey of Professional Forecasters: since January 1999, the ECB has surveyed experts at financial and non-financial institutions four times a year (in January, April, July and October) on inflation, core inflation, growth and unemployment, collecting point forecasts and probability distributions for the current year, the next year and the year after, as well as for the longer term (ECB, Survey of Professional Forecasters 2026).

July 2026 round	expected inflation (HICP)
2026	2.7%
2027	2.2%
2028	2.0%
longer term (2031)	2.0%

53 responses, surveyed July 1–6, 2026, published on July 24, 2026 (ECB, July 2026 survey 2026).

A survey and a market price measure different things. The survey shows what experts expect without a risk premium, but only four times a year and without anyone putting money on it. The market price changes daily but contains premia and technical effects. When comparing them, the exact question also matters: “longer term” in the ECB survey means a single year, while 5y5y forward inflation is an average over five years.

A third source is prediction markets: contracts on rate decisions or inflation figures whose prices are read as probabilities. According to Wolfers and Zitzewitz (2004), they are usually fairly accurate; how much fees, interest and thin markets distort their prices is shown in the article on prediction markets.

Recession probability from the yield curve

In the US, an inverted yield curve preceded many recessions. For practical use, Estrella and Trubin (2006) at the New York Fed described a model: a probit model translates the spread between the 10-year Treasury yield and the 3-month T-bill into the recession probability twelve months later.

P(recession in 12 months) = Φ(α + β · spread)
spread = 10-year yield − 3-month bond-equivalent yield;   bond-equivalent yield = 365 · d / (360 − 91 · d)

Φ: cumulative distribution function of the standard normal distribution
α, β: estimated parameters; at the New York Fed α = −0.6045 and β = −0.7374, estimated from the years 1959 to 2005
d: discount rate of the 3-month T-bill as a decimal

Estrella and Trubin recommend Treasury yields, the combination of three months and ten years, and the 3-month rate on the secondary market, converted into a bond-equivalent yield, because T-bills are quoted as a discount rate. A recession is defined as the months after an NBER peak up to and including the trough. In every recession in the period studied, the estimated probability had exceeded 30%, and in the 1981–82 recession it reached as much as 98% (Estrella and Trubin 2006). What matters is the level:

“Levels of the spread are more informative than changes.”

Estrella and Trubin (2006)

And monthly averages count rather than daily values: all six recessions since 1968 were preceded by at least three months with a negative monthly average, and an inversion in the monthly average was never a false signal from 1968 to 2005. Negative daily values, by contrast, occurred on 100 days in months whose average stayed positive (Estrella and Trubin 2006).

Recession probability by spread (2006 parameters)

Φ(−0.6045 − 0.7374 × spread) with the parameters of Estrella and Trubin (2006). Left of zero, the curve is inverted; from around −0.82 percentage points, the model shows more than 50%.

The false signal of 2022–2024

The model is fitted with hindsight: its parameters come from the same history it explains. One clear test outside this sample went wrong: in October 2022, the curve from three months to ten years inverted and stayed inverted until December 2024, longer than ever before in recent history (J.P. Morgan Asset Management 2025). With the 2006 parameters, every inversion implies a recession probability of more than 27.3%, and the deeper it is, the higher the probability. To this day, the NBER has not identified a recession: the last peak in its dating is February 2020 and the last trough April 2020 (NBER (2026), as of October 27, 2026).

The lesson: a model with two parameters, estimated from a handful of recessions, is a rough indicator, not a forecast.

For advanced readers: risk-neutral and actual probabilities

Option prices contain an entire distribution of future prices. Breeden and Litzenberger (1978) showed that the price of a claim that pays $1 if a portfolio lies between two values on a future date can be derived from the second derivative of the call price function. From many options, you can thus obtain a risk-neutral density: the probabilities with which the market values payoffs. Without a full option chain, an approximation using implied volatility and a lognormal distribution without drift helps:

q(K) = e^(rT) · ∂²C / ∂K²
P(S_T / S_0 ≤ 1 − k) = Φ((ln(1 − k) + σ²T / 2) / (σ√T))

C(K): price of a call with strike K and maturity T; r is the risk-free rate
q(K): risk-neutral density of the price at expiry
σ: annual implied volatility, e.g. VIX / 100
k: decline, e.g. 0.10 for −10%

Priced-in probability of a 10% decline by VIX level

Lognormal distribution without drift, 30 and 186 days respectively; risk-neutral, i.e. including the hedging premium. For six months, the VIX 6-month is used in practice.

These figures are risk-neutral. They contain the premium investors pay for protection: Carr and Wu (2009) show that implied variance is on average above the variance realized later, and Bakshi and Kapadia (2003) find that delta-hedged long index options perform worse than zero on average. The actual probability of a decline is therefore usually lower than the priced-in one. Skew works in the opposite direction: deep out-of-the-money puts are more expensive than they would be under a lognormal distribution (see VIX), so the approximation tends to underestimate large declines in risk-neutral terms.

For the probit model, you maximize the log-likelihood Σ [y · ln Φ(z) + (1 − y) · ln(1 − Φ(z))] with z = α + β · spread and y = 1 for a recession month twelve months later, for example using Fisher scoring (see maximum likelihood). Because the forecast horizons of neighboring months overlap and recessions are rare, the usual standard errors are likely to overstate the precision.

What the research shows

Forward rates. Fama (1984) estimated expected returns on US Treasury bills from forward rates: they did not rise steadily with maturity but peaked at eight to nine months. For almost all combinations of maturities between one month and ten years, Campbell and Shiller (1991) found that a large gap between long and short rates heralded rising short rates in the long run, but a falling long yield in the short run. That is the opposite of what the expectations hypothesis requires. According to Cochrane and Piazzesi (2005), a single combination of forward rates predicts the excess returns of one- to five-year bonds over one year, with an R² of up to 0.44.

Futures. Piazzesi and Swanson (2004) document that excess returns on fed funds futures were positive on average and strongly countercyclical, and could be predicted from employment growth and interest rate spreads; anyone who ignores the premium reads off a distorted rate path. Their conclusion is in the title: futures prices are risk-adjusted forecasts.

Inflation. Gürkaynak et al. (2010) estimate a daily TIPS yield curve and find that inflation compensation is not a pure measure of expectations but is distorted by the inflation risk premium and the liquidity premium. According to D'Amico et al. (2018), TIPS yields were up to 100 basis points above risk-free real rates when they were introduced and up to 300 basis points above them in the 2007–2008 financial crisis, mainly because of their lower liquidity.

Recessions. Estrella and Hardouvelis (1991) showed that a steep curve heralds higher real growth and beats survey forecasts in doing so. Estrella and Mishkin (1998) found that for forecasts more than one quarter ahead, the slope was clearly the best single leading indicator of US recessions.

Limitations and common mistakes

Treating the forward rate as a forecast: a forward rate of 3.63% does not mean that the market expects exactly 3.63%, but expectation plus premium. A high term premium makes priced-in cuts look smaller than they are expected to be.
The wrong policy rate: the comparison has to match the money market. Short-term Treasury yields are often above or below the fed funds rate; fed funds futures or overnight index swaps would be more precise.
Mixing rate conventions: discount rate, semi-annual yield, continuously compounded rate and annual rate differ by a few basis points, in the example above 5.10% against 5.24%. With a rate move of 25 basis points, that matters.
Taking risk-neutral literally: a priced-in probability of 11.9% for a 10% decline within a month does not mean that one occurs in every eighth month.
Market expectations as a timing signal: what is priced in is already reflected in prices. You can only win if you are more right than the market (see market timing).

How Portfolio Lab calculates it

On the Macro page, the Market expectations tab shows, for each region, what the market is pricing in. In the euro area and Germany, Portfolio Lab compares the instantaneous forward rate in one year from the ECB yield curve for AAA-rated government bonds (ECB, yield curves 2026) with the ECB deposit rate; in the US, it compares the forward rate for 6 to 12 months from constant-maturity Treasury yields with the midpoint of the Fed's target range. The difference is shown in percentage points and in rate moves of 0.25, together with the rate path today, 3 months ago and a year ago, and a weekly history.

In the US, the inflation expectation comes from breakevens; in the euro area, from the ECB survey, labeled as a survey. For the US only, the tab shows the recession probability and risk-neutral probabilities of declines of 10% and 20% and of a rise of 10%, derived from the VIX, VIX 3-month and VIX 6-month. Portfolio Lab re-estimates α and β for the recession probability on every request, using all months from 1959. This is not a forecast by Portfolio Lab; details are in the macro methodology.

Frequently asked questions

How many rate cuts is the market pricing in?