Financial coverage is full of statements like "traders see a 56% chance of a rate hike in September". It sounds precise. It is worth understanding where that number actually comes from, because it is not a survey and nobody is being asked their opinion.
These probabilities are derived from prices in markets where people commit real money. This tutorial explains how the calculation works and how to use the result.
The basic idea
If a contract pays out depending on whether an event happens, its price reflects the collective view of how likely that event is.
Suppose a contract pays $1 if the Fed raises rates and nothing if it does not. If that contract trades at 56 cents, buyers are collectively saying there is roughly a 56% chance of a hike. Pay much more and you lose money on average if the true probability is lower; pay much less and someone else will bid it up.
That is the whole principle. The rest is mechanics.
Fed funds futures
The main source of these numbers is the fed funds futures market, which has existed for decades and is used by institutions to hedge interest rate exposure.
These contracts settle based on the average federal funds rate over a given month. That averaging detail matters, and it is the reason the calculation is slightly more involved than it first appears.
Because the contract settles on a monthly average, and a rate change takes effect partway through the month, the implied rate in the contract is a blend of the old rate and the new one, weighted by how many days each applies.
Working backwards to a probability
The calculation runs in reverse from the contract price.
- Take the rate implied by the futures contract for the relevant month.
- Work out what that implied rate would be if the Fed held, given the current target range.
- Work out what it would be if the Fed moved by a quarter point.
- See where the actual implied rate sits between those two values.
If the implied rate sits a bit over halfway toward the "hiked" scenario, the market is pricing a bit over a 50% chance of a hike. That is where a figure like 56% comes from.
Prediction markets
The second source is prediction markets, where contracts pay out directly on a stated outcome. These are more intuitive because the price is the probability without any calculation.
In late August 2026, one such platform showed roughly 48% odds of a quarter-point hike in September, while fed funds futures implied closer to 56%.
Small differences between the two are normal. They have different participants, different costs and different liquidity. Large persistent gaps are unusual and usually reflect a technical quirk rather than genuine disagreement.
Why the numbers move so much
These probabilities update continuously, and in 2026 they have been remarkably volatile.
Before Fed chair Kevin Warsh spoke at Jackson Hole in August, the odds of the Fed simply holding in September were close to 70%. After his firm comments on fighting inflation, hike odds jumped toward 48% and 56% on the two measures.
Then governor Christopher Waller said he would support holding if inflation continues toward target, and hike odds fell back to about 50%, down from around 63% the previous day.
Two speeches. Roughly forty percentage points of movement in the implied probability. That is the market repricing in real time.
What the numbers are good for
Understanding what is already priced. This is the most valuable use. If a hike is 90% priced, the hike itself will not move markets much - but a hold would move them enormously. The surprise, not the event, drives the reaction.
Gauging asymmetry. When odds are heavily skewed one way, the risk is concentrated in the other direction. That is a genuinely useful input into position sizing.
Tracking sentiment shifts. Watching the series over days tells you how the balance of opinion is moving, which is more informative than any individual commentary.
What they are not good for
Predicting the outcome. A 56% probability is barely different from a coin flip. It is not a forecast that a hike will happen.
Taking a directional position by itself. Knowing the odds tells you what is priced, not what will occur.
Precision. The difference between 52% and 56% is well within the noise created by liquidity and rounding. Treating small changes as meaningful is over-reading.
The multi-meeting complication
When looking further ahead, the calculation gets harder. A contract for a meeting several months away reflects the combined probability of everything that might happen in between.
The market might imply a rate a full quarter point higher in six months, but that could mean a near-certain single hike, or a coin flip between two hikes and none. The same implied rate is consistent with very different distributions.
Published probabilities for distant meetings therefore rest on assumptions about the shape of that distribution, and deserve less weight than near-term figures.
Are the odds any good?
Reasonably, with a caveat.
Market-implied probabilities have historically been fairly well calibrated for the next meeting or two. Events priced at 80% have happened roughly 80% of the time.
They are considerably less reliable further out, and they are systematically poor at anticipating turning points. Markets tend to extrapolate the current regime and are repeatedly surprised when it changes - which is precisely the pattern behind the repeated failure of the rate-cut trade in recent years.
How to check them yourself
Several exchanges and data providers publish implied probabilities free of charge, updated continuously. You do not need to perform the calculation.
What is worth doing is checking them before a major release, so you know what is priced, and again after, so you can see how much the data actually changed expectations. The size of that shift explains the size of the market reaction better than the data itself.
The summary
Rate probabilities come from prices in markets where participants commit money, converted into percentages through a straightforward calculation. They tell you what is already priced rather than what will happen.
Used properly, they are one of the more useful free tools available: they show you where the surprise risk sits, they update faster than any commentary, and they impose a helpful discipline by replacing opinions about what the Fed should do with a number describing what the market currently expects.
Why these prices are usually trustworthy
There is a reasonable question here: why should a market price be a better guide than an expert forecast?
The argument is about incentives. Someone offering an opinion on television faces no direct cost for being wrong. Someone buying a contract at 56 cents loses money if the true probability was lower.
Money at risk concentrates attention. It also aggregates information from many participants, each holding a different piece of the picture, into a single figure.
This is not a claim that markets are always right. They are demonstrably not, particularly at turning points. It is a claim that a price reflecting committed capital is generally a better starting estimate than an uncommitted opinion.
Where the approach breaks down
Several situations reduce reliability, and knowing them prevents over-trusting the number.
Thin liquidity. Prediction markets on niche questions can be moved by relatively small amounts of money, making the implied probability unreliable.
Hedging demand. Fed funds futures are used by institutions to hedge real interest rate exposure, not only to speculate. A large hedging flow can distort the implied probability without reflecting anyone's genuine view.
Regime changes. Markets extrapolate current conditions and are systematically poor at anticipating turning points - which is the underlying reason so many priced rate cuts failed to materialise in recent years.
Very low or very high probabilities. The difference between 2% and 5% is hard to price accurately and often reflects transaction costs more than genuine assessment.
Using odds for position sizing
The most practical application is not directional but about risk, and it is underused.
When an outcome is priced at 90%, the market is heavily positioned for it. The move if it happens is small; the move if it does not is large. Holding a position aligned with the consensus in that situation offers poor asymmetry.
When odds sit near 50%, as they do for September 2026, roughly half of all positioning is wrong whichever way it resolves. That forced unwinding is what produces outsized moves, and it argues directly for reducing size through the event.
So the probability figure is best read as a volatility forecast rather than a direction forecast. It tells you how violent the resolution is likely to be, which is information you can actually act on without predicting anything.
Watching the change rather than the level
A final practical habit: the movement in the odds is more informative than the current reading.
A jump from 30% to 56% after a speech tells you something happened that genuinely shifted expectations. The level of 56% alone tells you only that the outcome is uncertain.
Recording the series over a few weeks around a decision gives you a clear picture of what actually moved the market. In the current cycle, that record shows two speeches producing roughly forty percentage points of combined movement - which is a far more useful summary of the situation than any article about what the Fed might do.
This article is general information, not financial advice. Do your own research or speak to a licensed professional before making money decisions.
General information, not advice. This article is published to everyone who reads it and takes no account of your circumstances, so it is not a personal recommendation. Trader Suite is not authorised or regulated by the FCA. Trading and investing involve a substantial risk of loss, and you should seek independent advice before acting. Our articles are researched and drafted with AI assistance and reviewed before publishing. Full risk disclosure.
TraderSuite Team
TraderSuite builds indicators and automated strategies for NinjaTrader 8. Our articles are written by the team, researched and drafted with AI assistance, and reviewed before publishing.






