What a 20% probability actually means
A 20 percent probability is the most misread number on any board. Too small to feel real, too large to dismiss, so most people round it to zero. The July 2026 Federal Reserve market, which closes on 29 July 2026, is a clean place to practice: it prices a 25 basis point increase at 19.4 percent and no change at 79.8 percent.
One time in five is common, not rare
An event at 20 percent happens roughly one time in five, the frequency of rolling a one or a two on a die. Watch five situations priced near this level and the honest expectation is that one of them lands. The core mistake is reading 19.4 percent as a synonym for no. It is a market saying the outcome is uncommon and entirely live.
The price is a live estimate, not a published forecast
A forecast in an article is a snapshot: written, filed, and frozen at whatever the world looked like that morning. A market price is not that. 19.4 percent is where buyers and sellers currently agree to transact, and it updates every time an order hits the book. New data, a shift in tone from an official, a repricing elsewhere in rates, all of it reaches the number before it reaches the commentary.
Notice too that 19.4 and 79.8 do not sum to 100. The rest sits in other outcomes and in the spread between bid and ask, a reminder that a screen price is a trading level, not a constant.
The arithmetic, in cents
A binary share pays $1 if the outcome resolves yes and nothing if it does not. At 19.4 percent, one share of the increase outcome costs 19.4 cents:
- Resolves yes, you collect 100 cents, a gain of 80.6 cents per share.
- Resolves no, you lose the 19.4 cents you paid.
Weigh those branches by the probability you actually believe, call it p, and nearly everything cancels: expected value per share, in cents, is (p x 100) minus 19.4, which is just your probability minus the price. The decision was never "will the Fed raise." It is one comparison: is your honest number above or below 19.4?
Take a hypothetical estimate of 25 percent. A quarter of the time you gain 80.6 cents, contributing 20.15 cents. Three quarters of the time you lose 19.4 cents, contributing minus 14.55 cents. Net expected value is plus 5.6 cents per share, exactly 25 minus 19.4. Bring 15 percent to the same price and the identical arithmetic returns minus 4.4 cents. Same market, opposite decision, and the only input that changed was your number. That 5.6 cents is gross, before spread and fees, so a thin gap may be no gap at all.
SmartX is an independent AI trading terminal for prediction markets. It streams live odds as they reprice, ranks markets by volume, and shows what proven wallets are doing, useful context when you are working out whether the price or your own estimate is the one that is wrong. Fees are a flat 0.5 percent.
Open SmartX →Unlikely and mispriced are different claims
You can agree a hike is unlikely and still think 19.4 is the wrong number. Someone who believes the true odds are 12 percent and someone who believes 28 percent both agree the outcome is improbable, and they belong on opposite sides of the trade. The 79.8 percent side is the mirror image: paying 79.8 cents to collect 100 makes sense only if no change is more likely than that, and the comfortable outcome is usually the one already picked over.
Base rates, used as a way of thinking
Build your own number from frequency, not from how the headline feels. Ask how often, across a long run of comparable situations, the surprise happened, then adjust for what is different now and write the adjustment down. Treat this as a method, not a figure to quote: pick the reference class too narrowly and you have three examples and a coincidence, too broadly and you are averaging events that share only a label. Our guide to base rates in prediction markets runs the same logic.
Sizing, and what invalidation looks like
A position at these odds fails around four times out of five even when the reasoning is sound, so size it so that losing several in a row changes nothing about how you act the next time. Then decide in advance what would show the read was wrong rather than early: a data release pointing clearly the other way, an official saying something the position assumed they would not, or the price moving through the level where your claimed gap no longer exists. Write it down before you enter.
FAQ
Does a 20 percent probability mean the event will not happen?
No. It happens roughly one time in five. Reading 19.4 percent as no throws away the difference between improbable and impossible, and that difference is where the interesting decisions live.
Is 19.4 cents a cheap price?
Cheap is not a property a price has on its own. It is worth paying only if your honest estimate sits above 19.4 percent by enough to cover spread, fees, and your own error. Below that, the contract is expensive however small the ticket looks.
Why do the two prices not add up to 100?
19.4 and 79.8 sum to 99.2. The rest sits in the other outcomes on the board and in the spread between bid and ask, so read any single quote as an estimate carrying friction.
How should I size a position on a low probability outcome?
Small enough that losing four times in a row is unremarkable, because at these odds that sequence is normal rather than evidence of a broken method. Fix the size and the invalidation before entering.
None of this says what the Federal Reserve will do. It says what the number in front of you is worth reading as: an outcome that is uncommon and fully possible, repriced by traders until the market closes on 29 July 2026. Produce your own number, compare it to the price, and act only when the difference survives your uncertainty. Nothing here is financial advice, and prediction markets carry risk.