# Slippage

Also written slippage tolerance, price impact.

*https://predictionmarkets.tools/glossary/slippage · next to Prediction Market Venues*

**Definition:** The gap between the price shown when you decided to trade and the average price your order actually filled at. On a prediction market it is counted in cents on a one-dollar contract, and each cent is a percentage point of probability: pay 53 cents on average for a contract shown at 50 and you need the event to be more than 53 percent likely just to break even, before fees.

Every other market prices slippage in money. Here it also has a second unit, and the second one is
the one that matters: the contract pays a fixed dollar, so the price you fill at is the probability
you have bought. A few cents of slippage is not a rounding cost on the side of a forecast. It moves
the forecast you are actually holding.

## How it works

**The reference is the price you saw, and the result is the average you got.** A market shows 50
cents. You send an order for a thousand contracts. What comes back is a set of fills — some at 50,
some at 51, the last few at 56 — and their average, say 53. The three cents between 50 and 53 are
slippage. There are three ways to get them, and a prediction market has all three.

- **Walking a book.** On a [central limit order book](https://predictionmarkets.tools/glossary/central-limit-order-book) the order
  takes the best resting offer, then the next, then the next. How far it walks depends on the size
  resting at each level, and on a thin market there may be very little at any of them.
- **Moving a curve.** On an [automated market maker](https://predictionmarkets.tools/glossary/automated-market-maker) there is no
  list of offers. Each contract bought moves the formula's price, and the average of the path is
  worse than where it started. A curve always quotes, and always charges for being used.
- **Arriving late.** The price moved between the moment it was on your screen and the moment the
  order reached the venue. On a market reacting to a vote count, a score or a data release, that
  interval can contain the whole move.

### Cents are probability points

Pre-fee, the break-even probability of a Yes position is its average fill price. Slippage raises it
point for point. That changes how to read a fill in two ways.

**Near the middle it is a shift in the claim.** Buying at an average of 53 on a contract shown at 50
is a statement that the event is more than 53 percent likely. If what you believed was 52, the
slippage turned a position you agreed with into one you do not.

**Near the edges it is a large share of the stake.** Two cents on a 6-cent contract is a third of
the price paid. The same two cents on a 94-cent contract is about two percent of the price, but it
is a third of the six cents that contract can still earn. Either way the edge of the range is where
a small absolute slip is a large fraction of something.

### How a curve spreads it across the range

Hanson's 2002 paper on the logarithmic market scoring rule gives the price response of that market
maker as a differential equation. For a two-outcome market, read from the trader's side, it
reduces to: the change in price per
contract bought is p(1 − p) divided by b, where p is the current price and b is a liquidity
parameter the operator chooses. Two things follow. For a given b the price moves fastest at 50
cents and slowest near 1 and 99, so the same order slips more on a coin-flip than on a longshot.
And b is set by the venue, not by the traders, so two markets with the same displayed price can
slip very differently. The [Futuur](https://predictionmarkets.tools/tools/futuur) card records a liquidity-sensitive variant,
LS-LMSR, in which slippage falls as a market gets busier.

### One word, three different numbers

Venue documents use "slippage" for things that are not the same measurement, and a reader comparing
them should know which is which.

- **What happened** — the realised gap on a fill, as defined above. No venue prints it for you in
  advance; it exists only after the order.
- **What you allow** — a *tolerance*, the furthest a market order may fill from the price shown
  before the rest is cancelled. The [Crypto.com](https://predictionmarkets.tools/tools/crypto-com-prediction-markets) card records
  a default of $0.05 per contract on a market order, and the venue's help centre describes it as a
  set number of points from the displayed price, with the order filled immediate-or-cancel and
  not at all if the price has moved past it. That is a ceiling, not an
  estimate, and on a 20-cent contract it is a quarter of the price.
- **What the currency swap costs** — on a crypto venue the deposit or withdrawal can pass through a
  token swap with its own slippage limit. The [how money gets out](https://predictionmarkets.tools/guides/how-money-gets-out)
  guide records a Polymarket bridge quote on a ten-dollar transfer with a 0.5 percent maximum
  slippage. It is a real cost, and it has nothing to do with the price of any contract.

## Why it matters here

**It is in no fee schedule.** The catalogue's [effective fee at fifty cents](https://predictionmarkets.tools/glossary/effective-fee-at-fifty-cents)
is a fee measurement: what the schedule charges on one contract at the midpoint.
Slippage is the cost that measurement leaves outside, and on a thin market it can exceed the
fee several times over. A card can tell you the rate. It cannot tell you what your order will do to a book it
has never seen.

**A displayed price on a thin market is a price for one small order.** The catalogue's liquidity
guide found every open Myriad market on 19 September 2026 quoting a two-decimal probability with a
median liquidity figure of about $1,000 among the stablecoin markets. The number on the screen is
real; the size behind it is what decides the fill, and it is shown in a different place, if at all.
The mechanics are in [where liquidity comes from](https://predictionmarkets.tools/guides/where-liquidity-comes-from).

**You pay it twice.** Getting in walks the book one way; getting out walks it the other, on
whatever depth is there on the day you leave. A position that has to be exited early — because of a
[close-only](https://predictionmarkets.tools/glossary/close-only) status, a withdrawal need, or a changed view — meets the bid, not
the midpoint.

**Tools model it, and mostly optimistically.** The [Polyrama](https://predictionmarkets.tools/tools/polyrama) card records terms
describing its simulated results as relying on optimistic fill models that do not reflect slippage or
market impact. The [Polycool](https://predictionmarkets.tools/tools/polycool) card records the opposite kind of control, a
slippage guard that skips a copied trade when the price has moved 1 to 10 cents past the leader's
entry. A backtest or bot that fills at the displayed price is measuring something other than what
the venue will do.

**Price-as-probability comparisons need it subtracted.** When a dashboard or a scoring page treats a
market's price as the crowd's forecast, the relevant price for a trader is the fill, not the quote.
On a deep market the two are close. On a shallow one they can be different forecasts.

## Where you will meet this

- [Crypto.com Prediction](https://predictionmarkets.tools/tools/crypto-com-prediction-markets.md)
- [Futuur](https://predictionmarkets.tools/tools/futuur.md)
- [Manifold API](https://predictionmarkets.tools/tools/manifold-api.md)
- [Polycool](https://predictionmarkets.tools/tools/polycool.md)
- [Polyrama](https://predictionmarkets.tools/tools/polyrama.md)
- [Synthesis](https://predictionmarkets.tools/tools/synthesis.md)

## Sources

1. [Logarithmic Market Scoring Rules for Modular Combinatorial Information Aggregation (working paper, January 2002)](https://mason.gmu.edu/~rhanson/mktscore.pdf) — Robin Hanson, George Mason University, read 2026-09-27
2. [Prediction Trading](https://help.crypto.com/en/articles/11373970-prediction-trading) — Crypto.com Help Center, read 2026-09-27

*Last updated 2026-09-27. A reference page, corrected in place — not a dated post.*