> ## Documentation Index
> Fetch the complete documentation index at: https://docs.useslip.xyz/llms.txt
> Use this file to discover all available pages before exploring further.

# Product overview

> Why Slip uses one variable-stake parimutuel primitive for football prediction markets.

# Why Slip is one parimutuel primitive

Football prediction products often separate head-to-head bets, pools, props, and multi-outcome
markets into different user flows and settlement systems. That makes the product harder to understand
and multiplies the amount of escrow and resolution code that must be trusted.

Slip uses one model instead: a market partitions one provable football value into two to five named
outcomes. A user chooses one outcome and stakes any allowed amount. Pool weight produces the implied
probability; there is no bookmaker price, CLOB, AMM curve, or external pricing agent.

The underlying value can be a single TxLINE stat or a combination of two stats. Examples include:

* home goals minus away goals, partitioned into Away / Draw / Home;
* home corners plus away corners, partitioned into five total-corner ranges;
* a single team's yellow-card count, partitioned around a line;
* first-half goals or corners using TxLINE's period-prefixed stat keys.

Multiple disjoint value bands can point to the same outcome. That permits Draw / Not draw without a
separate binary-market account type: negative margins and positive margins both map to Not draw,
while zero maps to Draw.

## Why settlement is proof-driven

The keeper is an availability mechanism, not an authority. Anyone can submit the same terminal
TxLINE proof. The program reconstructs the proof path and compares it with TxLINE's on-chain daily
root before selecting a winning band. A keeper cannot choose the winner, change the labels, or route
the vault somewhere else.

This design depends on TxLINE for football facts and Solana for execution. It does not depend on
FullTime, Tissue, an admin signer, or an off-chain price engine.

## Product boundaries

Slip supports TxLINE-validatable team and match statistics. It does not expose player props because
TxLINE supplies no player-stat leaves for the settlement path. It does not offer early cash-out,
continuous trading, synthetic demo markets, or sample-data fallbacks.

Natural language is interpreted by an AI SDK structured-output compiler. The model emits semantic
concepts and complete integer bands, not raw program data. Slip then derives visible TxLINE keys and
applies deterministic program-equivalent validation. Unsupported requests, including player props,
fail closed. The resulting Rulebook remains auditable before money moves.