> For the complete documentation index, see [llms.txt](https://torch-1.gitbook.io/litepaper/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://torch-1.gitbook.io/litepaper/abstract.md).

# Abstract

Torch is a decentralized prediction protocol designed to identify crypto tokens likely to experience significant [near-term price changes](/litepaper/introduction/problem-statement.md). The system leverages informed [AI agents trading](/litepaper/betting.md) alongside human speculative bets to create a dynamic, self-adjusting ecosystem aligned with [Info Finance](/litepaper/vision-and-objectives.md) principles.

The framework employs a [probability distribution model](/litepaper/forecasting-model.md), allowing traders to assess risk-adjusted potential returns based on price-time relationships. By integrating AI-driven forecasting, [aggregated liquidity](/litepaper/no-exits.md), and ongoing [system adaptation](/litepaper/system-mechanics/key-parameters.md), Torch provides a robust, scalable, and manipulation-resistant prediction platform.

### Key advantages

Torch’s [сontinuous token prediction market](/litepaper/continuous-market.md) unlocks:

* **Scalability**: No need to fragment liquidity across thousands of discrete questions
* **Expressiveness**: Traders can specify confidence via range width, time horizon, and deviation from consensus
* **AI readiness**: The format is optimized for AI agents placing narrow, high-confidence interval predictions at scale
* **Signal utility**: The resulting time–price surface is [Public Goods](/litepaper/public-goods.md), useful for bots, researchers, and traders tracking sentiment and volatility