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

# Prediction Market

> A Conductor app: paste a Polymarket or Kalshi market and the Conductor models the outcomes with a Timepoint simulation, showing the model's probability read — with honest uncertainty bands — alongside the market's implied odds.

# Prediction Market

<Note>**Beta.** A [Conductor App Store](/products/app-store) app, available to any authenticated user.</Note>

**Hand the Conductor a prediction market and get the model's read.** Paste a [Polymarket](https://polymarket.com) or [Kalshi](https://kalshi.com) market URL and the [Conductor](/products/conductor) inspects the question and its outcomes, runs a Timepoint simulation over them, and shows you the **model's probability read** for each outcome — alongside the market's own implied odds, so you can see where the two disagree.

## What you ask it

* *"Model this market for me: \<Polymarket URL>."*
* *"What does the simulation think the odds are on this Kalshi market, and where does it disagree with the market?"*
* *"Pull up that market I modeled earlier."*

## How it works

1. **Fetch (free).** The Conductor reads the market's question, outcomes, and the market's implied odds. Inspecting a market costs nothing.
2. **Model.** It runs a branching Timepoint simulation over the outcomes and returns the model's probability read for each, with an uncertainty band.
3. **Compare.** It lays the model's read next to the market's implied odds and flags a meaningful disagreement only when the result is large, robust, and grounded.

## The honesty stance

This app is deliberately careful about what it claims, and the Conductor echoes these rules every time it narrates a result:

* **It's the model's read — never "true odds."** The numbers are what a simulation estimates, presented as an opinion, not a statement of real-world probability.
* **Always a band, never a single number.** Results come with an uncertainty range (a p50→p90 band), never collapsed to a false-precision point estimate.
* **An "edge" is flagged sparingly.** A disagreement with the market is called an *actionable edge* only when it's large, robust, and well-grounded — otherwise it's just noise.
* **An undefined simulation says so.** If the simulation can't produce a meaningful distribution, the Conductor tells you it's undefined rather than inventing numbers.

<Note>
  Polymarket markets work out of the box. Kalshi support may require configuration; if a Kalshi market isn't reachable, the Conductor tells you so rather than guessing.
</Note>

## Where it lives

Enable **Prediction Market** from the App Store in the authenticated web app at [app.timepointai.com](https://app.timepointai.com), then just paste a market URL into the chat. See the [App Store](/products/app-store) overview for how apps work.