Before You Bet

Inspiration

Prediction markets are powerful tools for understanding real‑world events — but many users are intimidated by the technical rules, contract terms, and economic indicators behind each market. Kalshi events often involve detailed definitions, timing rules, and resolution conditions that can confuse even experienced users.

We wanted to build something that makes these markets clearer, more transparent, and more approachable for everyone. So we created Before You Bet — an AI‑powered explainer that breaks down any Kalshi event into simple, digestible insights.

What it does

Before You Bet automatically reads a Kalshi event and produces a structured, easy‑to‑read explanation that helps users understand what the event actually measures and how it will resolve. It also adds probabilities based on external data.

Our system: Extracts the event title, description, and rulebook Pulls in factual information (like release calendars or reporting processes) Analyzes public conversations to identify common misunderstandings Detects ambiguities or resolution edge cases in the official rulebook Produces a clean, Community‑Notes–style summary that clarifies everything

How we built it

We combined multiple data sources and a structured multi‑stage analysis pipeline:

Event Intake Parse Kalshi event URLs

Extract title, description, and contract terms PDF

Clean and structure rulebook text

Data Collection

Retrieve factual news snippets and institutional context (e.g., BLS release dates)

Fetch current and historical market structure (prices, dates, etc.)

AI Analysis Layer We used a multi‑agent reasoning approach:

Interpreter Agent Breaks down the event into plain language.

Context Agent Adds background and factual information the user should know.

Resolution Agent Identifies ambiguous rulebook sections or common misunderstandings.

Note Writer Agent Produces a one‑sentence clarity note.

Output Synthesizer All agent outputs are merged into a clean JSON structure that the UI renders.

The entire flow is encapsulated in our backend and integrated with a simple frontend interface.

Challenges we ran into

Distinguishing meaningful discussion patterns from noise

Ensuring the AI stays factual and educational

Normalizing terminology across different types of Kalshi events

Building a multi‑agent system that produces concise, consistent explanations

Creating a UX that surfaces insights clearly without overwhelming users

Accomplishments that we're proud of

Designing an approachable interface for a highly technical domain

Successfully integrating multiple external data sources into one flow

Making prediction markets more understandable and transparent for newcomers

What we learned

How to build multi‑agent systems that collaborate instead of compete

The importance of context and clarity when dealing with economic and political indicators

How to distill complex real‑world systems into simple explanations

What's next for Before You Bet

Add support for discovering related events for better understanding

Introduce visual timelines for events with scheduled releases

Generate multi‑language explanations

Build a browser extension to overlay notes directly on Kalshi event pages

Expand the knowledge base for better domain‑specific reasoning (inflation, employment, weather, elections, etc.)

For the video demo, make sure to clikc on

Built With

• digitalocean
• fastapi
• kalshi
• newsapi
• openai
• python
• streamlit