mcpField
P
prediction-markets-analysis

prediction-markets-analysis

github.com/octagonaiData & databases130250 installsUpdated 21w ago
Repository

About prediction-markets-analysis

A skill that researches Kalshi prediction markets using the Octagon Prediction Markets Agent. It compares live market prices with model probabilities and summarizes historical outcomes, contract data, and potential price drivers for market research and trading preparation.

Key features

  1. 01Compare market and model probabilities
  2. 02Research event catalysts and price drivers
  3. 03Review historical resolution outcomes
  4. 04Retrieve live contract metrics
  5. 05Cover markets across multiple categories

Use cases

  1. 01Compare crowd prices with model estimates
  2. 02Research a market before trading
  3. 03Track events that may move prices
  4. 04Review sourced analysis of fast-moving events

MCP support

Requires Octagon MCP configuration and uses the Octagon Prediction Markets Agent. The prediction_markets_history tool retrieves historical market data.

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/octagonai/skills --skill prediction-markets-analysis

After installing, type/prediction-markets-analysisin Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

marketplace.json
README.md
references/
interpreting-results.md
mcp-setup.md

Details

From the project README

Prediction Markets Analysis

Generate comprehensive research reports on prediction market events using the Octagon Prediction Markets Agent. This skill combines real-time Kalshi market data with AI-driven analysis to identify mispricings, surface price drivers, and provide actionable trading insights.

Prerequisites

Ensure Octagon MCP is configured in your AI agent (Cursor, Claude Desktop, Windsurf, etc.). See references/mcp-setup.md for installation instructions.

Note: Report generation requires Octagon Plus, Pro, or Enterprise subscription (3 credits per report). Cached reports are free for all users.

Core Functionality

The Prediction Markets Agent provides:

CapabilityDescription
Market vs. Model ProbabilityComputes independent model probability and compares against live market price
Event-Driven ResearchIdentifies key claims, catalysts, and decision-flipping events
Historical Resolution TrackingSurfaces past resolution outcomes for context on base rates
Contract SnapshotReturns live bid/ask spreads, last traded price, volume, and open interest
Cross-Category Coverage120+ active markets across Politics, Economics, Crypto, Sports, Entertainment, Science & Tech, Climate

Model Variants

VariantDescriptionUse Case
octagon-prediction-markets-agentDefault. Returns cached report if available, otherwise generates newGeneral use
octagon-prediction-markets-agent:cacheAlways retrieves cached reportFastest response, may not be latest
octagon-prediction-markets-agent:refreshAlways generates fresh reportMost up-to-date, takes several minutes

Example Queries

Federal Reserve Policy
Analyze this Fed decision prediction market: https://kalshi.com/markets/kxfeddecision/fed-meeting/kxfeddecision-26jun
Cryptocurrency Price Range
Research this Bitcoin price prediction market: https://kalshi.com/markets/kxbtcminy/how-low-will-bitcoin-fall-this-year/kxbtcminy-27jan01
Sports Matchup
Analyze this NHL playoff prediction market: https://kalshi.com/markets/kxnhleast/eastern-conference-championship/kxnhleast-26
Entertainment Awards
Research this Grammy nomination market: https://kalshi.com/markets/kxgrammynomsoty/grammy-nominees-for-song-of-the-year/kxgrammynomsoty-69
IPO Timing
Analyze when SpaceX will IPO: https://kalshi.com/markets/kxipospacex/when-will-spacex-ipo/kxipospacex
Force Fresh Report
Generate a fresh report (bypass cache) for: https://kalshi.com/markets/kxbtcminy/how-low-will-bitcoin-fall-this-year/kxbtcminy-27jan01

Example Response Structure

## Market Overview
- **Event**: Fed Decision June 2026
- **Resolution Date**: June 18, 2026
- **Current Market Price**: 72% (Hold rates)
- **Model Probability**: 68%
- **Mispricing Signal**: Market slightly overpriced

## Price Drivers
1. **Recent Economic Data**: CPI trending lower, supporting rate hold
2. **Fed Communications**: Dovish tone in recent FOMC minutes
3. **Labor Market**: Unemployment stable at 4.1%

## Key Catalysts
| Date | Event | Potential Impact |
|------|-------|------------------|
| June 10 | CPI Release | High - could shift probabilities 5-10% |
| June 12 | FOMC Statement | Medium - language on future path |

## Historical Resolution Context
- Last 6 meetings: 4 holds, 2 cuts
- Model accuracy on this series: 78%

## Contract Details
| Outcome | Bid | Ask | Last | Volume | OI |
|---------|-----|-----|------|--------|-----|
| Hold | 0.71 | 0.73 | 0.72 | 15,234 | 89,000 |
| Cut 25bp | 0.18 | 0.20 | 0.19 | 8,456 | 45,000 |
| Hike | 0.08 | 0.10 | 0.09 | 2,100 | 12,000 |

## Recommendation
**Slight Sell** on Hold contract. Model sees 4% overpricing. Consider waiting for CPI data before positioning.

Market Categories

The agent covers 120+ active markets across:

CategoryExample Markets
PoliticsElections, legislation, executive actions
EconomicsFed decisions, inflation, employment data
CryptoBitcoin/Ethereum price ranges, ETF approvals
SportsChampionship outcomes, playoff matchups
EntertainmentAwards shows, box office, streaming
Science & TechSpace launches, AI milestones, IPO timing
ClimateTemperature records, hurricane landfalls

Use Cases

Identify Mispricings

Compare crowd consensus against model-derived probability to find trading opportunities.

Pre-Trade Research

Research key drivers behind a prediction market price before placing a trade.

Catalyst Monitoring

Monitor upcoming events that could move contract prices.

Event Analysis

Get structured, sourced research on fast-moving events in politics, finance, and beyond.

Historical Data Access

Use the prediction_markets_history tool to fetch historical data:

Fetch historical data for prediction market ticker KXFEDDECISION-26JUN from January to June 2026

Parameters:

  • ticker: Market ticker symbol
  • captured_from: Start date filter
  • captured_to: End date filter
  • limit: Number of records
  • cursor: Pagination cursor

Integration with Other Skills

Combine with other Octagon skills for comprehensive analysis:

SkillIntegration Use
earnings-call-analysisAnalyze CEO comments relevant to prediction market events
sec-8k-analysisTrack material events that could impact market outcomes
stock-performanceCorrelate stock moves with prediction market probabilities

Analysis Tips

  1. Check model vs. market spread: Larger spreads suggest potential mispricings

  2. Monitor catalyst calendar: Know when decision-flipping events are scheduled

  3. Review resolution history: Past accuracy provides context for current probabilities

  4. Consider liquidity: Check volume and open interest before sizing positions

  5. Use cache wisely: Use :cache for quick checks, :refresh before trading

  6. Cross-reference sources: Combine prediction market analysis with fundamental research

Supported Platforms

  • Kalshi: Full support for all active markets
  • Polymarket: Coming soon

For More Markets

Browse the full list of supported markets at octagonai.co/markets.

Comments

Markdown supported