manifold-analyzer

Manifold Markets Analyzer

Analyze and visualize prediction market data from Manifold Markets. Fetches trade history via API, aggregates by trader and time period, and generates interactive HTML visualizations.

Features

• API Data Fetching: Paginated fetching of all bets for any market
• Trader Analytics: Volume, trade count, YES/NO breakdown, bull/bear stance
• Time Series: Monthly cumulative volume charts
• Interactive Visualizations: Standalone HTML files with Chart.js
• P&L Estimation: Estimate trader profits based on entry prices vs current probability
• Market Impact: Track how much each trader moved the price
• Trader Classification: Auto-tag traders as WHALE, BULL, BEAR, ACTIVE, WINNER, LOSER

Quick Start

1. Find a Market ID

Search for a market:

curl "https://api.manifold.markets/v0/search-markets?term=your+search+term"

Or grab from the URL - the market ID is in the API response.

2. Fetch Market Data

python3 scripts/fetch_market_data.py --market-id MARKET_ID --output all > market_data.json

Options:

• --output all - Summary, traders, monthly, last 100 trades
• --output json - Full data including all trades
• --output traders - Just trader leaderboard
• --output monthly - Just monthly aggregates

3. Resolve Usernames (Optional)

The API returns user IDs, not names. To resolve top traders:

import json
from urllib.request import urlopen

def get_username(user_id):
    url = f"https://api.manifold.markets/v0/user/by-id/{user_id}"
    return json.loads(urlopen(url).read())['username']

Note: Be conservative with API calls - add 1 second delays between requests.

4. Create Visualization

Use one of the example HTML files as a template:

• iran_market_viz_chartjs.html - Binary YES/NO market
• ai_imo_gold_viz.html - Long-running market with many months
• agi_2030_viz.html - Medium-sized market

Example Markets Analyzed

Project Structure

manifold-analysis/
├── scripts/
│   ├── fetch_market_data.py    # API fetcher with pagination
│   ├── analyze_traders.py      # Deep trader analysis (P&L, impact, timing)
│   ├── generate_viz.py         # Generate enhanced HTML visualizations
│   └── parse_trades.py         # Parse text/CSV trade data
├── examples/
│   ├── asl3_danger_data.json   # Example: ASL-3 market data
│   ├── asl3_danger_usernames.json
│   └── asl3_danger_viz.html    # Example visualization
├── references/
│   └── visualization_template.md   # React/Recharts template
├── SKILL.md                    # Claude Code skill definition
├── *_viz.html                  # Generated visualizations
└── *_data.json                 # Cached market data (gitignored)

Example: ASL-3 Danger Market

The examples/ folder contains a complete worked example:

# The data is already fetched - just open the visualization
open examples/asl3_danger_viz.html

Market: "Will any LLM released by EOY 2025 be dangerously ASL-3?"

• 145 trades, 60 traders, Ṁ18k volume
• Shows how data flows from API → JSON → HTML viz

Trader Analysis

Run deep analysis on any market:

python3 scripts/analyze_traders.py market_data.json --current-prob 0.95

Outputs:

• P&L Leaderboard: Estimated profits based on entry price vs current
• Biggest Losers: Who's underwater on their positions
• Market Movers: Traders with most price impact
• Position Flippers: Who changed direction
• Classifications: WHALE, BULL, BEAR, ACTIVE, WINNER, LOSER tags

Generating Enhanced Visualizations

python3 scripts/generate_viz.py data.json usernames.json -o output.html --url "username/market-slug"

The enhanced viz includes P&L estimates, ROI%, market impact, and trader type badges.

API Reference

Endpoints Used

Rate Limiting

Be nice to the API:

• 1 second between paginated bet requests
• 0.5-1 second between user lookups
• Consider skipping bulk user resolution for large markets

Visualization Features

The generated HTML files include:

• Stats Cards: Probability, volume, trades, unique traders
• Stacked Area Chart: Cumulative YES/NO volume over time
• Trader Leaderboard: Top 20 by volume with stance indicators
• Responsive Design: Works on mobile
• Dark Theme: Easy on the eyes

Polymarket Support

In addition to Manifold, the repo analyzes real-money Polymarket markets via their public APIs (no key required):

# Single market: fetch + deep-dive viz (price curve, volume, order flow, holders)
python3 scripts/fetch_polymarket.py --slug <slug> -o examples/<name>/data.json
python3 scripts/generate_poly_viz.py examples/<name>/data.json -o examples/<name>/viz.html

# Multi-market topic report — overlaid probability curves, comparison table,
# AND an auto-detected term structure (today's odds across resolution dates)
python3 scripts/generate_poly_report.py --slug <slug1> --slug <slug2> --title "Topic" -o report.html

# Price-range ladder for multi-outcome threshold events (e.g. oil price)
python3 scripts/generate_poly_ladder.py --slug what-price-will-wti-hit-in-june-2026 -o ladder.html

# Multi-outcome dashboard (scenarios, party races, seat distribution) — e.g. 2026 midterms
python3 scripts/generate_poly_midterms.py -o examples/midterms/midterms_dashboard.html
# Refresh it, print what moved, archive history/<date>.json (add --commit to commit)
python3 scripts/refresh_midterms.py

Note: Polymarket exposes no full trade-by-trade history for busy markets, so analysis uses current holders (positions) plus aggregate volume windows. See examples/hormuz/ for a worked multi-market example (Iran / Strait of Hormuz crisis) including time-series overlay, a term-structure cross-section, and an oil price-range ladder.

Claude Code Skill

This repo includes a skill definition (SKILL.md) that enables Claude Code to:

1. Fetch market data from any Manifold or Polymarket URL
2. Process and aggregate trade history / current positions
3. Generate interactive visualizations and multi-market topic reports
4. Identify key traders and market dynamics

Trigger phrases:

• "Analyze this Manifold market: [URL]"
• "Analyze this Polymarket: [URL]"
• "Show me the traders on [market name]"
• "Create a visualization for [market]"
• "Produce a topic report across these markets"

Tech Stack

• Python 3: Data fetching and processing
• Chart.js 4.x: Interactive charts (CDN loaded)
• Vanilla HTML/CSS/JS: No build step required

License

MIT