manja316/crypto-prediction-market-signals

Crypto + Prediction Market Cross-Signal Dataset

BTC, ETH, SOL prices + funding rates + open interest + gold + Polymarket crypto probabilities — synced at 15-minute intervals.

The only dataset that combines crypto market microstructure with prediction market sentiment in one place.

What's Inside

Table	Rows	Description
candles	8,800+	15-min OHLCV for BTC, ETH, SOL (Binance Futures)
funding_rates	460+	8-hourly funding rates + mark prices
open_interest	190+	Open interest in contracts + USD value
gold	60+	Gold spot price (XAU/USD)
pm_crypto_signals	6,300+	99 Polymarket crypto-related markets with YES probabilities
cross_metrics	60+	Combined snapshot: all prices, funding, OI, PM sentiment, BTC/Gold ratio, ETH/BTC ratio

Coverage

• Date range: 2026-03-23 to present (auto-updating)
• Crypto pairs: BTCUSDT, ETHUSDT, SOLUSDT
• Granularity: 15-minute candles, 8-hour funding, hourly OI
• Source: Binance Futures API, Gold-API, Polymarket Gamma API

Unique Value

No other dataset combines:

1. Crypto price data (candles, funding, OI) with
2. Prediction market probabilities (99 crypto markets on Polymarket) and
3. Gold spot prices for macro regime analysis

This enables research on:

• Do prediction markets lead crypto prices?
• Funding rate → price direction signals
• BTC/Gold ratio regime shifts
• Cross-asset sentiment analysis

Usage

import pandas as pd

candles = pd.read_parquet("candles.parquet")
funding = pd.read_parquet("funding_rates.parquet")
pm = pd.read_parquet("pm_crypto_signals.parquet")
metrics = pd.read_parquet("cross_metrics.parquet")

# BTC price vs PM crypto sentiment
btc = candles[candles.symbol == "BTC"].set_index("ts")
print(f"BTC range: ${btc.close.min():.0f} - ${btc.close.max():.0f}")
print(f"PM crypto markets tracked: {pm.market_id.nunique()}")

Full Dataset

This is a free sample (30 days). For the full historical dataset with 9.5M+ Polymarket prices across 9,550 markets:

• Gumroad — $9 full dataset
• Gumroad — $29/mo live subscription

Auto-Updated

This dataset refreshes weekly via automated pipeline. Star the repo to get notifications.

Contact

[email protected] | @gmanjuu