OmGoStudio/polypulse-polymarket-sample

PolyPulse — Empirical Polymarket Order Book Dataset

Empirical, isochronous discrete snapshot matrix combining Polymarket CLOB Best Bid/Ask (BBO), Binance Spot Ticks, and Chainlink RTDS Oracle feeds.

Live Hub: https://ogs.biz.id/polypulse
Open-Source Python Arbitrage Backtest: GitHub Gist
Instant 24-Hour Flash Pass ($2.49 with ALPHA50): Gumroad Checkout

---

Dataset Description

PolyPulse eliminates survivorship bias and polling jitter in prediction market research. While standard public API endpoints suffer from rate limits and variable network latency, this dataset aligns three synchronized liquidity venues onto strict 1.0-second (1Hz) and 0.25-second (4Hz Vilnius Node) grids:

1. Polymarket Central Limit Order Book (CLOB): Microsecond-stamped Bid/Ask depth for UP/DOWN binary contracts.
2. Binance Spot Anchor Feed: Underlying reference asset spot price.
3. Chainlink Real-Time Data Streams (RTDS): Decentralized oracle TWAP and benchmark index prices.

---

Schema & Column Reference

Column	Type	Description
sec	float	Elapsed seconds within continuous observation candle
timestamp_utc	string	ISO 8601 UTC timestamp synchronized across feeds
bnb_price	float	Underlying reference spot price (Binance)
pm_token_up_bid	float	Polymarket Best Bid for YES/UP contract ($)
pm_token_up_ask	float	Polymarket Best Ask for YES/UP contract ($)
pm_token_down_bid	float	Polymarket Best Bid for NO/DOWN contract ($)
pm_token_down_ask	float	Polymarket Best Ask for NO/DOWN contract ($)
cl_twap	float	Chainlink Oracle TWAP benchmark
parity_edge	float	Empirical sum arbitrage deviation ($1.00 - (Bid_Up + Bid_Down))
parity_edge_bps	float	Parity mispricing expressed in Basis Points (bps)
maker_pair_cost	float	Combined cost to acquire full outcome set at Bid ($)
taker_pair_cost	float	Combined cost to acquire full outcome set at Ask ($)
trade_count	int	Aggregated trade execution events in the 1-second slice

---

Quickstart in Python

Using Apache Parquet (Recommended)

import pandas as pd

# Load Apache Parquet directly
df = pd.read_parquet("https://huggingface.co/datasets/OmGoStudio/polypulse-polymarket-sample/resolve/main/polypulse_sample_preview.parquet")

print(f"Loaded {len(df)} discrete snapshot ticks.")
print(df[["timestamp_utc", "pm_token_up_bid", "pm_token_down_bid", "parity_edge_bps"]].head())

Full Arbitrage Execution Backtest

Check out the complete runnable open-source arbitrage backtest script on GitHub Gist:
👉 polymarket_4hz_arbitrage_backtest.py

---

Instant Commercial Feeds & Flash Passes

Use coupon code ALPHA50 for 50% off all packages:

Package	Granularity	Rows	Price (with ALPHA50)	Checkout
24-Hour Flash Backtest Pass	4Hz (0.25s) Vilnius Feed	345,600	$2.49	Instant Access
3 Days Standard	1Hz (1.0s)	259,200	$9.00	Instant Access
3 Days Vilnius Feed	4Hz (0.25s)	1,036,800	$24.00	Instant Access
1 Week Institutional Bundle	1Hz & 4Hz	3,024,000+	$49.50	Instant Access
1 Month Full Archive	Master Grid	10,000,000+	$124.50	Instant Access

Crypto payments (USDC / USDT on Polygon POS) are also supported natively on our web terminal:
🌐 https://ogs.biz.id/polypulse

---

Maintained by OmGoStudio (OGS) Quant Research Division.