Raw Bitcoin 5-minute market microstructure data with panic-event annotations. 637,000+ ticks across 2,600+ markets. CSV format. Updated weekly.
Real tick data from BTC 5-minute markets. Every price movement captured.
Weak side hit 3¢ at TTE=87s. Recovered to 78¢ before expiry. Panic flag triggered at 12:43:14 UTC.
| timestamp | up_price | down_price | tte | panic |
|---|---|---|---|---|
| 12:43:10 | 0.99 | 0.01 | 91 | true |
| 12:43:12 | 0.98 | 0.02 | 89 | true |
| 12:43:14 | 0.97 | 0.03 | 87 | true |
| 12:43:16 | 0.94 | 0.06 | 85 | true |
| 12:43:18 | 0.88 | 0.12 | 83 | true |
| 12:43:42 | 0.72 | 0.28 | 59 | true |
| 12:43:45 | 0.68 | 0.32 | 56 | true |
| 12:44:12 | 0.78 | 0.22 | 29 | - |
Most datasets give you open and close. We give you every tick from 2,600+ Bitcoin 5-minute markets.
Identify when BTC 5m markets hit extreme pricing (≤30¢). Study how often panic levels occur, at what TTE, and which side is affected.
Measure velocity of price reversion after panic events. How quickly do markets recover? What's the distribution of recovery times?
Analyze how behavior changes as expiry approaches. How do prices behave in the final 60 seconds? Where do inefficiencies persist?
Every field documented. No black boxes. Know exactly what you're buying.
| Field | Type | Description |
|---|---|---|
| timestamp | ISO-8601 | Observation time in UTC |
| market_id | string | Unique BTC market identifier (e.g., BTC-1774716300) |
| up_price | decimal | Best ask on UP side (0.00-1.00) |
| down_price | decimal | Best ask on DOWN side (0.00-1.00) |
| tte_seconds | integer | Seconds remaining until market expiry |
| panic_flag | boolean | True when either side ≤0.30 |
| recovery_flag | boolean | True when price returned >0.50 after panic |
| expiry_outcome | enum | UP, DOWN, or null if unresolved at capture |
Three ways serious traders use this data.
Backtest strategies on real BTC microstructure data. Study market efficiency at tick granularity. Publish findings with reproducible datasets.
Train models on BTC panic/recovery sequences. Test execution logic against historical paths. Validate edge cases before deploying capital.
Understand how your markets behave under stress. Identify regime changes. Build intuition from data, not anecdotes.
Transparency in methodology. No black boxes.
BTC 5-minute markets are sampled via Polymarket CLOB every 2 seconds. Data is timestamped at collection time. No interpolation. No fill-forward. Raw ticks only.
A panic event occurs when either side's best ask ≤0.30. Recovery is flagged when price returns >0.50 on the same side within the same market lifecycle.
Analysis of 2,602 BTC 5m markets (March 28 – April 7, 2026).
Markets with panic events showed observable recovery patterns in path data.
These are historical observations from collected data, not predictive indicators.
Past patterns may not repeat. This data is for research purposes only.
No subscriptions. No tiers. Just the data.
Current week — most recent complete dataset
ETH, SOL, XRP, DOGE datasets coming soon. Join the waitlist.
Everything you need to know.
Every tick captured from Bitcoin 5-minute prediction markets. Each row includes timestamps, UP/DOWN prices, time-to-expiry in seconds, and computed flags for panic and recovery events.
Yes. Current dataset is Bitcoin 5m markets only. We have the most comprehensive coverage here—637K+ ticks across 2,600+ markets. ETH, SOL, XRP, and DOGE datasets are in development.
We've already collected 637K+ records with <2s granularity. Replicating this would require weeks of infrastructure work, API integration, and data cleaning. We deliver clean, structured data immediately.
When either side of a BTC 5m market trades ≤30¢, indicating extreme positioning. Our dataset flags these automatically so you can study how markets behave under stress.
Raw data only. No trading recommendations. No strategy. Just clean, structured market microstructure for you to analyze and build upon.
CSV with header row plus JSONL for streaming processing. Schema documentation is included. A free sample is available for download before purchase.
Yes. Once purchased, the data is yours to use as you see fit. Attribution is appreciated but not required.
Weekly batches. Each purchase includes the most recent complete week of BTC data. Check the changelog for coverage stats.