Crypto Lead-Lag & Market Microstructure
This Ospina case study documents how Carlos Rico-Ospina approached a specific risk, infrastructure, revenue, or research problem and what was built to address it.
Built cross-Pacific infrastructure to measure price propagation between exchanges—and discovered why the arbitrage edge doesn't exist.
The Problem
Cryptocurrency markets trade 24/7 across global exchanges with fragmented liquidity. Binance (Tokyo) is the global volume leader; Coinbase (Virginia) dominates US flow. If Binance leads Coinbase by even 100ms, and I could measure and act on that signal faster than the ~82ms speed-of-light limit between Tokyo and Virginia, there might be a statistical arbitrage opportunity.
The Insight
Statistically significant Granger causality exists—Binance does lead Coinbase. But the measured lag was only ~25ms. The speed of light through fiber from Tokyo to Virginia is ~68-82ms. The lag is faster than physics allows. Sophisticated quant funds aren't racing between exchanges—they're intercepting information before it reaches Binance and executing simultaneously on both venues.
What I Built
- Deployed Terraform-managed AWS infrastructure with VPC peering between Tokyo and Virginia regions
- Achieved 82ms round-trip latency—within 14ms of theoretical physics limit
- Collected ~90 million tick observations across 16 symbols with dual-timestamp methodology
- Implemented full statistical battery: Engle-Granger cointegration, Johansen tests, cross-correlation, Granger causality, Hasbrouck information share
Outcomes
- Confirmed statistically significant causality (Binance → Coinbase, p < 10^-200 for some pairs)
- Measured ~25ms lag—faster than the 68-82ms speed of light limit
- Discovered that sophisticated players intercept information upstream of exchanges, not between them
- Research revealed market microstructure truth: the arbitrage is already captured by players with earlier information access
Why It Matters
Built infrastructure to measure something precisely and discovered why the edge doesn't exist—more valuable than finding a fragile signal.
Started with physics (speed of light), tested the hypothesis rigorously, and followed the data to its logical conclusion.
Related Projects
View All
InfrastructureTrading Infrastructure Pivot
2024Turned an existential platform shutdown into a new B2B infrastructure offering—and invented the pricing model the industry adopted.
InfrastructureSegregated Portfolio Company — Broker Procurement
2023Designed a Cayman SPC structure to scale seven-figure broker relationships across multiple brands without compounding legal overhead.
Have a Similar Challenge?
Let's discuss how I can help you achieve similar results.
Start a Conversation