Technical note · On-chain arbitrage · Co-developer & strategy lead · 2022–24
Latency as the real constraint
Watched several exchanges and the next block before it landed, and raced other bots around three-step trades.
Loading figures…
What it was
An automated system that traded price differences between several decentralized exchanges: three-step routes through different pools, and opportunities visible in pending transactions before the next block. We ran our own node to see them sooner. I co-developed it and led strategy, and worked with the CTO on the execution-latency infrastructure.
The problem
Every competitor sees the same prices. Detecting a profitable cycle is a textbook graph problem; capturing it depends on how fast the trade lands compared with everyone else's.
Under the hood
Re-creation of the problem with a standard method and synthetic data · not production code
Loading figures…
Result
- CVAbout $38,000 net profit from $7,500 of starting capital.
- CVIdentified execution latency as the competitive constraint.
What carried forward
Judging a system by outcomes against real competitors, not by whether the code runs, is how I evaluate agents now.