Services
Four kinds of work, priced by the week.
Every engagement starts with a fixed two-week block. If it is not obviously worth continuing after that, we say so and stop.
Market data pipelines
Collection from exchange websocket and FIX feeds, normalisation into a single internal schema, and storage that can be replayed bar-for-bar. Includes gap detection, clock-skew monitoring and an archive format that is still readable in five years.
Research infrastructure
Event-driven backtesting with explicit modelling of fees, borrow, latency and partial fills. We care most about the boring part: making the research path and the production path share the same code, so a strategy does not quietly change behaviour when it goes live.
Execution and risk plumbing
Order routers, position reconciliation against venue state, pre-trade limits and a kill switch that a non-engineer can reach at three in the morning. We assume every component will fail and design the failure mode rather than hoping.
Second opinions
A structured read of an existing system, delivered as a written report: what is solid, what is load-bearing but fragile, and what would break first under a doubling of volume. Usually one to two weeks, no code changes.
What we do not do
We do not manage money, take discretionary positions, sell signals, or advise on what to trade. We are not licensed to and have no interest in it. The models are yours; the plumbing is ours.