HFT Architecture builds short-horizon price prediction and the trading system around it. The prediction comes first: without it there is nothing to trade on, nothing to bound and no reason to be fast. This role works on that prediction. It is fully remote: you can work from anywhere in the world.
What you will do
- Build and test forecasts of the next mid-price move from order-book data, one market and one horizon at a time.
- Decide whether a forecast says anything the trading rules do not already know, and show it out of sample.
- Keep the future out of the past: leakage controls, embargoed walk-forward validation, a result that survives a second look.
- Take a result apart when it looks too good. Most of them are a data problem.
- Write up what was measured, on what sample, and where the result stops.
What you need
- Research on market data that you carried from a question to a result somebody else could check.
- Statistics you can defend: what the estimate is, how wide its uncertainty is, what would falsify it.
- Python, and enough engineering to make a result reproducible by another person.
- A working picture of an order book: levels, queue, spread, mid-price, what a fill costs.
- Writing. A result nobody can read does not count as one.
Good to have
- Tick or order-book data handled at full depth, with its gaps and resets.
- Market making or execution research: markout, adverse selection, inventory.
- A backtest of yours that turned out to be wrong, and how you found out.
Read before you apply
Two articles show how we write about a measurement, and the form asks about one of them: Latency Arbitrage Won 30 Days of 30 and Lost to Fees and Why Your Market Maker Loses Money.
How we hire
- You apply on this page, with a link and an answer to one question about the article.
- A technical conversation about research you have done.
- One practical task on real data, small enough to finish in an evening.
- A decision, with the reason.