We view financial markets as dynamic state estimation challenges.

We trade perpetual futures: instruments that track a price, never expire, and exchange a funding payment between long and short every few hours.

The work is the loop around them. Form a hypothesis. Test it on recorded data. Simulate execution with fees, slippage and funding. Deploy. Measure. Revise or retire.

We build that loop ourselves — market data, research tooling, backtesting, execution, risk and monitoring. Nothing in the order path we did not write.

We do not manage outside money, licence the software, or disclose what we trade.

We do not trade other people’s money. There is no fund, no mandate, and no outside account whose loss we can socialise.

Every position is the firm’s own capital. That is not a slogan: it is the constraint that sets the risk standard. When a strategy fails, the loss is ours. When a control is weak, we are the ones who live with it.

We take no outside money and we never will. Nothing on this site is an offer of investment services, and nothing we build is sold as a product for other people’s capital.

We hire researchers, engineers and operators who want the whole problem — from first question through production systems — and who do their best work with autonomy and a high standard of evidence.

Technology is not a support function here. It is how strategies exist at all. If you prefer handing a signal to another team and moving on, this firm is a poor fit. If you want to own it end to end, we would like to hear from you.

Venues, banks, and market-data and infrastructure providers deal with one counterparty that writes things down. Every order is traceable to the decision that produced it, and the book of record is reconciled against your records rather than our expectation of them.

Onboarding, due diligence and vendor conversations each have their own address, and each is read by a person. We review carefully and decide plainly.

Each row traces to a record we keep. Where a fact is not yet verified it appears as a slot rather than a guess.

A result is wrong until it survives. We require purged cross-validation, walk-forward testing, and an explicit count of every trial we ran. The count is what most work omits, and it decides whether a number means anything.

Costs are modelled before a result is believed, not after: taker and maker fees, spread, documented slippage, funding, and maintenance margin. A strategy that only works gross is not a strategy.

The head quant certifies that a strategy meets the firm’s standard of evidence before it receives capital. The principal is the sole signer. Those roles are deliberately separate; neither can quietly become the other.

Nobody outside audits us, because there is nobody outside. That raises the bar rather than lowers it. Controls are written down, run in continuous integration, and a merge that weakens one fails rather than warns.

Every order is traceable to the decision that produced it. The book of record is append-only and reconciled against the venue, not against our own expectation of it.