The edge is in the process.

Markets are noisy and mostly efficient. What is left over belongs to whoever measures it most carefully — and then executes without flinching.

01 — Research

Begin with a mechanism, not a backtest.

A pattern found by searching hard enough is not a discovery; it is a coincidence with good publicity. So every project starts with a written account of why a price should move: a behavioural bias, a structural flow, a constraint that forces someone to trade.

Only then do we look at data. The prediction — direction, horizon, magnitude — is recorded before the first test runs, which makes it possible to tell a real result from a story told afterwards.

Point-in-time everything. No revised fundamentals, no restated earnings, no survivorship in the universe.
Costs from the first run. A signal that only works at zero slippage is not a signal.
Paper trail. Every idea keeps its hypothesis, tests and failures in a permanent record.

02 — Risk

We size from distributions, not conviction.

Drawdown is the brief
Portfolios are built to a loss budget first and a return target second. When the two conflict, the budget wins.
Correlation is a guess
Diversification assumed in calm markets tends to arrive late in a crisis. We test the portfolio under adversarial correlation, not historical average.
Capacity is a constraint
Every strategy has a size beyond which it eats its own edge. We measure that number and stop below it.
No single point of failure
Independent risk oversight reports outside the investment team, and can cut exposure without asking permission.

03 — Execution

Microstructure is part of the model.

The last mile decides whether a strategy's theoretical return survives. Our execution research sits with the portfolio team, and its output feeds straight back into what we are willing to trade.

We measure implementation shortfall trade by trade, attribute it to spread, impact and timing, and treat any persistent gap as a defect to be fixed rather than a cost to be accepted.

What we deliberately do not do

Trade on discretionary override. A human can halt a model; a human cannot improvise inside it.
Launch a product to meet demand for a product.
Publish return forecasts. We do not know them.
Take capital we have no capacity to deploy.

Who does
the work.

Small teams, long tenures, and an unusually low tolerance for untested claims.