Research
Notes from the lab.
How we test ideas, why most of them fail, and what we have learned about building systems that hold up out of sample. Method, not marketing: these notes discuss process and contain no performance figures.
Look-ahead bias: the quietest way to fool yourself
Most backtests that look too good are quietly using information that did not exist yet. How we keep every test honest about time.
Read →When a rule stops working
Every rule eventually meets a market it was not built for. How we watch for that moment, and what we do when it comes.
Read →How a rule earns capital
Walk-forward windows, placebo tests and full trading costs: the three hurdles every idea must clear before it trades.
Read →Exits are a strategy too
Why we study when to leave a position as carefully as when to enter it.
Read →Why risk has a veto
The independent layer between every signal and the market, and what it is allowed to do.
Read →