Principles
Out-of-sample or it didn't happen
No strategy goes live without 12+ months of walk-forward results and 60 days of shadow trading.
Retire in public
When a strategy stops working, we say so — and we say why. The post-mortem stays on the site.
Risk before return
Position sizing is volatility-scaled. No leverage cocktails, no martingale, no doubling down on losers.
Receipts, always
Audited broker statements are available on request for every live track record.
Research methodology
- 01
Hypothesis spec
Every strategy starts as a one-page written hypothesis: the inefficiency, the regime it depends on, and the conditions that would falsify it.
- 02
Clean-data backtest
Backtests run on tick data with realistic spreads, slippage, and commissions. Survivorship bias and look-ahead are tested with adversarial date splits.
- 03
Walk-forward + monte carlo
Parameters are optimised on rolling windows. We require stable Sharpe across folds and acceptable tail behaviour in 1,000 monte-carlo trade resamples.
- 04
Paper-trade in shadow mode
A minimum 60-day live paper run on the same data feed as production. We compare expected slippage to realised slippage before any capital is committed.
- 05
Capital-allocated live
Live capital starts at the smallest tradable size and scales only after the live track record matches the walk-forward distribution.