Field notes on reading markets.
·6 min read·correlation · risk · GroupiFi
Correlation windows that actually matter
Two stocks can look almost unrelated over a full year and still move together nearly every afternoon. Which window is right depends on how long you plan to hold the trade.
·8 min read·backtesting · TestaFi
What makes a backtest real
Look-ahead bias, survivorship bias, and overfitting can make an ordinary strategy look exceptional. Here is how to check a backtest before trusting the equity curve.
·5 min read·probability · market structure
Reading hourly base rates without fooling yourself
Historical base rates can be useful, but only if you filter them by regime and check the sample size. A short guide to using probability grids without cherry-picking.