Forecasting that admits what it doesn’t know.
DClaw Forecast started from a simple frustration: most forecasting tools present a single confident number and hide the uncertainty behind it.
The principle
Every number in DClaw comes from your data or a calculation performed on it. When the tool can’t know something — an external shock, a competitor’s move, a one-off event — it says so instead of inventing a value. Every forecast is shown with the range around it, and that range widens honestly the further out you look.
How it works, briefly
DClaw studies the trend and seasonality in your historical series, projects those patterns forward with genuine statistical models, and validates them by backtesting on data the model didn’t see. You can compare models, run scenarios, and read exactly why a forecast looks the way it does. The Forecast Lab goes further: change one setting against a baseline and see what moves, replay past forecasts against what happened, try to beat the model, and work with real public data (observed weather, US GDP and stock indices).
Who it’s for
Learners (students and self-learners) who want to understand forecasting by changing one thing at a time and seeing what moves. And analysts and planners who need forecasts they can explain, test against history and defend, rather than a single unexplained number.
Nothing invented. Nothing hidden. History → Pattern → Range → Decision.