How accurate the Oliva fair value model is

Every week we test our fair values against real DLD transactions the model has not seen. This page shows how wrong we are, overall and by area, dated and measured on held out data.

No published runs yet

The first measured backtest has not been published yet. This page fills in automatically after the next weekly run.

When the gap is smaller than our error

When a project's price gap is smaller than our measured model error for its area, we say fairly priced instead of pretending precision. The directional verdict is suppressed inside the error band and the project carries a within model error label.

Read the full methodology