ORA is a research project studying probabilistic forecasting of financial-market quotations using AI and a metric developed within the project. It uses publicly available market data and records model outputs for selected instruments, time horizons, and operating modes.

Each output is an estimate with a probabilistic character, not a certain prediction. Current estimates are assessed in relation to the prior record for the same model configuration and compared with subsequent market quotations. This makes it possible to examine how often, and under which conditions, an observed pattern was reproduced—and where it was not.

Published statistics are meaningful only together with their observation period, sample, calculation method, instrument, mode, and relevant assumptions. ORA's research focuses on testing the metric against historical and subsequent data across changing market regimes, rather than presenting one result as universally applicable.

Historical, simulated, or illustrative observations do not predict future outcomes. A model-indicated level does not by itself show that a trade was executable or profitable after spreads, fees, slippage, and other costs.