Forecast methodology
Every forecast on Forecast Economy is produced by a statistical model trained on the official historical series of the indicator. We do not fit results to expectations or add expert assumptions: the forecast is determined by the source data and the algorithm. Below is the full calculation path — from series preparation to the confidence interval — and the indicators we deliberately do not forecast.
Principles
We forecast only series from official primary sources. The forecast is produced from the series history by a fixed algorithm and can be reproduced from published data. Alongside the central estimate we show a confidence interval that widens with the forecast horizon.
Calculation steps
The series is checked against the source and aligned to its frequency. We then assess trend, seasonality, and stationarity (augmented Dickey–Fuller), choose a stable transform — levels, differences, or log differences — and fit a statistical model: regression on lagged values with several training windows; for seasonal series, ARIMA and SARIMA-family models. Window estimates are combined with weights inverse to their dispersion, the forecast is mapped back to original units, and a confidence interval is built around it.
Model by indicator type
Monthly indicators (wages, rates, money supply, budget, trade) use a general autoregressive model; quarterly positive series (GDP and its components, exports, imports, external debt) use a model on log differences; series that change sign (current account, balances) use a model on level differences; inflation uses a combined model with pronounced seasonality. Derived series (annual totals, year-on-year and period-on-period change) are obtained from the base-series forecast, so they stay consistent across chart modes.
What we do not forecast
Statistical extrapolation loses meaning where current news, not series inertia, drive the path: exchange-traded quotes and indices, intraday FX, cryptocurrencies, and daily or intra-week series. For those we publish full history without a forecast line, and the forecast toggle stays inactive.
Cross-country GDP ranking
On the world map and in the country ranking, GDP and GDP per capita in current US dollars for foreign countries come from the IMF World Economic Outlook annual estimate. For Russia the ranking uses a platform calculation: Rosstat’s annual GDP in rubles converted to US dollars at the Bank of Russia average annual official exchange rate. Opening a country leads to the fund’s series. Years the fund publishes as projections, including the current calendar year, appear on the map as Outlook estimates. Russia’s quarterly actuals are published by Rosstat and collected in the GDP category of the Russia section.
Limitations
The forecast relies on stable patterns in the past and may diverge from outcomes under economic shocks, changes in monetary or fiscal policy, or revisions of historical data by the source. Materials are informational and are not personalized investment advice.