TradeWeave is a free, open-access global trade analytics platform covering 238 countries, 5,115 products, and 29 years of bilateral trade data from BACI/CEPII. It provides interactive visualizations, country profiles, tariff simulations, economic complexity rankings, and trade forecasting tools.

What data sources does TradeWeave use?

TradeWeave combines data from BACI/CEPII (bilateral trade flows), WITS/TRAINS (tariff rates), World Bank WDI (macroeconomic indicators), ESCAP-World Bank (bilateral trade costs), FRED and BLS (freight indices, exchange rates), and FAOSTAT (agricultural trade).

How can I analyze a country's trade profile?

Visit the Country Profiles page and select any of 238 countries to see their export composition, top trade partners, revealed comparative advantage (RCA), Economic Complexity Index (ECI), trade diversification, and 30-year trends.

What is the Economic Complexity Index (ECI)?

The Economic Complexity Index (ECI), developed by Hidalgo and Hausmann, measures the productive knowledge embedded in a country's export basket. Countries that export a diverse set of sophisticated products score higher. TradeWeave ranks all 238 countries by ECI from 1996 to 2024.

Can I simulate the impact of tariff changes?

Yes. TradeWeave's Tariff Simulator lets you model how tariff changes affect import volumes using real product-level demand elasticities and MFN tariff rates for over 1,200 HS4 products.

How far back does the trade data go?

The primary BACI dataset covers 1996-2024 with HS6 product detail. TradeWeave also includes historical trade data spanning over 200 years (1800-present) from academic compilations.

Model methodsForecasting

Trade-flow forecast ensemble

Annual bilateral trade values one to eight years ahead, with a cross-validation-based uncertainty band.

Surface: /forecast

Data

Vintage

BACI 202601 (CEPII), HS96, 1996-2024 (retrieved 2026-06-01)

Tables

data/parquet/bilateral_year/year=*/
data/parquet/country_year_product/year=*/
data/parquet/products.parquet
data/parquet/countries.parquet

Sources

CEPII BACI

Method

The forecast combines Holt's linear trend method, OLS linear regression, and a weighted moving average with drift.

Holt smoothing parameters are selected by a grid search that minimises in-sample mean squared error. The moving average uses up to the latest five observations with linearly increasing weights and a similarly weighted annual drift.

Each model's ensemble weight is proportional to inverse leave-last-out cross-validation MSE, using at most five folds. Equal weights are used when the cross-validation errors cannot determine weights. Negative ensemble forecasts are clamped to zero.

The uncertainty band uses the ensemble's pooled leave-last-out residual error, with fold-local weights recomputed so held-out observations do not enter their own forecast weights. The margin grows with the square root of the forecast horizon.

Total-trade forecasts read the bilateral total directly. Commodity forecasts approximate bilateral commodity trade as the bilateral total multiplied by the exporter's share of exports in the selected HS chapter group. BACI values are converted from thousands of USD to plain USD before modelling.

Equations

Model weightweight(m) = [1 / MSE(m)] / sum_j [1 / MSE(j)]

Ensembleforecast(k) = max(0, sum_m weight(m) * forecast(m,k))

Moving averageWMA = sum_i i * value(i) / sum_i i

Linear trendOLS_value(year) = intercept + slope * year

Uncertainty bandband(k) = forecast(k) +/- 1.96 * residual_SE * sqrt(k)

Validation

Model weights are selected by leave-last-out cross-validation MSE.
The returned backtest reports ensemble RMSE, MAPE when nonzero actuals exist, and the number of folds, using fold-local weights with no look-ahead.
No fixed benchmark accuracy values or external forecast comparison are documented in the named repository sources.

Limitations

Commodity-level bilateral history is estimated from the exporter's commodity share of total exports; it is not observed bilateral commodity trade.
The model uses annual univariate history only. It has no policy, price, macroeconomic or structural-shock covariates.
The residual error pools forecast horizons across the validation folds, so it is an average-horizon error rather than a one-step error.
The endpoint requires at least five historical observations and limits the forecast horizon to eight years.
The confidence band is an internal cross-validation construction, not a documented calibrated predictive interval.

References

Endpoints

Path	Method	Returns	Example
/api/prediction/forecast	POST	The historical bilateral series in plain USD and forecast periods, predicted values, lower and upper bands, model weights, and leave-last-out backtest diagnostics.	Not linkable

Loading chart...

Method

The forecast combines Holt's linear trend method, OLS linear regression, and a weighted moving average with drift.

Validation

Model weights are selected by leave-last-out cross-validation MSE.

The returned backtest reports ensemble RMSE, MAPE when nonzero actuals exist, and the number of folds, using fold-local weights with no look-ahead.

No fixed benchmark accuracy values or external forecast comparison are documented in the named repository sources.

Limitations

Commodity-level bilateral history is estimated from the exporter's commodity share of total exports; it is not observed bilateral commodity trade.

The model uses annual univariate history only. It has no policy, price, macroeconomic or structural-shock covariates.

The residual error pools forecast horizons across the validation folds, so it is an average-horizon error rather than a one-step error.

The endpoint requires at least five historical observations and limits the forecast horizon to eight years.

The confidence band is an internal cross-validation construction, not a documented calibrated predictive interval.

Path

Method

Returns

Example

/api/prediction/forecast

POST

The historical bilateral series in plain USD and forecast periods, predicted values, lower and upper bands, model weights, and leave-last-out backtest diagnostics.

Not linkable