XTech Global Macro Economic Forecasts™ Core

XTech Global Macro Economic Forecasts Core offers institutional investors an alternative to consensus macro forecasts. Powered by machine learning and bottom-up modeling, this dataset delivers one-period-ahead forecasts of key U.S. economic indicators — including CPI, retail sales, and consumer sentiment — days to weeks before official releases.
Unlike broker-driven estimates that rely on lagging or survey-based inputs, XTech ingests high-frequency, real-time data and applies statistical models to deliver forward-looking insights. The methodology leverages textual data such as news archives and market data, providing an independent source of macroeconomic market intelligence. This integration incorporates data including Machine Readable News, Point-in-time Economics data, and Tick History, enabling AI inference by correlating datasets to uncover relationships that enhance forecast accuracy.
The dataset updates continuously with accuracy as new data becomes available, giving traders and portfolio managers the ability to pre-position ahead of the crowd. Successive updates are provided up to the official publication time, enabling investors to re-position ahead of macroeconomic releases and arbitrage consensus views.
With granular component-level forecasts (e.g., CPI gas, shelter, food) and accuracy diagnostics — from confidence intervals to Theil’s U-statistics — users can assess model precision and bias in detail. Delivered via API, the dataset is lightweight (<1MB/day) and integrates into quant workflows.
Use XTech to anticipate macro surprises, arbitrage mispriced consensus, manage risk exposure, and time trades around economic catalysts. Whether you’re running macro strategies or supporting research and asset allocation, this dataset provides foresight.
Asset Class
Commodities, Fixed Income, Equities, Futures, Options, Options on Futures, ETFs, Crypto, FX
Data Classification
Short Term Signals
Data Source Name
lseg_macro_us_core_predictions
Regions
Global, Americas, Frontier Markets, Asian Pacific (APAC)
Data Frequency
Day
Data Dictionary
Name
Type
Description
asof_date
string
Datetime (YYYY-MM-DD HH:MM:SS) at which the prediction was computed in UTC timezone.
date
string
Date-only (YYYY-MM-DD) format of the asof_date timestamp above.
timestamp
string
Business period over which the prediction applies; the measurement period for the economic figure being reported.
identifier
string
Unique identifier for each macro prediction product.
frequency
string
Data frequency (e.g., "m" for monthly).
actual
float
Observed actual value of the target economic release variable.
predicted
float
Forecasted value of the target economic release variable.
difference
float
Difference between actual and predicted values (Actual - Predicted).
ci_lb_95
float
Lower bound of 95% confidence interval of the prediction.
ci_ub_95
float
Upper bound of 95% confidence interval of the prediction.
zscore_diff
float
Z-score of difference (residuals) of the prediction.
zscore_pred
float
Z-score of predictions of the prediction.
mae_cum
float
Cumulative mean absolute error of the prediction.
rmse_cum
float
Cumulative root mean squared error of the prediction.
smape_cum
float
Cumulative symmetric mean absolute percentage error of the prediction.
correlation_cum
float
Cumulative correlation between actual and predicted of the prediction.
directional_accuracy_cum
float
Cumulative percentage of times the forecast correctly predicted whether the actual value increased or decreased compared to the previous period.
signed_correlation_cum
float
Cumulative correlation between forecasted and actual direction of changes, measuring alignment of up/down movements.
sign_accuracy_cum
float
Cumulative percentage of times the forecasted value had the same sign (positive or negative) as the actual value.
theil_u1_cum
float
Cumulative Theil’s U1 statistic (forecast vs. actual accuracy).
theil_u1_bias_cum
float
Cumulative Theil’s U1 bias component.
theil_u1_variance_cum
float
Cumulative Theil’s U1 variance component.
theil_u1_covariance_cum
float
Cumulative Theil’s U1 covariance component.
theil_u2_cum
float
Cumulative Theil’s U2 statistic (forecast vs. naive model).
hit_rate_1bp_cum
float
Cumulative proportion of forecasts within 1 basis point of actual.
hit_rate_5bp_cum
float
Cumulative proportion of forecasts within 5 basis points of actual.
hit_rate_10bp_cum
float
Cumulative proportion of forecasts within 10 basis points of actual.
forecast_bias_cum
float
Cumulative average forecast error (bias).
mae_ttm
float
Trailing 12-month mean absolute error of the prediction.
rmse_ttm
float
Trailing 12-month root mean squared error of the prediction.
smape_ttm
float
Trailing 12-month symmetric mean absolute percentage error of the prediction.
correlation_ttm
float
Trailing 12-month correlation between actual and predicted of the prediction.
directional_accuracy_ttm
float
Trailing 12-month directional accuracy of the prediction.
signed_correlation_ttm
float
Trailing 12-month signed correlation of the prediction.
sign_accuracy_ttm
float
Trailing 12-month sign accuracy of the prediction.
theil_u1_ttm
float
Trailing 12-month Theil’s U1 statistic of the prediction.
theil_u1_bias_ttm
float
Trailing 12-month Theil’s U1 bias component of the prediction.
theil_u1_variance_ttm
float
Trailing 12-month Theil’s U1 variance component of the prediction.
theil_u1_covariance_ttm
float
Trailing 12-month Theil’s U1 covariance component of the prediction.
theil_u2_ttm
float
Trailing 12-month Theil’s U2 statistic of the prediction.
hit_rate_1bp_ttm
float
Trailing 12-month proportion of forecasts within 1 basis point of actual.
hit_rate_5bp_ttm
float
Trailing 12-month proportion of forecasts within 5 basis points of actual.
hit_rate_10bp_ttm
float
Trailing 12-month proportion of forecasts within 10 basis points of actual.
forecast_bias_ttm
float
Trailing 12-month average forecast error (bias) of the prediction.
Python Client
Query Results
AMENITIES |
|---|
24-hour front desk |
Free WiFi |
Parking |
Breakfast |
Pool |
Gym |
AMENITIES |
|---|
24-hour front desk |
Free WiFi |
Parking |
Breakfast |
Pool |
Gym |
AMENITIES |
|---|
24-hour front desk |
Free WiFi |
Parking |
Breakfast |
Pool |
Gym |
AMENITIES |
|---|
24-hour front desk |
Free WiFi |
Parking |
Breakfast |
Pool |
Gym |
AMENITIES |
|---|
24-hour front desk |
Free WiFi |
Parking |
Breakfast |
Pool |
Gym |
AMENITIES |
|---|
24-hour front desk |
Free WiFi |
Parking |
Breakfast |
Pool |
Gym |
AMENITIES |
|---|
24-hour front desk |
Free WiFi |
Parking |
Breakfast |
Pool |
Gym |
AMENITIES |
|---|
24-hour front desk |
Free WiFi |
Parking |
Breakfast |
Pool |
Gym |
AMENITIES |
|---|
24-hour front desk |
Free WiFi |
Parking |
Breakfast |
Pool |
Gym |
More Like This
Loading...

Collapsible text is perfect for longer content like paragraphs and descriptions. It's a great way to give people more information while keeping your layout clean. Link your text to anything, including an external website or a different page. You can set your text box to expand and collapse when people click, so they can read more or less info.
Loading...

Collapsible text is perfect for longer content like paragraphs and descriptions. It's a great way to give people more information while keeping your layout clean. Link your text to anything, including an external website or a different page. You can set your text box to expand and collapse when people click, so they can read more or less info.
Loading...

Collapsible text is perfect for longer content like paragraphs and descriptions. It's a great way to give people more information while keeping your layout clean. Link your text to anything, including an external website or a different page. You can set your text box to expand and collapse when people click, so they can read more or less info.
Loading...

Collapsible text is perfect for longer content like paragraphs and descriptions. It's a great way to give people more information while keeping your layout clean. Link your text to anything, including an external website or a different page. You can set your text box to expand and collapse when people click, so they can read more or less info.
