How more GDP models are better than one

• Third quarter GDP of 3 percent beat estimates of 2.8 percent by CNBC's Rapid Update and the Atlanta Fed's 2.5 percent.
• Over the three quarters of this year, the average error rate for CNBC has been 0.2 percent, half that of the Atlanta Fed's GDPNow.
• No single forecaster's model has proven to be the best for tracking GDP.

A worker installs parts on the engine of a Chrysler SUV as it moves along the assembly line at the Jefferson North Assembly Plant in Detroit Michigan.

Geoff Robins | AFP | Getty Images

CNBC's Rapid Update estimate of GDP has continued to outperform the widely followed Atlanta Fed's GDPNow.

For Friday's third quarter estimate of GDP, which came in at 3 percent, Rapid Update estimated 2.8 percent, compared with 2.5 percent for the Atlanta Fed. Over the three quarters of this year, the average error rate for CNBC has been 0.2 percent, half that of GDPNow' s error. Since 2016, the CNBC Rapid Update has had an average quarterly error rate of 0.41 percent, compared with 0.53 for the Atlanta Fed. Since Rapid Update began in 2014, it has been modestly more accurate by .05 percent.

CNBC Rapid Update is compiled by Moody's Analytics from the tracking forecast models of up to a dozen leading Wall Street forecasters. It is updated every day economic data is released that bears on Gross Domestic Product. Behind the concept is that no single forecaster's model has proven to be the best for tracking GDP, but that an average of the models will offer a better estimate.

The results are updated here on days when data is released that affects GDP.

WATCH: What fueled the economy to 3% GDP