Census says new home sales rose 6.4% but the change isn't statistically significant. Does that change how you trade the number?

The Census Bureau reported new home sales of 684,000 in August, 6.4% above July. In the same release it says the 90% confidence interval for that change is ±19.5 percentage points, which means the actual change could be anywhere from about −13% to +26%, and that it isn't statistically significant (full report). The construction spending release says the same about its +0.9% rise: the interval includes zero, so there is insufficient evidence that the actual change differs from zero (report). The durable goods report says statistical significance cannot be measured for that survey at all (report).

Most traders react to the headline number against the forecast. The agencies are telling you how much weight the number can bear. That does not mean you should ignore it; a market that trades the headline moves on it whatever its precision. But it is useful to know which releases are sharp and which are fuzzy.

How do you approach this?

• Do you read the confidence intervals or "statistically significant" notes, or only the headline?
• Does a release with a wide margin of error change whether you trade it, or how you size the trade?
• Which releases do you consider the most reliable, and which the least?
• Do you wait for revisions before you form a view?

Trade balance and current account explained is another data guide with caveats worth knowing.

Please share your own approach and reasoning, not calls for others to copy. Posts that promise a direction or sell signals will be removed.