With such tests, it is all about the resulting p-value. If the p-value is really low you reject the null hypothesis that the distribution of your data is similar to the statistical distribution you compare your data to. Typically, people use a p-value of 0.10 or lower as a threshold that you can reject the null hypothesis that your data distribution is the same as the distribution you compare it to.

I can't comment on that. However, before pronouncing the JB test dead, it should be benchmarked against its competitors. In isolation, stating a test is good or bad is not informative until you can actually rank that test against its alternatives. And, I would venture that the JB test on a relative basis holds up just fine. In the end, all tests have their flaws, but many are still pretty informative. I would think the JB one falls into that category.

I beg to differ on several counts. Welch is specifically earmarked for both unequal variance and unequal sample if necessary. en.wikipedia.org/wiki/Welch%27s_t-test. The t-distribution is leptokurtic just as the one of stock returns. Often stock returns are not that far off a Normal distribution anyway, especially if factoring for fatter tails as in the t distribution. The MW null is that when the data are combined together, each sample has the same average rank... which is a nonparametric correspondence to comparing nominal average values.

When you construct independent variables that are simply recombinations of existing ones, the plain arithmetic suggests you would be more likely to break multicollinearity and exogeneity assumptions. I am unclear why such position would be controversial. Your comment suggests my question as phrased was somewhat insulting. I beg to differ. I did not claim at all that people don't know what they are doing. I am simply seeking diagnostic testing interpretation ammunition to figure out how to get out related conundrum.

Stephan Kolassa, thanks for your informative comment. I added a small paragraph to incorporate your suggested additions of other forecasting and estimating measure metrics.