Let's start with the concept of regression to the mean. In simplified terms, very high scores are extremely rare. Two very high scores are much, much rarer. So if one score is extremely high, the likelihood is that the next score will be closer to the mean (in this case, lower). This is why, when looking at achievement scores predicted from cognitive scores, the range of expected scores is fairly large for cognitive scores at the extremities.
Secondly, the cognitive index most relevant to the WJIV Applied Problems subtest is the WISC-V QRI. The difference between the QRI and AP is only 15 SS points--not a huge difference when you consider regression to the mean, and also when you consider that these are comparisons across different (and not co-normed) instruments.
Thirdly, consider the way standard scores and norms are built. The 2,200 children age 6-16 in the WISC-V standardization sample were divided such that each 3-month age bracket had about 67 children. (The WJIV sample was similar, with just under 4,000 school-age children K-12.) Think about how many children a 135 represents. Roughly 2/3s of a child from the standardization sample corresponds to all the scores from 130 up. The distinctions finer than that had to be derived from decisions in statistical analysis of the standardization data. Now, to be clear, I have no problem using these norms, even knowing how little raw data goes into the extremes of the bell curve, mainly because these are the best tools we have at the present time.
Finally, you are talking about a very young student, most likely in the beginning of 1st or 2nd grade. Expectations for age-peers are very low, with a bell curve that is rather tightly packed in the middle, and below the middle, so a child who has mastered a few arithmetic skills (pretty much anything beyond single-digit addition and subtraction), and understands how to apply them, will rapidly climb the bell curve into its upper extremes. Also, based on the Broad Math score you report, I assume that the other math subtest scores were in the 130s, which would be exactly where the QRI is.
From an advocacy standpoint, the simplest approach might be to present the QRI and AP scores, both of which are in the >99 %ile. The presumptive 20 pt gap between AP and math calculations would support the contention that this is native math ability, and not hot-housing, as one would imagine that it is easier to drill procedures into children, than conceptual understanding.
