Presumably one could use the mean and standard deviation listed on the SBACC site to determine the percentile of a given scaled score at the given grade level. And yes, a perfect scaled score likely means 99th %ile in most cases, except where the ceiling of the test is too low to spread the last few percentiles.
For advocacy purposes, perhaps you could consider combining data from his scaled score and the level 3 score expectations for the level of instruction in higher grades. For example, if he is entering sixth grade in the fall, with a max obtainable scaled score on last year's grade 4 test of 2663, he would have scored in the upper quartile and at level 3 for end of grade eight in ELA. (Except, of course, that this close to the top of the test, the standard error gets a bit unwieldy, so you probably wouldn't want to interpret the score within an inch of its life. So maybe a slightly less generous grade placement. But still high.)
https://portal.smarterbalanced.org/library/en/2015-16-summative-technical-report.pdfsee page 5-5, table 5-3 (the 160th page of the pdf) for the highest obtainable scaled score, to make sure that you are not placing undue interpretive demands on tail scores.
http://www.smarterbalanced.org/assessments/development/percentiles/compare your kiddo's scaled scores to scores in the upper quartile of higher grades.
http://www.smarterbalanced.org/assessments/scores/compare to level 3 or 4 scores in upper grades.
These are, of course, only one source of data, and don't take into account whether he has received content-specific instruction (such as prerequisite math), or has gaps in essential skills that weren't sampled by the test.