Just to add to Maverick's comments, this marketing piece is about equal weighting vice cap based weighting. It in fact specifically states that the effect is most probably based on cap effect, not random selection. Not to mention its not peer reviewed nor does it give any info needed to check the strength of the effect, like p value (you didn't realize it but you pointed out the need for a p value with your "99%" of the time comment, which was not supported, and would in fact be shocking to see). Nor does it rule out data snooping, for example does this hold for 45 stocks, 65, or did they just run it with a bunch of sizes until they got the most striking result and publish that? Your level of trust in someone because of their title is also striking. Most likely this was ginned up to meet the authors goals, which obviously didn't extend to impressing anyone with more than a passing familiarity with stats.
