First...a hearty shout out to JR and everyone else who has contributed valuable resources to our hobby (ROAR, VASL, ASL Players Ratings, AREA, VASL, ASL Armory, Desperation Morale website, VASL, Klass with Klas, and many more I can't think of at the moment). Bravo Zulu.
Regarding a lack of ROAR balance...some of those scenarios are indeed mutts. However, some of them can be far more balanced than they appear. Yes, 14 playings (11-3) is statistically insignificant, but I understand why someone might raise an eyebrow. That said nobody (including me, and I'm a mathematician) is interested in a debate on what can be properly inferred from some skewed initial results.
My point is this...for some of those canine candidates the culprit is a steep learning curve. ROAR can't tell you which 11-3 record is a dog, and which one shows that the initial playings didn't represent a complete grasp of the challenge that was presented. This is not a criticism (there's no way ROAR can do that); it's just a recommendation that you should temper any preliminary ROAR feedback with your own assessment of the scenario. Steep learning curve scenarios will always appear unbalanced initially.