This sounds like an easy enough thing for you to track and post on your own with a fictitious BR to support your claim.
Take any closing line you want in MLB, as long as you can get a few thousand plus results. you can even tweak it as you see fit.
EX: look at results of the previous 5,000 division dogs that closed at +125 and faced LHP. (you get to pick the scenario, really doesn't matter to me).
i'm guessing you're going to be hard pressed to discover a data set large enough to be considered statistically significant without the W/L record being within a few percentage points of the implied probability as expressed by the closing line.
unless you are not using historical data to drive your implied probabilities. if you're using market prices for that piece, they are built on models using historical data so either way...
so identify a scenario where you think you have a consistent edge and Kelly the ish out of it. then you can clearly validate your assumptions that Kelly is awesome.