Oh so it's so "outlandish" to think that someone scoring a 39 is more likely to do better on Step 1 than someone scoring a 24? In other words, since you think there is "no predictability," you believe someone scoring a 24 on the MCAT is equally likely to pass the Step 1 as that of someone scoring 39. If that were really the case, then there would be no point in taking the MCAT to being with...
Btw, here are multiple studies showing that, indeed, there is a statistically significant correlation between MCAT and Step 1 performance.
http://www.internationalgme.org/Resources/Pubs/Donnon et al (2007) Acad Med.pdf
It might not be a strong correlation (it's cited as a "small to medium" correlation in that meta-analysis*), but it is statistically significant. So no, I don't find my claim outlandish. Doing well on MCAT is associated with doing well on Step 1.
*To be precise, the predictive validity coefficient for the MCAT predicting Step 1 is r=.6. That's not low.