A race-selected accelerator is filtering based on race first, then on quality. It's not business investment, it's social entrepreneurship.

Step 1) Throw out 87% of applications randomly. Step 2) Select based on quality. Step 3) Profit?

Compare that to a quality-selected accelerator:

Step 1) Select based on quality. Step 2) Profit!

A race-based accelerator will only outperform a quality-selected accelerator if there's a race that inherently, innately outperforms all other races. Science has pretty well debunked that one.

As to #2, you'd have to be really, really, really stupid to go with a niche accelerator that no one's ever heard of IF given the option of going with YC or TS.

NewMe could barely get a mention in TechCrunch, and, when they did, the social entrepreneurship side story beat out the "here are our awesome companies!" story.