> Cool, as a rough approximation, it is close to (1/50)^6, right? But couldn't the first ball be any ball? That doesn't really matter. So wouldn't it just be 1/

> Cool, as a rough approximation, it is close to (1/50)^6, right? But couldn't the first ball be any ball? That doesn't really matter. So wouldn't it just be 1/50^5 for 6 balls? It's very, very far, between non-replacement (~1.36x), order (5!=120x), the fact that final ball is out of 20 instead of out of 50 (2.5x), and the number of possible sequences which would be notable (at least 16x, but actually probably a fair bit more because recognizable patterns are pretty broad). The upshot of which is that estimating at 50^6 is off by multiple orders of magnitude...