Thanks, everyone. I have a blast putting this stuff together. I've got ideas for more content and I'm always open to input.
The Projected Records thing is kinda fun, but should be taken for what it is - an algorithmic approach to predicting something that doesn't always follow "rules". When I back-tested it against last season, it was surprisingly accurate, though, so I'm interested to see how things match up. Also, since it uses Coach Effect data, if a coach is new to a level, I don't have CE data and I assign a neutral (0) value. Obviously, a good coach will have a positive effect and could overachieve in this sense; likewise, a bad one could underachieve with a negative effect.