An interesting observation from the FIDE rating data was that the lower-rated player outperforms expectation at all rating levels. That is, the Elo system predicts an expected score based on rating difference, which the higher rated player would have to match to keep their rating. But in real life, the lower rated player actually scores much better than the prediction (over a large sample). As far as I know, a similar analysis has not been done for USCF ratings, but a similar dynamic may be present.
An interesting observation from the FIDE rating data was that the lower-rated player outperforms expectation at all rating levels. That is, the Elo system predicts an expected score based on rating difference, which the higher rated player would have to match to keep their rating. But in real life, the lower rated player actually scores much better than the prediction (over a large sample). As far as I know, a similar analysis has not been done for USCF ratings, but a similar dynamic may be present.