If I have heard it once I have heard it a thousand times over 20 years of league play. i.e. look at my scores and why have I been moved up (or down). In reply come our gods(posters) and say loftily well you should have known from your scores you are playing above (or below) your rating. Easy for you to say. But so long as you say the scores reflect the rating you will be incorrect. The answer to the question is that the rating reflects the interpretation placed on the scores by the USTA and no one knows the algorithm. The scores are merely facts and what counts is how the USTA interprets those facts.

Ok, my understanding of how the NTRP calculation works is this. In a singles match between two computer-rated players each player enters the match with a dynamic NTRP rating value. For purposes of discussion, let's assume this is a 3.5 league match and the dynamic rating of player A is 3.25 and the dynamic rating of player B is 3.45 so that there is a dynamic rating difference of .2 between the two players.

The NTRP algorithm has programmed into it an expected game differential for a .2 dynamic rating difference. Again for purposes of discussion assume that the expected differential is 6 games, which means the higher rated player would be expect to win 6 more games than their opponent which could be accomplished by scores such as 6-3/6-3, 6-2/6-4, 6-2/7-5 or 6-1/7-6.

If the actual score differential was less than 6 then the lower rated player's dynamic rating would move up by some predetermined amount and the higher rated player's dynamic rating would move down by some predetermined amount. If the actual game score differential was more than six then just the opposite would occur and if it was exactly six then no one's dynamic rating would change. Of course if the expected differential was a fractional number then both player's dynamic ratings would change every match.

The actual winner or loser of the match is not considered in the algorithm. For example there are match game scores where you lose more games than your opponent but win the match, such as where you win a match 0-6, 7-6, 1-0. However, extending this example, if your expected game differential before the match was +6 and the actual game differential was -4 as in my example, you would win the match but see your dynamic rating go down because you did not win as many games as you were expected to win by the algorithm.

I would love to hear how you believe the NTRP actually works since you are so adamant in proclaiming that you know how it doesn't work.