I have a question 
When rating a tournament, is it rated by round or all at once? 
For OTB events, all games for each player in a section are rated at the same time using the formulas described in the paper by Mark Glickman and Tom Doan, which is available in the ratings section of the USCF website.
Note that if a player has more than one pairing # in a section, for example as a result of a re-entry or an event with multiple schedules that are merged together, all games under that USCF ID are rated as one block of games.
Correspondence games are rated as the individual completed games are reported, since a correspondence section could take a year or longer to complete.
but in what order are the players rated?
I can’t say for sure about the current programming, but I assume it probably does them in pairing # order. That’s what I’m doing in the new implementation.
However, as I understand the ratings formula (which I hopefully do at this point, having studied it for several weeks, including finding a typo in the paper on the USCF website), the order in which players are rated within a section is irrelevant.
The ratings process has several stages it goes through before it computes the new rating, mostly to come up with a reasonable pre-event rating for unrated and provisionally rated players. (That’s a bit of a simplfication of the first several stages. Unless you’re a mathematics major or incredibly masochistic, don’t try to read the paper on the website.)
However, once a pre-event rating has been chosen for everyone, the order in which players are processed should have no impact on the post-event rating.
Are not unrated players rated first? I’m not sure if provisional ratings are next or not.
Steve
Since the time of the 1980’s, the unrated and the provisional players are rated first: that was the party line back then. Many different rating theories have come and gone, not sure what the federation has as the ethos of the rating department. The empirical evidence says ‘yes they still do’, has not been re-affirmed in some years. With the demand of the ‘USCF executive board’ for the wish of rating inflation: like the ‘failed’ demand of the ‘2 point activity points per-USCF game’.
If you want to know EXACTLY how it works, I refer you to the paper by Mark Glickman and Tom Doan, available online at
math.bu.edu/people/mg/ratings/rs/rs2.html
Here is the summary given in that document of the five steps that the program goes:
The first step sets temporary initial ratings for unrated players.
The second step calculates an ``effective’’ number of games played by each player.
The third step calculates temporary estimates of ratings for unrated players only to be used when rating their opponents on the subsequent step.
The fourth step then calculates intermediate ratings for all players.
The fifth step uses these intermediate ratings from the previous step as estimates of opponents’ strengths to calculate final post-event ratings.
I’m not a PhD mathematician, though I did have two years of math beyond calculus in college.
The paper does not appear to specify doing unrated or provisionally rated players ahead of players with established ratings, though steps 1 and 3 apply only to unrated players.
I believe the point of steps 1 through 3 is to ensure that a reasonable pre-event rating is used for all players, including unrated and provisionally rated players.
The two stage iteration in steps 4 and 5 ensures (among other things) that the effect of the bonus formula is taken into account when computing the final rating of the opponents of someone who received bonus points as a result of a much-better-than-expected performance.
If you have questions regarding the theory behind the ratings formula, I suggest you contact the Ratings Committee.