USCF Tournament/Event Performance Rating Equation

After searching around for a while trying to find the actual equation that USCF uses to determine a tournament (or an event) performance rating, I thought I had it figured out.  However, after playing around with different sets of numbers to verify my findings I find myself a little stumped.  The page http://www.mark-weeks.com/aboutcom/aa04l04.htm explains how to calculate the tournament/event performance rating.  At first glance the equations seem accurate, but it can't be the exact equation that USCF uses on it's on Rating Estimator page (http://www.uschess.org/content/view/13317/898/).

Using the set of example ratings from http://www.mark-weeks.com/aboutcom/aa04l04.htm a tournament/event in which a player plays the following ratings (1400, 1650, 1575, 1625 & 1700) would yield the following performance ratings:

-.5 (or 2 of 5 tournament/event points)

(1400+1650+1575+1625+1700+400+400-400-400-400)/5 = 1510

= (or 2.5 of 5 tournament/event points)

(1400+1650+1575+1625+1700+400+400-400-400)/5 = 1590

+.5 (or 3 of 5 tournament/event points)

(1400+1650+1575+1625+1700+400+400+400-400-400)/5 = 1670

Now comes the curve ball... Using the calculations above each individual opponent's rating wouldn't matter as long as their overall rating average was the same (1400, 1650, 1575, 1625 & 1700 should yield the same results as 1590, 1590, 1590, 1590 & 1590).  However if you enter those sets of numbers into USCF's Rating Estimator you get slightly different performance ratings for what are the same overall rating average.

Does anyone know the actual equation that USCF uses to determine tournament/event performance that would yield different results for the same overall opponent rating average?

Its used to be easy, then they changed the k factor.  now its like trying to figure out tax law.

It's been a long time since I've delved into the USCF formulas in detail, but I suspect the rating estimator uses the approximating formulas:

http://www.glicko.net/ratings/approx.pdf

The actual equations used for the official ratings is an iterative process as described here:

http://www.glicko.net/ratings/rating.system.pdf

For completeness, here's Glickman's USCF web page:

http://www.glicko.net/ratings.html

I completely understand how to calculate post event ratings.  However, single event performance isn't the same thing and I couldn't find a reference to single event performance equation in any of the references you linked.

The performance rating formula used on that page can be viewed by looking at the page source, under

function performanceRating

OK, yeah, performance ratings...

What Martin said. lol

(I was amazed that the page source actually had all of the code!)

Thanks!  I'll take a look at it.