The Ratings Committee (RC) had another busy year. We proposed two changes to the rating system that were approved by the Executive Board (EB) in February, 2014, as well as addressing several other issues in an advisory role. Some of the ongoing tasks, such as developing a method to connect the Quick Chess (QC), blitz, and regular rating systems, have been hindered by the lack of a testbed server at the USCF office on which to perform the testing. We have been assured that the backlog work can proceed imminently, so we can expect progress on some of our outstanding tasks. In this report, we describe our main achievements over the year, several of the minor issues we have addressed, and provide a sense for some of our upcoming work in the next few months. Each year the RC performs a set of diagnostic analyses to monitor trends in the rating pool. Overall rating levels deflated from the mid-1990s through 2000 when rating floors were decreased by 100 points without a counteracting inflationary mechanism. With the new rating system implemented in 2001, ratings started to re-inflate. The RC.s goal has been to re-inflate and then maintain rating levels roughly where they were at the end of 1997. Our analyses have focused on players with established ratings who have been active over the current and previous three years and who are aged 35-45 years old in the current year. Based on our analyses of rating changes, current ratings are now at the levels that they were at the end of 1997. However, with the change in the .effective N. formula and K-factor formula made effective in May 2013, we have estimated from simulation analyses that players on average should experience noticeable inflation relative to the formulas previously in place. Even though ratings are now currently at appropriate levels, the concern of ongoing rating inflation due to the new formulas has resulted in our proposing to change the bonus point threshold from B=8 to B=10 in the rating formula. The EB has approved this motion, and the change went into effect in late March 2014. In late January 2014, we reviewed a particular aspect of the rating system in which players. ratings always increase or decrease by at least 1 point after an event is rated. This is an artifact of the rule that ratings are rounded up when the post-event rating is greater than the pre-event rating, and rounded down when the post-event rating is less than the pre-event rating. Interest in this issue was motivated by players who were defeating much lower-rated players in one-game events, but still gaining 1 rating point per event based on this rule. Players can take advantage of this feature of the system by repeatedly winning in 1-game events against much lower-rated opponents and increase in rating by potentially large amounts over many event. The reason this rule existed in the first place was that an earlier EB felt it was important for players to see their rating change after an event even if the performance in the event would have resulted in no rating change. The solution proposed by the ratings committee chair was to change the rating system updates from integers to floating point precision, and store ratings as floating point values. The underlying idea is that if a higher-rated player defeats a much lower-rated player, then instead of gaining 1 point the player would only gain a small fraction of a point. Along with storing ratings as floating point values, ratings would still be presented as rounded to the nearest integer, though players would have the ability to see their ratings as floating point values on the MSA pages online. The motion to store ratings and compute rating changes in floating point accuracy was approved by the EB in February 2014. Mike Nolan is in the process of implementing the motion, and performing tests to make sure the updated system is accurate before the floating point system is rolled out. The USCF has partnered with ICC (internet chess club) and chess.com so that some games and tournaments on these platforms will be under USCF jurisdiction. The USCF and these partnering groups agreed that USCF-sanctioned games/tournaments with quick and blitz time controls will be rated under a newly created online blitz and online quick system. The systems will be set up for now to follow the over-the-board rating computations. The USCF office asked the ratings committee for input on initializing unrated online ratings. After RC discussion, we agreed upon a procedure to initialize the different rating systems. To initialize an over-the-board/online rating in the blitz/QC/regular system, we suggested using the following order of precedence: . A rating that matches the over-the-board/online dichotomy first (e.g., for initializing over-the-board ratings, use an over-the-board system); if that.s not possible, use a rating from the other branch . Within over-the-board or online systems, choose the blitz/QC/regular system closest to the one being initialized . When initializing QC, blitz is closer than regular . Have the rating be based on 10 games when initializing based on regular or blitz ratings; have the rating be based on 0 games when initializing based on QC ratings. As an example, to initialize an online QC rating, the order of precedence would be to choose the rating (if it exists) from online blitz, online regular (if/when implemented), over-the-board QC, over-the-board blitz, and over-the-board regular. The RC chair was asked to review the FIDE-to-USCF conversion in September, 2013, for the purpose of initializing unrated USCF players who have FIDE ratings. The most recent conversion was performed in 2008, so the formulas being used were out of date. Based on all players with both FIDE and USCF ratings at the time of the analysis in September, the following conversion was determined: USCF = 180 + 0.94FIDE, if FIDE <= 2000 USCF = 20 + 1.02FIDE, if FIDE > 2000 Similarly, for players who are under 20 years old, the following conversion was determined: USCF = 560 + 0.76FIDE, if FIDE <= 2000 USCF = 80 + 1.0FIDE, if FIDE > 2000 The first set of conversions would be used in general for initializing USCF ratings from FIDE ratings, and the second set would be used for updating USCF ratings from FIDE youth events. The RC was asked to consider how to initialize an unrated USCF player who has a FIDE title but not a FIDE rating. Players can earn candidate master or FIDE master titles at events such as the World Youth Championship without actually earning a FIDE rating, so it is important to assign these players an initialized rating when they first compete in a USCF event. After some discussion by the RC, we agreed with Mike Nolan.s recommendation to assign the following initial ratings based on the earned FIDE titles: CM 2000; FM 2100; IM 2200; GM 2400; WCM 1800; WFM 1900; WIM 2000; and WGM 2200. After further discussion, we agreed that these assigned ratings would be based on 5 games in all cases. One of the tasks charged to the committee back in November 2009 was to propose changes to the quick rating system that would enable ratings to track players' abilities more reliably. Because many players that play quick chess events do so infrequently, quick ratings are often stale relative to players' current abilities. We have also recently noticed a substantial divergence between regular ratings and blitz ratings, even though the blitz system is only one year old. The proposed remedy to this problem is to increase the impact of regular (slow) game events on quick and blitz ratings, as well as increase the K-factor for quick events in general, and even more so as a function of infrequent play. The current system already has quick ratings impacted by games played in the "dual-rated" time controls of G/30-G/60, but the proposed changes will attempt to expand the scope of the impact. The development of this system has been hindered by the lack of availability of a testbed server on which to perform simulation analyses to test the proposed connections among the different systems. We have been informed that the testbed server is now running, and we should imminently be able to perform the required testing. While waiting for the testbed server availability, the RC has come up with an alternative approach to the original .activity point. method derived from the original proposal. The new method proposes to update QC ratings from regular events using a value of K given by K = 800/(1.5(N* + A r^t) + m) where N* is the effective number of games for the regular system, m is the number of games in the current event, A is an activity measure for the QC system, t is the time since the last time the player has competed in a QC event (measured in, say, months), and r is a constant (to be estimated from simulations) between 0 and 1 that discounts activity due to the passage of time. This activity measure, A, measured at the last time the player has competed in a QC event, is defined recursively as A = m.qc + r^s A.prev where m.qc is the number of rated games the player played in the last QC event, r is as defined above, s is the time in months since the previous event, and A.prev is the activity measure prior to the last event. In our development of this system, we will optimize the parameter r, as well as explore the impact of the multiplier 1.5 in the K-factor formula above, on the predictability of QC ratings when we perform our testing.