FIDE Adjusts Ratings For 350,000 Players In Major Change

Today, FIDE, the chess world-governing body, made a significant move in an attempt to combat rating deflation and improve the accuracy of the ranking system by adjusting the rating for hundreds of thousands of players. Nonetheless, the extensive overhaul has not escaped criticism.

Last December, Chess.com reported on the drastic changes proposed by FIDE's Qualifications Commission (QC) with the objective of refining the accuracy of the rating system and addressing the purported issue of deflation. The changes came following recommendations by chess statistician Jeff Sonas after an announcement in July last year.

The proposals were approved by the FIDE Council on December 14, with one adjustment: Instead of January 1, the four changes came into effect today, on March 1:

• An adjustment of up to 400 rating points
• An adjustment of the rating floor
• Changes in the initial rating
• Reinstallment of the 400-point rule

If you are FIDE-rated and have a rating of less than 2000 you may, along with approximately 350,000 others, want to check your new rating. As many as 85% of all players will see a one-off change of up to 400 points added to their rating, while there are no adjustments for the close to 70,000 players rated above 2000.

The goal, according to Sonas, is to compress the ratings from 1400 to 2000 without affecting the order of the players, only spacing them. The formula is:

(0.40) x (2000 - Rating)

The formula means players will have somewhere between 0 and 400 Elo points added to their ratings. Players rated at exactly 1000 would get an increase of 400 points, while a 1500-player would increase their rating by 200 points. A player rated at 1950 will only get 20 points added.

Happy free rating day!! 📈 #FIDE pic.twitter.com/6pIanDiUe1

— ChessDojo (@chessdojo) March 1, 2024

While the changes were approved with few objections, there have been critics. One was Otto Milvang, an international arbiter, Chairman of the Arbiters Commission in the Norwegian Chess Federation, and a member of FIDE's Technical Commission.

In a response running over 12 pages to Sonas' analysis, Milvang said the changes would not solve the problem of rating deflation. "A rating floor [of] 1400 [divides] the players into rated and unrated players, and in the 1400-1600 range the rating is highly unreliable," he writes.

"The paper also shows that the rating floor is a border that creates unreliable [ratings] for the lowest rated players. It also destroys a natural rating [distribution across] the players. In the proposed model, [the] rating floor is removed, and the simulation shows that this only has advantages."

He pointed out that many young players being underrated is the challenge, noting that they decrease the rating of established players. "The proposal from QC does not solve this problem. We will still have talented young players that climb up the rating at the expense of established players."

Milvang also notes that ratings in many cases are "closed ecosystems" and "evolve differently in different countries." As an example, the mean rating in Germany is 1803, while in India it's 1298.

"[The table] shows a huge difference in rating [between] players in Germany and India. It is impossible to say if this difference is real or artificial. Will a rating compression help? No!" he notes.

Moreover, Milvang criticized the increase of the rating floor to 1400 after it was reduced to 1000 in 2012. He emphasized the need for mechanisms to prevent rating deflation rather than solely relying on rating adjustments. 

"Sonas claims that decreasing the rating floor will start pulling large amounts of rating points away from the established pool. This is true also for the QC proposal, so the problem is not that low rated players increase their playing skills, but that there are no mechanisms to prevent deflation."

He noted that the Norwegian chess community had a rating system down to 600 until 2017 when FIDE ratings became the main one. 

"Already in 2017 it was clear that the fact that it is a rating floor with 2/3 of the players with rating and 1/3 without rating is challenging. The floor gives [a] high degree of uncertainty within the first 300 rating points (1000-1300)."

In his proposal, Milvang suggested revising the formula for rating calculation, eliminating the rating floor, and incorporating adjustments based on the number of games played.

Milvang said the change "will reset the current deflation in the rating system, however it will not stop the deflation for the future."

"FIDE [needs] to introduce a long-term rating regulation that reverses the deflation, and with a rating system that covers all chess players," he wrote in his conclusion.

2100 FIDEs looking for their rating adjusment pic.twitter.com/fTRi49MvCR

— Strong Chess (@thestrongchess) March 1, 2024