"Everyone Has Lost Points": A Rating Deflation Deep Dive
By Cameron W. · Wed, 07 Oct 2026
Ian Nepomniachtchi didn't set out to make news on a Tuesday livestream, but his aside landed squarely inside one of chess's hottest arguments in recent years.
Ian Nepomniachtchi didn't set out to make news on a Tuesday livestream, but his aside landed squarely inside one of chess's longest-running arguments. "There's just such a rating deflation going on right now," he said, putting a number on his own experience: he'd lost about 40 rapid rating points over the past year, dropping from 2762 before recovering about 20 of them to sit at 2748. He hasn't been above 2800 in rapid since July 2022. And he was careful to frame it as a systemic problem rather than a personal slump: "Apparently, everyone has lost points," he said, noting that even closing the 53-point gap to Magnus Carlsen's 2801 rapid rating doesn't mean much if Carlsen is bleeding points at the same rate.
Nepomniachtchi is simply the most prominent name to say out loud what a growing body of research and more than a few irritated grandmasters have been arguing for three years: that FIDE's rating pool is still quietly deflating, that the fixes applied so far haven't closed the gap, and that it's starting to show up even at the very top of the game.
A familiar complaint, now backed by new data
The theoretical groundwork here is not new. Statistician Jeff Sonas told FIDE in a 2023 proposal that "player ratings are spread apart far more than they should be, and it's getting worse every year," using data from millions of games to show favorites across the pool consistently underperforming their ratings. FIDE's own working group, which included GM Alexander Moiseenko, estimated deflation on the order of 200 points in parts of the pool. The response, a one-time compression that touched roughly 350,000 players on January 1, 2024, plus a higher rating floor and a reinstated 400-point rule, helped but didn't solve it. An independent follow-up that October found deflationary pressure "continued, albeit quite mild compared to previous years," with large regional gaps, players from India, Sri Lanka, and Armenia underrated by 100 to 200 points, still unaddressed.
Things got more personal in 2025. GM Oleksandr Bortnyk accused FIDE of "stealing" his rating points after a retroactive recalculation tied to the new 600-point rule; FIDE CEO Emil Sutovsky called it a technical fix, not a retroactive penalty, a distinction GM David Howell flatly rejected ("This is literally the definition of retroactive"). Then came the so-called "Hikaru rule": FIDE stripped the 400-point cap from players rated 2650 and up after Hikaru Nakamura farmed rating points against 1800-rated amateurs in small American opens to hit his Candidates qualification game count. Carlsen called the maneuver "absolutely shameless" but understandable; Hans Niemann welcomed the crackdown; Howell called it short-sighted for punishing working pros who depend on open tournaments; and streamer Levy Rozman asked the question that lingers over all of this: if grinding open events no longer builds a rating, how does anyone outside the elite break in?
The forensic audit
The most detailed new evidence comes from Vlad Gita, a Romanian arbiter, trainer, and physics graduate who left a tech career during the pandemic and spent roughly three years collecting data before publishing an ebook this year titled The Rating Revolution: A Forensic Audit of the Global Elo Economy. Speaking on the Perpetual Chess Podcast, Gita said FIDE gave him access to anonymized game-level data covering nearly a million classical over-the-board games, which he used to build a country-by-country model of who is actually underrated.
His central finding is a mechanism he calls the "deflation vortex." Young players carry a K-factor of 40, meaning their ratings move fast; most adults carry a K-factor of 20. When a fast-moving junior and a slow-moving adult play each other in a federation with a healthy age mix, like Germany or Spain, the junior tends to overperform and gains roughly double what the adult loses, injecting a small amount of inflation into the system. But in federations where the rated population is overwhelmingly young players facing other young players, that friction never happens, because two K-40 players trading points is a zero-sum game. Combine that with players who rarely travel abroad to meet a more mixed field, and a federation gets trapped, in Gita's words, "in a whirlpool of deflation" that never self-corrects.
Gita also built a four-way age segmentation that goes further than Sonas's earlier junior-versus-everyone split: players under 18 gain rating roughly in proportion to how many games they play; a "prime" cohort from 19 to 34 still nets gains from activity; from 35 to 49 the relationship inverts and more games mean fewer rating points; and from 50 onward, the effect is sharp. It's worth noting that Nepomniachtchi, 36, and Carlsen, 35, both sit squarely inside the bracket Gita's data says is working against them by default, regardless of form.
At the country level, Gita's model flags Vietnam as the most underrated federation by over 100 points, with China, India, Uzbekistan, and Kazakhstan also showing significant underrating, consistent with the usual complaints about Asian open tournaments being "a graveyard for Elo ratings," as Gita put it. Two results surprised him: Peru and Ecuador showed up as highly underrated despite rarely coming up in the conversation, and India turned out to be less extreme than its reputation suggests, dwarfed by China and Vietnam, something Gita attributes partly to India's much larger pool of over 50,000 FIDE-rated players and the fact that at least some of its players travel and bring points back, where Vietnam's far smaller rated population mostly doesn't. He also pointed to English GM Keith Arkell, now rated 2380 despite showing no real decline in strength, as a case study in what heavy exposure to those pools does to an established player's number, and recounted facing a USCF National Master, rated 2200 in the US system, who carried a FIDE rating of just 1700.
Gita argues the US Chess Federation's Glicko-based system, overseen by its co-creator Mark Glickman, handles this better by proactively tuning its deflation parameters and using rating floors, which he concedes are inflationary by design but useful as a backstop, something FIDE doesn't really have beyond a hard cutoff at 1400 where an inactive player's rating is deleted outright rather than protected. His own proposed fix is deliberately narrow rather than a blanket solution: a flat bonus of one Elo point per game for K-factor-40 players specifically inside federations his model identifies as statistically underrated, which he argues would speed convergence to real strength without the risk of overshooting it, since the bonus's relative effect shrinks as a player's rating climbs toward where it belongs.
Gita says his relationship with FIDE is informal: he's sat in on the Qualification Commission's open Zoom meetings and exchanged data and ideas, and believes an earlier draft of his work reached FIDE's desk, though no formal collaboration exists yet. His verdict on the federation's posture is more generous than most critics manage: "thumbs up to FIDE," he said, for being newly willing to let outside, data-driven work shape the conversation.
Where that leaves Nepomniachtchi's complaint
None of this proves FIDE is mismanaging the system out of indifference, and Gita himself credits the 2024 reforms with real, if partial, progress. But it does mean Nepomniachtchi's throwaway livestream comment lines up with what the data has been saying for a while: deflation hasn't been solved, it's unevenly distributed by country and by age, and even a player who spent years at world number two can watch the number drop for reasons that have very little to do with how he's actually playing.