Lambcast

Forecasts  ›  Metaculus Cup Spring 2026

Who Will Win the 2026 FIDE Candidates Tournament?

Published: February 28, 2026  •  Platform: Metaculus  •  Resolves: April 16, 2026
My Pick
Caruana
My Probability
26%
Community
29.6%
Biggest Disagreement
Nakamura -10pt

The Tournament

The 2026 FIDE Candidates Tournament is an 8-player double round-robin held in Paphos, Cyprus from March 29 to April 16. Each player faces every other player twice — once with white, once with black — for 14 rounds total. The winner earns the right to challenge World Champion Gukesh Dommaraju later in 2026. In the event of a tie, rapid and blitz tiebreaks determine the winner.

The Field

Fabiano Caruana (USA, 2795) — qualified via 2024 FIDE Circuit. Only player in the field to have won a Candidates Tournament (2018, 9/14). Six consecutive Candidates appearances. Most experienced player in the field by a significant margin.

Hikaru Nakamura (USA, 2810) — qualified via highest average rating (Aug 2025–Jan 2026). World #2 by rating. Last elite classical event: Norway Chess, June 2025. Since then, played four small North American tournaments against opposition averaging ~2090 Elo to meet FIDE's 40-game eligibility requirement.

Rameshbabu Praggnanandhaa (India) — qualified via 2025 FIDE Circuit. The dominant classical performer of 2025: won Tata Steel, SuperBet Classic Romania, UzChess Cup Masters, runner-up at Sinquefield Cup, and won the FIDE Circuit by a commanding margin.

Anish Giri (Netherlands) — qualified via 2025 Grand Swiss (8/11). Third Candidates appearance. Consistent performer at elite level with strong recent form.

Javokhir Sindarov (Uzbekistan) — qualified via 2025 World Cup. Age 20, first Candidates appearance. Limited head-to-head record against most of this field in classical chess.

Wei Yi (China) — qualified via 2025 World Cup. Re-emerged as a top-10 player after years away from the game. Won Tata Steel 2024. First Candidates appearance.

Matthias Bluebaum (Germany) — qualified via 2025 Grand Swiss (runner-up). First Candidates appearance. Best head-to-head percentage in the field, though over a small sample. Favorable color distribution early in the tournament (three whites in first four rounds).

Andrey Esipenko (Russia) — qualified via 2025 World Cup. Has scored only 1 out of 10 in decisive classical games against the rest of this field.

Model Approach: Weighted Scoring

This forecast uses a weighted scoring model rather than a pure Elo simulation. Each player was scored 1–10 on four factors, weighted as follows:

Factor Weight Rationale
Current Elo rating 20% Useful signal, but can lag real readiness
Recent classical form (Aug 2025–Feb 2026) 30% Preparation and sharpness matter at this level
Head-to-head record vs this field 30% Most predictive factor in round-robin formats
Candidates/elite tournament experience 20% 14 rounds is a format that rewards familiarity

Results

Player My % Community % Difference
Fabiano Caruana 26% 29.6% -4
Rameshbabu Praggnanandhaa 19% 13.1% +6
Hikaru Nakamura 17% 27.4% -10
Anish Giri 15% 9.1% +6
Wei Yi 8% 8.5% -1
Javokhir Sindarov 7% 9.1% -2
Matthias Bluebaum 6% 1% +5
Andrey Esipenko 2% 2% 0

Key Disagreements with the Community

Nakamura down 10 points. His 2810 rating reflects results, not readiness. Since Norway Chess in June 2025, Nakamura played exclusively in small North American tournaments against opposition averaging around 2090 Elo — a deliberate strategy to meet FIDE's 40-game eligibility requirement. Eight months without elite classical competition is a meaningful inactivity penalty at this level, where opening preparation and positional sharpness are built and maintained through high-level games. The bookmakers and community are pricing his rating. This model prices his preparation.

Praggnanandhaa up 6 points. By almost any measure, Praggnanandhaa was the dominant classical performer of 2025: Tata Steel winner, SuperBet Classic Romania winner, UzChess Cup Masters winner, Sinquefield Cup runner-up, and FIDE Circuit winner by a commanding margin. He enters Cyprus in the best form of anyone in the field.

Bluebaum up 5 points. The community has him at 1%, which implies near impossibility. For a player who qualified via Grand Swiss runner-up, has the best head-to-head percentage in the field (small sample), and gets favorable color distribution early — three whites in his first four games — 1% is too dismissive. 6% reflects a genuine long-shot, not an impossibility.

Model Limitations

The weighted scoring approach is interpretable and allows explicit adjustments for factors that ratings miss — but it introduces subjectivity that a pure Elo simulation avoids. The factor weights (why 30% on H2H and not 25%?) and the 1–10 scores themselves both involve judgment calls without a strong empirical basis. Two reasonable analysts could score the same player differently and produce meaningfully different outputs.

The more standard approach for an 8-player round-robin is an Elo-based simulation: derive each player's expected win probability in every individual game from pairwise Elo differences, simulate the full 14-round tournament thousands of times, and count how often each player finishes first. This approach is fully mechanical and replicable — anyone with the same inputs gets the same answer. It removes forecaster discretion almost entirely, which is why it is trusted.

The limitation of a pure Elo simulation, however, is that Elo is calculated from game results regardless of opponent quality or recency. Nakamura's 2810 rating was maintained in part by winning games against 2090-rated players — a fact that an Elo simulation has no mechanism to discount. This is the clearest case in the current field where a rating meaningfully overstates competitive readiness, and it is the primary reason this model departs from a pure Elo approach.

The ideal methodology would use an Elo simulation as a base, then apply targeted adjustments only where there is concrete evidence the rating is misleading. Nakamura is the strongest example of that case in this tournament.

Finally, lower probabilities for players like Giri, Sindarov, and Bluebaum reflect relative likelihood, not elimination. At this level the field is genuinely close, and upsets are part of the format.

Revision Log

Feb 28, 2026 — Initial forecast submitted to Metaculus. Weighted scoring model, 4 factors. Caruana 26%, Praggnanandhaa 19%, Nakamura 17%, Giri 15%.