The 2026 season is in the rankings: six World Climbing Series stops, the European Championship in Barcelona and two Europe Series events, in a model that reads every top, zone and attempt. Non-European continental events are not included. Grades below are posterior means with 95% intervals. Where we give a probability that one climber is better than another, it refers to estimated ability over the season, not to who had the better results.

Men: one clear result, then a cluster

Sorato Anraku won five of the six Series stops and is the top-rated man at 11.5 (11.0 to 12.0), with a 96% probability of being the strongest climber in the field. He did not enter Barcelona.

Behind him the data do not separate much. Dohyun Lee, Hannes Van Duysen, Mejdi Schalck, Rei Kawamata and Maximillian Milne sit between 10.5 and 10.75, with heavily overlapping intervals. Lee is ranked second, but the probability that he is better than Van Duysen is only 54%, and Lee’s chance of being a top-three climber is 56%. Those five places are best read as a group.

Women: an order that needs a caveat

Janja Garnbret is ranked first at 11.5, but the interval is wide (10.4 to 12.6) because she contested one event, Keqiao, and 13 boulders. The model gives her a 98% probability of being the strongest, which assumes those 13 boulders fairly represent her ability. Her season was too short to test that.

Among climbers who contested the full season, Erin McNeice (one win, four seconds) has a 78% probability of being the strongest and Oriane Bertone, who won the last two events, 20%. Five different women won the seven top-tier events. Annie Sanders won twice and ranks fifth, because the rating uses every boulder and not only finals.

Our older model, which we still fit alongside this one, agrees on both number ones and orders the climbers below them differently, another reason not to lean on those places.

How the forecasts did

We scored forecasts for Prague, Innsbruck, Barcelona and Salt Lake City, all made with the same method and using only results available before each round. Those for Barcelona, Salt Lake City and the Innsbruck men’s final were computed after the event. Across eight men’s and women’s competitions, pre-event forecasts improved on a base-rate forecast as follows.

Outcome Brier skill
Reaching the semi-final +46%
Reaching the final +41%
Podium +38%
Winner +17%

The favourite won three of eight, against 2.4 expected. Calibration was reasonable: outcomes given about 39% happened 42% of the time, and those given about 79% happened 86% of the time. Climbers in the same round are not independent, so these figures are less certain than their sample sizes suggest, and eight competitions is a small sample.

Salt Lake City men was the clear miss. Anraku had a 99% chance of reaching the semi-final and finished outside the qualifying places. Yonatan Katz won from a 0.1% pre-event chance and is still ranked 33rd at 9.2, because one win is a small share of 44 boulders. Two outcomes that unlikely in one event may be bad luck, or may reflect a weakness in the model, which holds each climber’s ability fixed across a season that had a four-month gap before this event. One event cannot tell us which.

We will address that assumption next, by allowing ability to change through a season.

Separately, our paper on the mathematics of the grade scale has now been published in the Journal of Quantitative Analysis in Sports.