Preprint with Nikkei journalist Junya Iwai on forecasting Nobel Prizes from prize records
Professor Naohiro Shichijo and Junya Iwai, a journalist in the Science News and Insights Group of Nikkei Inc.'s Editorial Division, have published a preprint on Zenodo (2 October 2026): "Forecasting Nobel Prizes from Prize Records. How to Read Precursor Prizes: A Comparison of Forecasting Methods and the 2026 Candidates." The paper asks how well laureates of the Nobel Prizes in Physics, Chemistry, and Physiology or Medicine can be predicted from the science prizes they received beforehand. It is available in English and Japanese; both authors contributed equally, and Naohiro Shichijo is the corresponding author.
On 27 September 2026, the Nikkei reported on its front page an analysis, written by Junya Iwai, using a mathematical model that assesses the chances of winning a Nobel Prize in the natural sciences from the records of 146 science prizes in Japan and abroad, including the Lasker Awards (in Japanese; Professor Shichijo also commented in the article). This study takes the approach of that model as its starting point. We re-collected the prize records independently from Wikidata and the official lists of each prize, rebuilt the model, and tested how well it predicts. The figures in the paper therefore differ from those reported in the article.
The rebuilt baseline model weights each prize by the share of its recipients who later won the Nobel Prize, and ranks candidates on six indicators such as prize counts and the sum of prize weights. Over 78 field-years from 2000 to 2025, a laureate was ranked in the top 30 in 39 of them (50%). We compared the baseline with three alternatives, including one that builds into the model the fact that only a few laureates are chosen each year; none significantly outperformed it. The comparison procedure and the rule for adopting a method were fixed before the validation was run.
The paper also shows that choosing the input prizes using outcomes known only in hindsight makes past performance look better than it was. Adding citation growth of signature papers and citations from patents produced no detectable improvement.
The 2026 rankings of all candidates were fixed before the announcements, and the existence of the record at that time was registered on the Bitcoin blockchain with OpenTimestamps. Among candidates affiliated with institutions in Japan, Masashi Yanagisawa, who received the 2026 Lasker Award, ranks highest in Physiology or Medicine (30th), followed by Kazutoshi Mori (42nd). Without the 2026 awards Mori ranks higher; the order depends on the information cut, the set of input prizes, and the model's settings. The ranks are predictions derived from public prize records, not assessments of scientific achievement.
Going forward, we will extend the analysis beyond the Nobel Prize to a wide range of awards, including those outside science, and study the process by which the breakthroughs behind them emerge and come to be recognized.
We publish the full 2026 candidate rankings from the M2 model for Physiology or Medicine, Physics and Chemistry. They are the rankings locked before the announcements, re-ranked after removing candidates whose death in 2026 has been recorded and very old candidates who are unlikely to be alive (the locked rank before removal is given in a separate column). Names come from Wikidata, and affiliations from the address on each researcher's latest paper in Dimensions; they do not indicate nationality. The ranks are predictions derived from public prize records, not assessments of scientific achievement. Column headings are in Japanese.
2026 candidate list from the M2 model (Excel, 3.1 MB)
Paper (Zenodo, CC BY 4.0)