This is a Preprint and has not been peer reviewed. This is version 1 of this Preprint.
Improvement of Daily Temperature Forecasts from a Prediction Market
Downloads
Authors
Patrick T Brown 
Abstract
Prediction markets aggregate and distill the collective views of participants who have a financial stake in being correct. They have been shown to produce more accurate forecasts in realms like election results, which are not easily amenable to mechanistic modeling because they entail forecasting human behavior. It is far less clear that prediction markets should improve the accuracy of predictions of physical quantities where credible mechanistic models already exist, and their output is further routinely augmented with statistical post-processing and expert human oversight. Here I evaluate ForecastEx prediction markets on the daily maximum temperature at 24 US locations from 11 February 2026 to 1 October 2026 (5,405 location-days) against 20 operational forecast systems at every hourly lead from 36 hours before the end of the target day. The alternatives include state-of-the-art numerical weather prediction models with biases removed, model output statistics applied to those models, optimal model blends, an artificial intelligence ensemble, and human-supervised forecasts with access to all the aforementioned information. I converted prediction market probabilities to a central value for comparison with alternative deterministic systems that output only a single value, and I compared them using mean absolute error. Where alternative systems provided probabilistic information, I applied basic ensemble model output statistics corrections and compared them to the prediction market probabilities using the continuous ranked probability score (CRPS). I made substantial efforts to verify alternative systems against the observations that align with their forecasts when they differed from the observations that the prediction market targeted. Whether lead time was counted to the end of the target day or to the time the daily maximum was observed, the prediction market central forecast value and CRPS had the lowest error at all meaningful lead times. At 6:00 PM local time on the evening before the target day, when none of the day’s temperatures had been observed, the prediction market’s mean absolute error was 1.49°F, compared with 1.97°F for the best deterministic alternative (24% [20%, 29%] lower), and its CRPS was 1.06°F compared with 1.19°F for the best EMOS-corrected probabilistic alternative (11% [8%, 15%] lower). The margins generally held or improved through the target day, when the market’s responsiveness to live information becomes more of an advantage. The advantage was a genuine gain in accuracy rather than an artifact of forecast latency, since the market was more accurate than each alternative even when sampled only at the times when the alternative issued a new value. The prediction market’s performance improvement over model output statistics deliberately designed to remove systematic biases, and over human-expert systems with access to the same information, indicates that the market combines information and potentially incorporates additional information more flexibly and dynamically than conventional alternatives.
DOI
https://doi.org/10.31223/X5522D
Subjects
Applied Statistics, Atmospheric Sciences, Climate, Meteorology, Natural Resource Economics, Oceanography and Atmospheric Sciences and Meteorology, Physical Sciences and Mathematics, Probability, Statistical Methodology, Statistical Models, Statistics and Probability
Keywords
Prediction Market, Weather Forecasting, Numerical Weather Prediction, Model Output Statistics
Dates
Published: 2026-10-03 16:23
Last Updated: 2026-10-03 16:23
License
CC BY Attribution 4.0 International
Additional Metadata
Conflict of interest statement:
I am employed by Trend Analytics, LLC, a subsidiary of Interactive Brokers Group, which owns and operates ForecastEx, the exchange whose market prices are evaluated in this study. As with much academic research, certain research questions and their downstream results and framings are more appealing to the researcher than others. This is the case here, and I absolutely concede that the findings were welcome to me and my organization. Nevertheless, to the best of my knowledge, the findings are presented accurately and forthrightly. All the underlying datasets are publicly available, and I invite others to investigate the same questions from the same or different angles.
Data Availability:
ForecastEx publishes its daily price and trade records at https://forecastex.com/data. All alternative systems compared here are publicly available from their originating services. The station MOS products and the National Blend of Models are published by the NOAA Meteorological Development Laboratory at https://vlab.noaa.gov/web/mdl, with LAMP at https://vlab.noaa.gov/web/mdl/lamp, GFS MOS and NAM MOS at https://vlab.noaa.gov/web/mdl/mos and the Blend at https://vlab.noaa.gov/web/mdl/nbm. The National Weather Service forecast is published from the National Digital Forecast Database at https://digital.weather.gov. The GFS, GEFS, NAM and HRRR are published by the National Centers for Environmental Prediction at https://www.nco.ncep.noaa.gov/pmb/products and https://rapidrefresh.noaa.gov/hrrr. The ECMWF IFS, its ensemble and AIFS are published at https://www.ecmwf.int/en/forecasts/datasets/open-data and https://data.ecmwf.int/forecasts. ICON, ICON-EPS and MOSMIX are published by the Deutscher Wetterdienst at https://opendata.dwd.de. GEM and the GEPS are published by Environment and Climate Change Canada at https://eccc-msc.github.io/open-data. The Unified Model is published by the Met Office at https://www.metoffice.gov.uk/services/data, ARPEGE by Météo-France at https://donneespubliques.meteofrance.fr, and the GSM by the Japan Meteorological Agency at https://www.jma.go.jp/jma/jma-eng/jma-center/nwp. The METAR observations, the daily climate reports and URMA are published by NOAA, and the MOS bulletins as issued are archived by the Iowa Environmental Mesonet [105] at https://mesonet.agron.iastate.edu, which is the source used here, each bulletin timed to its cycle plus the measured delay in its arrival. A public archive will hold the data behind every figure of this manuscript and sample scripts that demonstrate processing. The verification statistics are also updated daily at https://weather.weatherclimatehumansystems.org/accuracy.html [106], which runs on the current record and will therefore differ from the frozen build reported here.
Metrics
Views: 4
Downloads: 0
There are no comments or no comments have been made public for this article.