Benchmarks/3 studies

Benchmarks

Protocols, comparisons and clearly labelled study status for event discovery and forecasting.

Next-move identification · judged · %
Mimic One: 56.0
Claude Sonnet 5.5: 53.7
GPT-6.1 Sol: 50.0
DeepSeek V4 Flash: 49.3
Gemini 3.8 Flash: 44.8
Public benchmark comparison

Human Behavioural Simulation

Mimic One against Claude Sonnet, GPT, DeepSeek and Gemini on naming what real actors do next, across 609 sealed forecast points.

Read full benchmark
Public benchmark comparison

Mantic Crucible

Forecasting 50 real-world outcomes across markets, economics, elections and geopolitics.

50 questions · mean normalized CRPS · lower is better
Meridian Superforecast0.1215
Mantic0.1198
Questions with lower error
Meridian Superforecast20Mantic30
Overall loss gap: 1.4% · Meridian Superforecast’s loss is 1.4% higher than Mantic’s.
Read full benchmark
Forecast Leaderboard

ForecastBench

Forecasting accuracy on real-world questions, compared with leading forecasting systems.

Core domains · Brier loss · lower is better
Vaticinus / vaticinus-3.0-raw0.1423
Anonymous 11 / jforecast-router-v10.1532
Vaticinus / vaticinus-3.00.1539
Ryso AI / Apex0.1629
Meridian Superforecast0.1639Meridian Superforecast · Rank 5th of 50
hliu763 / ForecasterX-Coco-09130.1692
Google Cloud / NanoForecaster-MTV-09130.1702
hliu763 / ForecasterX-Diego-09130.1704
hliu763 / ForecasterX-Butter-09130.1705
Google Cloud / NanoForecaster-ATL-09130.1706
Google Cloud / NanoForecaster-SVL-09130.1707
Afterflow / e12-competition-portfolio-v1-state_research0.1719
Anonymous 33 / Nathan0.1722
Yhat Labs / vani0.1756
Afterflow / e12-competition-portfolio-v1-conservative0.1759
Afterflow / e12-competition-portfolio-v1-state0.1759
Torchcast AI / dimoo-nano0.1762
TRACES / TRACES0.1762
Anonymous 33 / agentic-v10.1763
PathFi / PathFi-F60.1763
Anonymous 24 / ginger-rook0.1768
Anonymous 24 / honeycomb-fox0.1768
Anonymous 24 / mint-badger0.1768
NeuroFusion / new ensemble 10.1768
NeuroFusion / new ensemble 20.1768
NeuroFusion / new ensemble 30.1768
Ryso AI / Frost0.1778
Torchcast AI / hirono-nano0.1779
limeforecast / superpuper0.1781
Torchcast AI / hacipupu-nano0.1781
Ryso AI / Linen0.1785
yarrow / yarrow-kestrel-202609130.1803
Yhat Labs / kaal0.1846
Anonymous 29 / claude-sonnet-50.1868
PolyLedger / polyledger-quant0.1926
Anonymous 19 / ASA ForecastBench Shadow (routed_market_dataset_hybrid)0.1928
Elementary Deduction / Apollo0.1935
Elementary Deduction / Zeus0.1935
DL Solutions / dl-superforecaster-v20.1939
Anonymous 33 / fvb0.1940
Anonymous 22 / ant-batch-pa0.1944
Cenva Intelligence / norma-market-v20.1951
Anonymous 29 / mars-debate-v10.2023
yhat-labs / 8-ball0.2312
Phinomial Labs / phinomial-forecaster0.2357
Anonymous 11 / fs-panel-v10.2454
Anonymous 22 / ant-batch-ra0.2503
PolyLedger / polyledger-core0.2544
Cenva Intelligence / norma-v20.2565
PolyLedger / polyledger-pure0.4108
Read full benchmark