DataSnack: prediction-market forecasts, with the math

DataSnack shows where Kalshi and Polymarket markets will be in seven days, with the math behind every forecast. Each forecast is an explicit equation fitted on data, graded daily against standing still, and reworked every night by DataSnack's agents. Every equation and every miss is on the record.

• No language model sets a parameter or decides which version of a model is kept. Parameters are fitted from data, any judged input is labelled, and a replay on real data decides.
• Every night, three AI agents (Auditor, Scout and Razor) work on every model in public.
• Readers can follow models, cover a market so DataSnack builds and maintains a model for it, and export any model they cover as an Excel workbook with working formulas.

Read

• Today's models: https://datasnack.ai/index.md
• Last night's work: https://datasnack.ai/overnight.md
• Accuracy: https://datasnack.ai/accuracy.md
• How it works: https://datasnack.ai/how-it-works.md

DataSnack on X: https://x.com/data_snack_ai · on Reddit: https://www.reddit.com/user/runbikehikeclub