Prediction Markets as Macro Instruments
What 352 resolved Polymarket macro markets, 121.7M Kalshi trades, a 110-position paper book, and a 14-market depth scan teach us about when a prediction-market price is information and when it is five wallets. Companion to the State of Macro Positioning, Q3 2026 (2026-09-12).
data as of 2026-10-08 · published 2026-10-12 · next update monthly (next: 2026-11-10)
TL;DR — the read in bullets
- 352 resolved Polymarket macro markets price information, not noise: Brier 0.027 against a 0.132 base-rate benchmark, a 79% skill score, confirmed out-of-fold by a cross-fitted Venn-ABERS check (0.026).
- The headline number hides its worst segment. Split by tag, Fed contracts score Brier 0.120 on 21 resolved markets — four times worse than the 0.027 headline — while Crypto (0.012) and Economy (0.013) are the strongest. The 21 Fed contracts are also the same markets the depth pillar flags as 100% concentrated on one side, though n=21 leaves the cause ambiguous: this could be (a) Fed contracts are intrinsically harder to forecast, (b) the small set of Fed contracts on Polymarket are dominated by NO-side whales, or (c) both. The by-tag split alone does not pick a side.
5 further findings, 3 charts and the rest of the analysis are in the full report.
They are absent from this page, not hidden. View-source and check.
The full report is on Full Access.
See the plansAlready subscribed? Sign in.
MacroTerminal Research is market commentary and data analysis. It is not investment advice, and nothing here is a recommendation to buy or sell anything.
MacroTerminal is a research publication and is not investment advice; nothing here is a recommendation to buy or sell anything.
MacroTerminal is a research publication and is not investment advice; nothing here is a recommendation to buy or sell anything.