NPDev FinPulse · A little more market clarity.
Public prediction markets
Markets discovered from Polymarket's public API, with its own implied probability and, where the public record supports one, an independent estimate from the NVIDIA Kumo Relational model.
Discovery coverage — open: 21980 markets over 8 pages · resolved: 437 markets over 8 pages · 997 cached model estimates. Every number is cached locally; reading this page never calls a provider.
21980 matching markets · newest discovery first within each scope
Uncategorised · Open · 9605 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
No cached estimate yet. Estimates are produced in the background and are never generated by a page view.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9605 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
No cached estimate yet. Estimates are produced in the background and are never generated by a page view.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9604 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
No estimate is published for this market: rejected.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9602 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
No cached estimate yet. Estimates are produced in the background and are never generated by a page view.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9602 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
No estimate is published for this market: rejected.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9602 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
Cached model estimate of the eventual "Yes" outcome from 120 resolved markets. Generated Sep 23, 2026 · 16:30 UTC. Refresh status: ready. Not independent evidence.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9601 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
No estimate is published for this market: rejected.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9600 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
No estimate is published for this market: rejected.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9599 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
No cached estimate yet. Estimates are produced in the background and are never generated by a page view.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9599 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
Cached model estimate of the eventual "Yes" outcome from 120 resolved markets. Generated Sep 23, 2026 · 18:45 UTC. Refresh status: ready. Not independent evidence.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9597 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
No cached estimate yet. Estimates are produced in the background and are never generated by a page view.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Uncategorised · Open · 9595 USDC tracked volume
Polymarket's own last-traded price for the first outcome, shown as a probability.
Cached model estimate of the eventual "Yes" outcome from 120 resolved markets. Generated Sep 23, 2026 · 16:30 UTC. Refresh status: ready. Not independent evidence.
Two separate numbers: the market's own last-traded price is market-derived evidence; the model estimate is a different, independent method. Neither is a recommendation.
Polymarket publishes a market's own last-traded price, which is its implied probability. That is market-derived evidence: it is the crowd's aggregate view, and it is shown separately from anything else.
The model estimate comes from a supervised relational model asked to predict how the market eventually resolves. It learns only from markets that have already settled, described only by what was observable a fixed horizon before their resolution — pre-resolution price paths, public trade history, and order-book snapshots recorded before the market ended. A market's outcome, final price and final volume are never inputs.
Order books are only published as a live snapshot, so no historical book exists to learn from; the section therefore records durable timestamped snapshots from now on, and reports book features as missing — never as zero — wherever none was recorded before an anchor.
Not investment advice. Prediction markets carry risk of total loss.