SimpleFunctions

Compare · Data providers

Prediction Market
Data Providers, 10 Compared.

10 prediction-market data providers — SimpleFunctions, Oddpool, FinFeedAPI, Probalytics, Marketlens, TREMOR, Hashdive, Verso, plus raw Kalshi and Polymarket public datasets. Delivery format (REST / WebSocket / SDK / SQL / CSV / Parquet), venue coverage, history depth, pricing — verified from public sources.

Providers compared

10

Multi-venue

7

Free tier

4

Calibration data

SimpleFunctions only

Quick recommendation

If you want X, use Y.

  • Cross-venue + indicators + calibration + CLI/API → SimpleFunctions
  • Sub-second WebSocket + trader attribution → Oddpool
  • Enterprise SLA-grade normalized REST → FinFeedAPI
  • 1ms orderbook + SQL warehouse + Parquet bulk → Probalytics
  • Polymarket-only Python SDK for backtesting → Marketlens
  • SQL-first PM data warehouse → TREMOR
  • Curated Smart Score signal layer → Hashdive
  • Bloomberg-style terminal UI → Verso
  • Kalshi-only raw exports → Kalshi public market downloads
  • Free Polymarket research dataset → Jon-Becker GitHub

Side by side

10 providers, comparison table.

ProviderDeliveryFormatVenuesPricing
SimpleFunctionsCLI + REST/Data API + WebSocket + MCP adapterJSON / streaming / CSV via /api/data/downloadKalshi, PolymarketPublic reads free; authenticated thesis/intents free up to 15M tokens
OddpoolWebSocket + RESTJSON / streamingKalshi, PolymarketPro $30/mo (1M req) · Premium $100/mo (5M req)
FinFeedAPIRESTJSONPolymarket, Kalshi, Manifold, MyriadEnterprise SLA contracts; quote-driven
ProbalyticsREST + ClickHouse SQL + Parquet bulk export; planned SSE streamingJSON / SQL / ParquetPolymarket, KalshiTiered — see probalytics.io/pricing (Bearer auth: api_xxx:sk_xxx)
MarketlensPython SDKPython objects / pandas DataFramesPolymarketFree 60 req/min · Pro $39/mo 720 req/min · Enterprise $199/mo 1,800 req/min
TREMORWeb terminal + SQL data warehouse accessSQL / CSVKalshi, PolymarketSubscription tiers
HashdiveWeb app + REST APIJSONKalshi, PolymarketSubscription tiers
VersoBloomberg-style terminal UI (no public API)UI onlyKalshi, PolymarketSeat-based SaaS
Kalshi public market downloadsCSV downloads + REST APICSV / JSONKalshiFree for public data
Polymarket public dataset (Jon-Becker)GitHub-hosted snapshots + scrapersCSV / JSON / parquetPolymarketFree / open-source

Detail · 10 providers

Each provider, in one paragraph.

SimpleFunctions

https://simplefunctions.dev

KalshiPolymarket

Continuous since 2025-Q4 — markets, prices, candles, indicators, calibration, theses

Cross-venue normalized PM data with the agent layer above raw prices: CLI command catalog, REST/Data API, computed indicators (IY/CRI/LAS/EE) across the indicator universe, live calibration scores at /api/calibration, cross-venue arb pair table, causal-tree thesis system, CC-BY-4.0 public datasets on HuggingFace + Kaggle, and an MCP adapter for compatible hosts. The only PM data provider that publishes its own Brier scores as a programmatic endpoint.

Best for: Builders + AI agents wanting normalized PM data plus computed indicators plus thesis system plus calibration in one install.

Delivery: CLI + REST/Data API + WebSocket + MCP adapter

Oddpool

https://oddpool.com

KalshiPolymarket

Tick-level orderbook streaming

Bloomberg-style PM data SaaS with sub-second WebSocket + trader attribution + whale tracking. Tier pricing scales with request volume. Strong on the orderbook microstructure surface; quiet on derived signals (no MCP, no calibration scoring, no thesis system).

Best for: HFT-style bots, quant teams, traders who need sub-second WebSocket on the orderbook + per-row trader attribution.

Delivery: WebSocket + REST

Full SimpleFunctions vs Oddpool comparison →

FinFeedAPI

https://finfeedapi.com/products/prediction-markets-api

PolymarketKalshiManifoldMyriad

Production-grade feed history

Mature normalized REST API across four venues, built on CoinAPI's enterprise infrastructure. Strong on uptime SLA + production-feed semantics; lighter on derived signals (no MCP, no calibration scoring, no thesis system).

Best for: Enterprise customers requiring SLA-grade uptime on a normalized PM data feed.

Delivery: REST

Full SimpleFunctions vs FinFeedAPI comparison →

Probalytics

https://probalytics.io

PolymarketKalshi

1ms orderbook resolution at 200-500M event updates/day · 813K Polymarket markets · 40M Kalshi markets

Microstructure-first PM data product: REST + direct ClickHouse SQL + Parquet bulk export. Highest-fidelity orderbook data on this list. Targets quantitative HFT-style buyers who need tick-level history and SQL access.

Best for: Quant researchers who need tick-level orderbook history + SQL queryable warehouse + bulk historical exports.

Delivery: REST + ClickHouse SQL + Parquet bulk export; planned SSE streaming

Full SimpleFunctions vs Probalytics comparison →

Marketlens

https://marketlens.trade

Polymarket

Tick-level Polymarket orderbook + trade history

Polymarket-exclusive Python SDK focused on tick-level orderbook history and quant backtesting. Single-venue (Polymarket only). Strong on the Python-native quant workflow; quiet on cross-venue + agent surface.

Best for: Python quants who want a Polymarket-native backtest harness with deterministic tick data.

Delivery: Python SDK

Full SimpleFunctions vs Marketlens comparison →

TREMOR

https://tremor.markets

KalshiPolymarket

Snapshot + historical data via SQL queries

SQL-first PM data terminal — query the warehouse directly with familiar SELECT syntax. Targets data-team buyers who already work in dbt/Snowflake/BigQuery patterns and want PM data accessible the same way.

Best for: Data teams that want PM data joinable in their existing SQL warehouse.

Delivery: Web terminal + SQL data warehouse access

Full SimpleFunctions vs TREMOR comparison →

Hashdive

https://hashdive.com

KalshiPolymarket

Continuous market history with Smart Score analytics

Analytics-first PM data product with proprietary Smart Score signal layer ranking markets by their internal heuristics. Strong on the curated signal surface; lighter on raw data export workflows.

Best for: Traders who want a curated signal layer rather than raw orderbook data they normalize themselves.

Delivery: Web app + REST API

Full SimpleFunctions vs Hashdive comparison →

Verso

https://verso.trading

KalshiPolymarket

Continuous via terminal

Institutional-grade Bloomberg-style PM terminal distributed via polymark.et. Self-reports use of GPT-5 with claimed 73% forecast accuracy. UI-first product — no public REST or WebSocket data API. Different buyer than the API-first providers on this list.

Best for: Professional traders who want a polished visual cockpit rather than a programmable data feed.

Delivery: Bloomberg-style terminal UI (no public API)

Full SimpleFunctions vs Verso comparison →

Kalshi public market downloads

https://kalshi.com/about/markets

Kalshi

Daily market exports + full REST history

Kalshi's own public data exports — CSV market downloads plus the public REST API. Source of truth for Kalshi data; no aggregation, no normalization, no Polymarket coverage.

Best for: Kalshi-only research projects, regulatory analysis, building custom downstream pipelines.

Delivery: CSV downloads + REST API

Polymarket public dataset (Jon-Becker)

https://github.com/Jon-Becker/polymarket-public-data

Polymarket

Periodic public snapshots

Community-maintained Polymarket public dataset on GitHub. Good for one-off analysis, academic research, or as a starting point before paying for production-grade feeds. Commit recency varies — check before relying.

Best for: Academic research, one-off analysis, exploratory data work where deterministic recency is not required.

Delivery: GitHub-hosted snapshots + scrapers

FAQ

Common questions.

What is the best prediction market data provider in 2026?

No single best — match the provider to the workflow. For builders + AI agents needing normalized cross-venue data plus calibration plus indicators plus a thesis system, SimpleFunctions ships the deepest agent-callable surface. For HFT-grade tick orderbook + SQL warehouse + Parquet bulk, Probalytics. For sub-second WebSocket + trader attribution, Oddpool. For SLA-grade enterprise feed, FinFeedAPI. For Python-native Polymarket backtesting, Marketlens. For institutional Bloomberg-style terminal, Verso. For raw single-venue data, the venues' own public exports.

Is prediction market data free?

Mostly yes for read access. Kalshi exposes a public REST API + daily CSV exports for free. Polymarket's gamma API and CLOB read endpoints are free and auth-light. Manifold and Metaculus offer free public APIs. Aggregators (SimpleFunctions, FinFeedAPI, Adjacent News) typically have free tiers for public read access plus paid tiers for SLA, write, or higher rate limits. Specialized derived-signal products (Oddpool, Hashdive, Verso, Probalytics, Marketlens) are subscription-based.

What is the difference between a data provider and an API?

Overlapping but not identical. A data provider sells the data product (CSV exports, SQL warehouses, normalized REST, WebSocket streams, Python SDKs). An API is the access protocol — a data provider may ship a REST API as one of several delivery channels alongside CSV downloads or SQL access. SimpleFunctions ships data via CLI + REST/Data API + WebSocket + CSV downloads, with MCP as an adapter for compatible hosts. Probalytics ships via REST + ClickHouse SQL + Parquet. Oddpool ships via WebSocket + REST. Marketlens ships via Python SDK only. Match the delivery channel to your workflow.

Which providers cover both Kalshi and Polymarket?

SimpleFunctions, Oddpool, FinFeedAPI, Probalytics, TREMOR, Hashdive, Verso, and Adjacent News all cover Kalshi + Polymarket from a single integration. They differ in additional venues (FinFeedAPI adds Manifold + Myriad), in derived signals (only SimpleFunctions ships calibration + thesis system + agent-ready CLI/API surfaces), and in delivery model (Marketlens is Polymarket-only Python SDK; Verso is UI-only).

How do I bulk-download historical prediction market data?

For Polymarket: the Jon-Becker public dataset on GitHub has periodic snapshots; Probalytics offers Parquet bulk export at higher tiers; Marketlens offers tick-level history through their Python SDK. For Kalshi: Kalshi's own daily CSV exports are the canonical source plus Probalytics Parquet exports. SimpleFunctions publishes daily CC-BY-4.0 dumps on HuggingFace (huggingface.co/SimpleFunctions) and Kaggle covering markets, prices, calibration, and indicators.

Which data provider publishes calibration data?

SimpleFunctions is the only provider on this list that publishes live Brier scores as a programmatic endpoint — GET /api/calibration returns scores by venue, category, and price bucket, currently Kalshi 0.20 / Polymarket 0.12 on T-24h price over the past 90 days. Verso self-reports a 73% forecast accuracy on its dashboard but does not expose a queryable calibration endpoint. Other providers ship raw price data only; calibration must be computed downstream.

What is the right data provider for an AI agent?

Agents typically need: (a) normalized cross-venue prices, (b) computed indicators ready to consume, (c) calibration to weight predictions, (d) a CLI or HTTP tool surface so the agent can invoke capabilities by name. SimpleFunctions is the only provider on this list that ships all four, plus an MCP adapter for hosts that require it. Other providers ship raw data — your agent layer would need to compose the indicators, calibration, and tool routing yourself before any reasoning can happen.

Can I get prediction market data in CSV or Parquet?

Yes — multiple providers support bulk export formats. SimpleFunctions exposes /api/data/download for CSV exports plus daily CC-BY-4.0 dumps on HuggingFace + Kaggle. Probalytics offers Parquet bulk exports. Kalshi publishes daily CSV market exports on its public site. Jon-Becker's Polymarket public dataset on GitHub ships JSON, CSV, and parquet. Pick the format your downstream pipeline already speaks.

Try the SimpleFunctions data feed.

Cross-venue normalized data + computed indicators + live calibration + CLI/API surface. Public reads free, no auth required. CC-BY-4.0 daily dumps on HuggingFace and Kaggle.

curl https://simplefunctions.dev/api/public/markets
Get API KeyBrowse data products →
All comparisonsPrediction market APIsMCP serversSimpleFunctions vs each toolHistorical dataReal-time dataCalibration