Stop building data infrastructure.
Start finding edge.
A data API for algorithmic traders and funds on Polymarket and Kalshi. Query live, export to Parquet, or stream in real time — without ever building a pipeline.
You've built this pipeline before.
Every prediction market trader building serious strategies knows the routine. On-chain fills, wallet history, market metadata, position tracking — none of it lives in one place. So you build a pipeline. Weeks pass. By the time the data is ready, the opportunity is gone.
Weeks of eng time, gone
Collecting, parsing, and merging raw events before you can validate a single hypothesis.
Pre-packaged data doesn't fit
Existing providers give you fixed views. You need to define what a maker is, what a competitor looks like, what counts as mispricing — on your terms.
Static snapshots in a live market
By the time a dataset is built, you need real-time feeds to act on what you found.
The data API for Polymarket and Kalshi.
Define the conditions. Titus executes them across 8 billion events in milliseconds. No schemas to work around. No dashboards to outgrow.
What a “maker” means in your strategy
// Your taxonomy, your terms define maker = wallet where maker_fill_ratio > 0.80 and win_rate > 0.65 and volume_30d > 100_000 select * from makers where market.category = “politics”
✓4 matches in 142ms
Flexible definitions
Makers, competitors, mispricing — all user-defined. Titus runs your taxonomy against 8B events.
Unconstrained queries
Combine wallet behavior, market metadata, fill history, and P&L in one query. No schemas to fight.
Real-time feeds
WebSocket subscriptions on wallets and markets. Pipe signals directly into your execution layer.
Query live. Export once.
Iterate offline — at local speed.
Every HTTP round-trip costs you 100+ ms. Multiply that across 10,000 backtest iterations and you're losing hours to the network. With Titus, you define your dataset once, export it as Parquet, and run backtests directly from disk.
From raw events to structured intelligence
An AI query planner translates natural language or structured requests into optimized queries across 10TB of market data. Pipe results into your execution layer in real time, or export as Parquet for offline backtesting.
What alpha traders are running today.
Actual queries from the two traders currently using Titus. Each returns structured data, ready for their systems.
Find all wallets with >80% maker orders, >65% win rate, and >80% weighted win rate.
Identifying consistent high-performing makers for competitive analysis.
Taker vs maker P&L per cricket match, tagged by consistent vs one-off players.
Roll-up analysis across a sports series for strategy validation.
On-chain transaction data for 20 specific wallets across 15 recent matches.
Raw data dump for offline analysis and custom modeling.
Natural language in. Market intelligence out.
Prefer the full walkthrough? Watch the 2-min Loom →
Common questions
Existing platforms give you pre-built dashboards or require you to write SQL against raw data. Titus lets you define your own conditions, tags, and classifications — then query across the full structured dataset in milliseconds. You define what a maker is, what a competitor looks like, what counts as mispricing. Titus executes it.
Titus is a data API for prediction markets. It indexes on-chain events across platforms like Polymarket and Kalshi, structures them, and provides sub-200ms queryable access via REST, WebSocket, and MCP endpoints. Direct AI agent connections run through an additional reasoning layer and can take longer than 200ms depending on query complexity.
Every fill, transfer, position change, market resolution, and wallet interaction across supported prediction markets. This includes derived data like P&L, maker/taker classification (by your definition), lot-level cost basis, and order aggregation.
Via REST API for queries, WebSocket for real-time wallet tracking and market notifications, MCP for AI agent integration, or direct AI agent connections. Sub-200ms response times on REST, WebSocket, and MCP endpoints. Direct AI agent connections include a reasoning step on top and typically exceed 200ms — use them for exploratory work, and the loop-optimized endpoints for latency-sensitive paths.
Usage-based API pricing. The private alpha is free for the current cohort. A paid tier rollout is expected by Q3 2026 — pricing will scale with query volume and real-time subscription load.
The data is ready.
You should be too.
Private alpha — two traders onboarded. Accepting applications for the next cohort.
Get alpha access
Early access to the data API for algorithmic traders on Polymarket and Kalshi.