Cryptarista Data API

A funded derivatives research desk, exposed as production infrastructure.
For AI agent builders and systematic traders · aristaflow.cloud · beta · April 2026
Abstract

Cryptarista runs a live derivatives research desk and ships its output as production infrastructure. The system collects funding rates, liquidations, open interest, and crowding data from Binance and Bybit, then layers evidence-linked news signals — structured events traceable from raw source to derived signal. Subscribe to fund the research. Consume it through REST, MCP, or Python SDK.

Request beta access →

1.Market Failure

Crypto derivatives data is fragmented across exchanges with proprietary schemas and rate limits. Aggregators consolidate it but charge $299/month for commercial access and restrict redistribution. The gap they leave: no one offers evidence-linked event intelligence — structured signals traceable to the source article, extraction logic, and market context that produced them. Technical buyers do not purchase transport. They purchase interpretation, confidence, and auditability.

The commercial promise is not universal market coverage. It is narrower coverage, stronger interpretation, and cleaner delivery into software.

2.Signal Wedge

The news signal pipeline is the primary differentiator. Raw articles and social posts are ingested, classified, and processed through event extraction that identifies regulatory actions, exchange incidents, and macroeconomic signals. Related events are clustered temporally, and each cluster produces a derived signal enriched with market context from the derivatives layer.

Every signal maintains a complete evidence chain: the raw source URL, the extracted event payload, the clustering rationale, and the market data snapshot at signal generation time. Consuming systems can audit signal provenance rather than treating signals as opaque inputs.

Raw Article / PostEvent ExtractionClassify, extract structured eventTemporal ClusteringGroup related events into clustersSignal DerivationDerive signal, attach confidenceMarket ContextEnrich with derivatives snapshotEvidence Chainsource_urls[], payload, context
Figure 1. Each pipeline stage preserves provenance \u2014 full traceability from raw source to derived signal.
The wedge is not “more endpoints.” The wedge is signals that can survive inspection when a desk, model, or agent asks why they fired.

3.Endpoint Surface

Eight endpoint families cover derivatives market data, news-derived signals, and computed features. All endpoints return JSON with consistent pagination and error schemas.

Table 1. Endpoint families, data coverage, and update characteristics.
EndpointDataUpdate FrequencyHistory Depth
/v1/snapshots/derivativesComposite snapshotReal-timeCurrent
/v1/history/fundingFunding rates8h settlements3 years
/v1/history/oiOpen interest15 min3 years
/v1/history/long-shortLong/short ratios15 min3 years
/v1/events/liquidationsLiquidation eventsReal-time90 days
/v1/news/signalsNews event signalsEvent-driven90 days
/v1/news/eventsClustered news eventsEvent-driven90 days
/v1/features/crowdingCrowding score15 min90 days

4.Delivery Modes

The same research surface is available through REST, an MCP server for AI agent frameworks, and a Python SDK. Authentication is via API key (X-API-Key header). MCP supports both stdio (local agents) and SSE (remote). TypeScript SDK planned for Q3 2026.

Python SDK
from cryptarista import CryptaristaClient

client = CryptaristaClient(api_key="cda_live_...")
snapshot = client.snapshots.derivatives(symbol="BTCUSDT")
signals = client.news.active_signals(asset="BTC")
MCP (for AI agents)
Tool: get_derivatives_snapshot
Input: {"venue": "binance", "symbol": "BTCUSDT"}
Returns: funding_rate, open_interest, crowding_score, regime

5.Access Model

Subscription is the primary commercial path — recurring users fund the ongoing research desk. Credits handle burst workloads without plan changes. Optional x402 pay-per-use covers low-volume agent-native calls.

Table 2. Commercial access paths and what each one buys.
Access PathPriceBest ForWhat You Get
Preview$0EvaluationBTC derivatives, 7-day history, 100 calls/day — enough to inspect the surface before paying
Research Subscription$49/moProduction useCommercial access, MCP, evidence-linked news signals, BTC + ETH (expanding), 3-year history
Credits$25+Burst workloadsReplay jobs, premium queries, overages — flexible spend without a larger recurring commitment
x402 / Pay-per-usePer requestAgent automationOn-demand calls for automated workflows that don’t need standing account access

The hierarchy is intentional: subscription is the business, credits are the pressure valve, x402 is the edge case.

Request beta access →

6.System Architecture

The system runs a three-stage pipeline: collection, storage, and serving. Exchange collectors normalize data from Binance and Bybit into a unified schema and write to ClickHouse. A parallel news pipeline extracts structured events from unstructured sources and attaches evidence lineage. Both paths converge at the serving layer.

Exchange APIs(Binance, Bybit)Collectorsnormalize & ingestClickHousetime-series storeFastAPI / MCPREST, stdio, SSEConsumersbots, agents, dashboardsNews SourcesTrendRadarextract, classifyEvidencecluster, signal, enrich
Figure 2. Exchange data and news signals are collected, normalized, and served through a unified API layer.

References

  1. Model Context Protocol Specification, Anthropic, 2024. modelcontextprotocol.io
  2. CoinGlass API Documentation. coinglass.com/api
  3. Binance Futures API. binance-docs.github.io