Global equity intelligence, as an API.

Python SDK — pip install git+https://github.com/Computational-Value-Investing/cvi-python-sdk

import cvi                                     # distribution: computational-value-investing
c = cvi.connect(api_key="cvi_live_…")

c.screen(top=20)                               # top 20 undervalued worldwide — one call, 25 markets
c.market("DAX").rankings(limit=10)             # one market, cheapest-quality first
c.company("CON.DE").valuation()                # intrinsic value + margin of safety
c.time_machine("CON.DE", "2026-07-15")         # the snapshot recorded for that date
c.predictions(market="SP500", top=20)          # hierarchical model picks
c.portfolio([{"ticker":"CON.DE","weight":.5},
             {"ticker":"AXON","weight":.5}])   # exposures, weighted scores, concentration

REST — every endpoint, one screen

BASE=https://www.computationalvalueinvesting.com/v1
KEY="X-API-Key: cvi_live_…"

curl "$BASE/health"                                        # status (no key)
curl "$BASE/platform/statistics"                           # canonical coverage counts (no key)
curl "$BASE/changelog"                                     # API changelog (no key)
curl -H "$KEY" "$BASE/markets"                             # 25 covered markets
curl -H "$KEY" "$BASE/markets/DAX/rankings?limit=10"       # ranked securities in a market
curl -H "$KEY" "$BASE/screen?classification=pass&min_score=0.8&limit=20"  # global screen
curl -H "$KEY" "$BASE/companies/resolve?ticker=CON.DE"     # ticker → canonical CVI IDs
curl -H "$KEY" "$BASE/companies/{id}"                      # profile (all listings)
curl -H "$KEY" "$BASE/companies/{id}/scores"               # scores + confidence
curl -H "$KEY" "$BASE/companies/{id}/valuation"            # valuation + margin of safety
curl -H "$KEY" "$BASE/companies/{id}/features"             # features + lineage      [features]
curl -H "$KEY" "$BASE/companies/{id}/history"              # every research run      [research]
curl -H "$KEY" "$BASE/companies/{id}/as-of/2026-07-15"     # Time Machine            [research]
curl -H "$KEY" "$BASE/predictions/hierarchical?market=SP500&limit=20"  #             [models]
curl -H "$KEY" -X POST "$BASE/portfolios/analyze" \
     -H "Content-Type: application/json" \
     -d '{"holdings":[{"ticker":"CON.DE","weight":0.5},{"ticker":"AXON","weight":0.5}]}'  # [portfolio]

AI assistants — pip install git+https://github.com/Computational-Value-Investing/cvi-mcp

CVI is a native MCP tool for Claude and other AI assistants — ask "the 20 most undervalued stocks worldwide" in plain language and the assistant queries the research engine directly, citing the immutable research run behind every number.

{ "mcpServers": { "cvi": {
    "command": "cvi-mcp",
    "env": { "CVI_API_KEY": "cvi_live_…" } } } }   // Claude Desktop config

Every response tells you where it came from

X-CVI-API-Version: 1.1.0            X-RateLimit-Limit: 600
X-CVI-Serving-Version: v1           X-RateLimit-Remaining: 599
X-CVI-Research-Run: CVIR2026…       X-RateLimit-Reset: 1784204152
X-CVI-As-Of: 2026-07-15T22:48:27Z   X-Total-Count + Link (paginated)

Envelope: {"data": …, "meta": {…}} — meta always carries api_version, serving_version, as_of and source_run_id; paginated endpoints add total and next_offset.

Interactive reference (/v1/docs) → · SDK on GitHub · OpenAPI 3.1 spec · Live status · Changelog · Plans & API tiers