Universe members (GET /api/universe/{name}/members)

GET /api/universe/{name}/members returns the active membership of a named universe at one trading day (latest by default). It's the canonical answer to "what's in uni_mc_3000 today?" — without requiring a local Python SDK or a stale cached list.

This is the foundational endpoint for any cross-sectional workflow that needs to align on the universe the model was built for: rank screens, basket queries, attribution sleeves, panel histories. Trust the response; don't reinvent the membership check at the client edge.

Active = mask AND validity

The membership the API serves is the same dual-gate the ERM3 pipeline applies to produce its output zarrs:

Symbols failing either gate are not in the response. The endpoint does NOT return a "would-be-in-the-universe-but-data-broken-today" list.

Why two cadences matter

Because the gates run at different cadences, membership can change between two adjacent teos for two distinct reasons:

• New month — the monthly mask gets restamped, names enter or leave based on the prior month's mcap / dollar-volume.
• Daily validity — the same monthly mask, but a stock had a data hole today.

The response carries a mask_as_of field (the month-end the mask was stamped at) so callers tracking turnover over time can disambiguate. If mask_as_of is the same on both teos, any membership change is validity-driven; if mask_as_of differs, it's a fresh monthly mask.

Request

GET /api/universe/uni_mc_3000/members          ← latest teo
GET /api/universe/uni_mc_3000/members?teo=2026-05-15

Registered universes (from ERM3 KNOWN_UNIVERSES):

Unknown labels return 400; the API does NOT leak zarr-path semantics by accepting arbitrary strings.

Response shape

{
  "universe":    "uni_mc_3000",
  "teo":         "2026-05-26",
  "mask_as_of":  "2026-04-30",
  "members": [
    { "symbol": "BW-US0378331005", "ticker": "AAPL" },
    { "symbol": "BW-US5949181045", "ticker": "MSFT" },
    ...
  ],
  "counts": {
    "active":                  2813,
    "in_universe_mask":        2822,
    "inactive_from_validity":  9
  },
  "_metadata": { "data_source": "zarr", ... }
}

The counts block matches the breakdown the ERM3 1e Excel report's "Mask Profile" tab surfaces — same logic, exposed as JSON.

Examples

Bash — what's in uni_mc_3000 today

curl -sS "https://riskmodels.app/api/universe/uni_mc_3000/members" \
  -H "Authorization: Bearer $RISKMODELS_API_KEY" \
  | jq '{teo, mask_as_of, counts, sample_tickers: [.members[:5][].ticker]}'

Track turnover across a month rollover

prev=$(curl -sS -H "Authorization: Bearer $RISKMODELS_API_KEY" \
  "https://riskmodels.app/api/universe/uni_mc_3000/members?teo=2026-04-30" | jq -r '.members[].ticker' | sort)
curr=$(curl -sS -H "Authorization: Bearer $RISKMODELS_API_KEY" \
  "https://riskmodels.app/api/universe/uni_mc_3000/members?teo=2026-05-30" | jq -r '.members[].ticker' | sort)

# Added in the new monthly mask
comm -13 <(echo "$prev") <(echo "$curr")

# Dropped
comm -23 <(echo "$prev") <(echo "$curr")

Python — feed a rank-screen with universe-aligned tickers

from riskmodels import RiskModelsClient
client = RiskModelsClient.from_env()

# 1. Get the active universe
universe = client.get_universe_members("uni_mc_3000")
tickers  = [m["ticker"] for m in universe["members"]]

# 2. Pull a residual-reversion signal aggregate over the full universe
basket = client.get_residual_signal_basket(
    tickers=tickers,
    signal_quality_min_quintile=4,
)
basket["aggregate"]["residual_z_5d"]

Pricing

0.005/request (billing_code: universe_members_v1). Cheap by design — this is foundational; you'll hit it once per workflow to align on the universe and then issue more expensive calls against the result.

When to use

• Before any cross-sectional workflow that needs universe alignment.
• Anytime your basket / portfolio query returns "fewer names than expected" — diff your input against this endpoint's response to find what dropped and why.
• For turnover tracking — compare across month boundaries (mask change) or within a month (validity change).

When NOT to use

• For per-ticker risk metrics — use GET /api/metrics/{ticker}.
• For ranked screens — use POST /api/rankings/screen (which already filters to the active universe internally).
• For mask aggregates (how many names per sector, etc.) — coming in a future industry-membership endpoint; for now, derive client-side from the members[] rows joined to symbols metadata.

Related

• POST /api/rankings/screen — cross-sectional rank filter (also masks to the active universe).
• POST /api/signals/residual-reversion/basket — basket aggregator; pair with /universe/{name}/members to operate against the whole universe.
• OpenAPI UniverseMembersResponse — full schema.