Risk Metrics

Cohort residual statistics (time series)

Cohort statistics over a date range, one series per cohort. This is the endpoint to use for demeaning: request residual_mean for the level your residual is defined against and subtract it. See /cohorts for why that is necessary — ERM3 fits residuals without an intercept, so the cross-sectional mean is not zero.

The panel runs from 2000-01-03, but full factor richness begins around 2006; earlier history leans on proxy or synthetic backfill. Each cohort reports proxied_fraction, the share of returned days whose factor came from a substitute instrument. This matters for long windows: two sector cohorts are majority-proxied over the full panel, so a chart that does not surface the substitution shows partly a different basket. Pass include_proxy_source=true to see which instrument backed each day.

get/cohorts/series

Request

  • Base URL: https://riskmodels.app/api
  • URL: https://riskmodels.app/api/cohorts/series
  • Auth: HTTP bearer

Query parameters

cohortsstring

Comma-separated cohort tickers. Default: all public cohorts.

variablesstring

Comma-separated variable names. Default: residual_mean, residual_sd, mean_pairwise_corr, n_names, n_effective.

start_datestring date

Window start (default panel start).

end_datestring date

Window end (default latest teo).

min_namesinteger

Drop days where the cohort had fewer than this many members.

include_proxy_source'true' | 'false'

Include the per-day instrument backing the cohort factor.

Response

Cohort time series

rangestring[] nullable

[first, last] date actually returned.

variablesstring[]
min_namesinteger
universestring
market_factor_etfstring
store_buildstring nullable
data_sourcestring

Example response

{
  "data_source": "zarr"
}

Changes

No recorded changes to this operation across all 1 revision of this API.