FIDC tranches and aging

Inside a receivables fund’s capital structure and credit book — each tranche’s quotas, return and promised vs realised performance with its subscriptions and redemptions, and the aging ladder of what is owed. Informe tabs X and VI, as filed, from 2025.

Parameters, response columns and types are generated from the SQL catalog and render under Endpoints → FIDC in the sidebar. This page carries what a schema cannot: where the history starts, what the labels mean, and what is not derived.

Overview

Two more members of the monthly FIDC informe, next to the concentration tabs:

Tab	What it publishes	Endpoint
X_2 / X_3 / X_6	per tranche (classe_serie): quota count and value, the month’s return, and the performance the series promised vs what it delivered	fidc_tranches
X_4	per tranche: the month’s operations — subscriptions, redemptions, amortizations — by CVM’s own label	fidc_tranches (flows)
VI	the receivables not yet due by days to maturity, and the overdue ones by days past due	fidc_aging

History begins in 2013-01. These tabs come from CVM’s yearly HIST archive through 2024-12 and from the monthly informe from 2025-01; the value columns have the same names in both. coverage() states it on the fidc_tranches and fidc_aging rows. The percentage fields carry CVM’s outliers as filed (magnitudes of 1e14 and more), never rescaled: range-check them yourself.

fidc_tranches

curl -s -X POST "$SILO_URL/rpc/fidc_tranches" \
  -H "apikey: $SILO_ANON_KEY" \
  -H "Content-Type: application/json" \
  -d '{"p_cnpj":"05754060000113","p_from":"2025-01-01"}'

silo.fidc_tranches("05754060000113")
silo.fidc_tranches("05754060000113", series="Subclasse Senior 1")

Argument	Default	Meaning
p_cnpj	required	the fund; punctuation is stripped
p_from / p_to	the whole history	window on period (CVM’s competency date)
p_series	every tranche	one classe_serie, matched exactly as filed

One row per (month, tranche), oldest first:

Column	Meaning
classe_serie	CVM’s tranche label, as filed (CVM-175 filings read Subclasse Senior … / Classe Sênior …; seniority is only ever in this text)
quotas / quota_value	quota count and quota value (TAB_X_QT_COTA, TAB_X_VL_COTA)
return_month	the tranche’s return in the month, percent (TAB_X_VL_RENTAB_MES)
performance_expected	the performance the series promised, percent (TAB_X_PR_DESEMP_ESPERADO)
performance_realised	the performance it delivered, percent (TAB_X_PR_DESEMP_REAL)
tranche_filed	false when the month has tab X_4 flows for this series but no X_2 row — then the columns above are null because nothing was filed, not because the value was blank
flows	the tranche’s tab X_4 operations: an array of {tp_oper, value, quotas}, or null when none were filed

The shape, with illustrative values:

[
  { "cnpj": "05754060000113", "period": "2026-06-30", "classe_serie": "Subclasse Senior 1",
    "quotas": 1520000.0, "quota_value": 1.0843, "return_month": 1.12,
    "performance_expected": 100.0, "performance_realised": 103.4, "tranche_filed": true,
    "flows": [ { "tp_oper": "Captações no Mês", "value": 2000000.0, "quotas": 1845000.0 },
               { "tp_oper": "Resgates no Mês",  "value":  500000.0, "quotas":  461000.0 } ] }
]

flows keeps CVM’s label verbatim and never buckets it. tp_oper is free text (typically Captações no Mês, Resgates no Mês, Amortizações) whose vocabulary has drifted over the years. A fixed “subscriptions” column would silently drop any label it did not anticipate; the array cannot. Match the labels yourself — and a label you did not find is not a zero.

Nothing is derived. There is no performance gap, no subordination ratio and no senior/junior classification: the realised-minus-expected gap is one subtraction in the notebook, and the tranche label is the only seniority signal the filing carries. The percent fields are dirty the way CVM’s percentage fields are — raw values reach 10⁸ — so they are served as filed, never clipped: range-check them before you average.

fidc_aging

curl -s -X POST "$SILO_URL/rpc/fidc_aging" \
  -H "apikey: $SILO_ANON_KEY" \
  -H "Content-Type: application/json" \
  -d '{"p_cnpj":"05754060000113","p_from":"2026-01-01"}'

silo.fidc_aging("05754060000113", start="2026-01-01")

Argument	Default	Meaning
p_cnpj	required	the fund
p_from / p_to	the whole history	window on period

Long: 21 rows per month, oldest first.

kind	Rows	Meaning
to_maturity	10	credits not yet due, by days to maturity
overdue	10	credits past due, by days overdue
overdue_total	1	CVM’s filed total of overdue credits (TAB_VI_B_VL_DIRCRED_INAD)

bucket is the band (1-30, 31-60, 61-90, 91-120, 121-150, 151-180, 181-360, 361-720, 721-1080, >1080; TOTAL on overdue_total), with days_from / days_to as integers (days_to is null on >1080). item names the source column. value is BRL as filed; a blank in the filing is null, never 0.

overdue_total is not a sum of the ten overdue bands. It is the total CVM’s form asks for separately, served as filed, and the two can disagree. Sum the bands yourself if you want the sum, and keep them apart.

Tab VI covers the receivables acquired without substantial retention of risk by the originator; tab V, its with-risk twin, is not ingested. The panel’s delinquency metric (FIDC, from 2013-01; on every row from 2020-11) is the fund-level total; fidc_aging is the ladder under it.

Limits and freshness

• One 1,000-row page, and both refuse above it (22023). No cursor: a window that passes one page means narrowing it by date or p_series, not walking. See Conventions.
• coverage() carries a fidc_tranches row and a fidc_aging row, clamped to the FIDC family’s latest complete month like the concentration rows, with notes stating that both come from CVM’s HIST archive through 2024-12 and the monthly informe after.
• An unknown CNPJ is an empty array, never a guessed row.