Input Data Schemas¶

This page documents the authoritative schemas for all input data files required by the RWA calculator. These schemas are defined in src/rwa_calc/data/schemas.py and represent the single source of truth.

Quick Reference¶

Data Category	File(s)	Required	Purpose
Counterparty	`counterparty/counterparties.parquet`	Yes	Borrower/obligor information
Facility	`exposures/facilities.parquet`	Yes	Committed credit limits
Loan	`exposures/loans.parquet`	Yes	Drawn exposures
Contingent	`exposures/contingents.parquet`	No	Off-balance sheet items
Collateral	`collateral/collateral.parquet`	No	Security/collateral
Guarantee	`guarantee/guarantee.parquet`	No	Credit protection
Provision	`provision/provision.parquet`	No	IFRS 9 provisions
Rating	`ratings/ratings.parquet`	No	Credit ratings
FX Rates	`fx_rates/fx_rates.parquet`	No	Currency conversion rates
Specialised Lending	`ratings/specialised_lending.parquet`	No	Slotting approach data
SFT Trade	`ccr/sft_trades.parquet`	No	FCCM SFT EAD (CCR — Art. 220–223)
SFT Collateral	`ccr/sft_collateral.parquet`	No	Optional collateral for FCCM SFTs
Equity Exposure	N/A	No	Equity holdings
CIU Holdings	N/A	No	Per-fund holdings for CIU look-through (Art. 132)
Model Permissions	`config/model_permissions.parquet`	No	Per-model IRB approach permissions

Mapping Files:

Mapping	File	Purpose
Facility Mapping	`exposures/facility_mapping.parquet`	Facility-to-loan hierarchy
Org Mapping	`mapping/org_mapping.parquet`	Organisation hierarchy (rating inheritance)
Lending Mapping	`mapping/lending_mapping.parquet`	Lending groups (retail threshold aggregation)

Counterparty Schema¶

Purpose: Defines borrower/obligor information used for exposure classification, risk weight determination, and hierarchy resolution.

File: counterparty/counterparties.parquet

Column	Type	Required	Description
`counterparty_reference`	`String`	Yes	Unique identifier for the counterparty
`counterparty_name`	`String`	No	Legal name of the counterparty
`entity_type`	`String`	Yes	Single source of truth for exposure class (see valid values below)
`country_code`	`String`	No	ISO 3166-1 alpha-2 country code
`annual_revenue`	`Float64`	No	Annual revenue in GBP (for SME classification - EUR 50m threshold)
`total_assets`	`Float64`	No	Total assets in GBP (for LFSE threshold: EUR 70bn under CRR Art. 142(1)(4); GBP 79bn under PS1/26 Glossary p. 78)
`default_status`	`Boolean`	No	Whether counterparty is in default
`sector_code`	`String`	No	Industry sector code (SIC-based)
`apply_fi_scalar`	`Boolean`	No	User flag: True = apply 1.25x FI correlation scalar (CRR Art. 153(2))
`is_managed_as_retail`	`Boolean`	No	SME managed on pooled retail basis (75% RW per CRR Art. 123)
`is_natural_person`	`Boolean`	No	True for individuals (natural persons). Required to qualify a counterparty for the SA retail exposure class (CRR Art. 123 / PS1/26 Art. 123 — the class is restricted to "natural persons or to a small or medium-sized enterprise"). Also drives the Basel 3.1 retail RE classification gates (PS1/26 Art. 124E(1) primary-residence exception, Art. 124H natural-person CRE branch)
`is_social_housing`	`Boolean`	No	True for social-housing providers. Under CRR Art. 123(d), exposures to a "social housing provider, or [a] non-profit association regulated by law" are eligible for retail-class treatment without the natural-person test. Under Basel 3.1 the same population qualifies for the Art. 124E(1) materially-dependent exception so the loan-splitter treats it as non-income-producing residential RE (PRA PS1/26)
`is_financial_sector_entity`	`Boolean`	No	True for financial sector entities (FSEs) as defined in CRR Art. 119(5) — credit institutions, investment firms, financial institutions, insurance and reinsurance undertakings, and similar regulated/unregulated entities. Used in conjunction with `apply_fi_scalar` and `total_assets` for the LFSE 1.25x correlation gate (Art. 142(1)(4) — "large financial sector entity")
`is_ccp_client_cleared`	`Boolean`	No	True when the counterparty is a CCP whose exposures arise from client-cleared trades — drives the QCCP 4% RW (vs 2% proprietary) under CRR Art. 306 / PS1/26 Art. 306
`borrower_income_currency`	`String`	No	ISO 4217 currency in which the borrower earns income (or, for RRE indexation, the currency in which the borrower's repayment cashflows are denominated). Triggers two independent currency-mismatch tests: (1) PRA PS1/26 Art. 123A unhedged-FX retail uplift — when the exposure currency differs from `borrower_income_currency` for an unhedged retail exposure, the RW is multiplied by 1.5 (capped at 150%); (2) PS1/26 Art. 124(2)(c) RRE primary-residence carve-out — only available where the exposure currency matches the currency of the borrower's repayment cashflows
`local_currency`	`String`	No	Counterparty's domestic currency (ISO 4217). Two use sites: (1) CRR Art. 114(4) sovereign domestic-currency carve-out — exposures to a central government or central bank denominated and funded in that sovereign's own domestic currency may receive a 0% RW regardless of CQS (compared to `currency` on the loan/facility row); (2) currency-mismatch test for retail under PS1/26 Art. 123A when `borrower_income_currency` is unavailable — `local_currency` is then used as the proxy "income" currency. Distinct from `borrower_income_currency`, which is the obligor's actual income currency
`sovereign_cqs`	`Int32`	No	External CQS for the counterparty's sovereign (i.e. the central government of the country in `country_code`). Counterparty-level — distinct from the row-level `external_cqs` derived on each exposure (which captures the obligor's own rating, not its sovereign). Primary use site is the SA sovereign risk weight table (CRR Art. 114(2) / PS1/26 Art. 114(2) — exposures to central governments and central banks rated by a nominated ECAI). Drives the domestic-currency preferential treatment (CRR Art. 114(4): exposures to the sovereign and central bank denominated and funded in the domestic currency may receive 0% RW), the third-country preferential treatment (CRR Art. 114(5)–(6): same treatment for non-EU/UK sovereigns where the third country applies an equivalent supervisory regime). Also feeds the sovereign-derived institution floor (CRR Art. 119(2) — institution RW must not be more favourable than its home sovereign's RW; PS1/26 Art. 121(1) Table 5 derives unrated-institution SCRA grades from the home sovereign's CQS), and SCRA derivations
`institution_cqs`	`Int8`	No	External CQS for the counterparty when treated as an institution — counterparty-level field, distinct from the row-level `external_cqs` derived on each exposure. Used by ECRA institution risk weights (CRR Art. 120 Table 3 / PS1/26 Art. 120 Table 3). Under ECRA, rated institutions take a CQS-driven RW; unrated institutions fall through to SCRA (PS1/26 Art. 121, where unrated grades may themselves be derived from the home sovereign via Art. 121(1) Table 5). When both `sovereign_cqs` and `institution_cqs` are populated, the sovereign floor at Art. 119(2) caps the institution treatment
`scra_grade`	`String`	No	SCRA grade for unrated institutions: `"A"`, `"A_ENHANCED"`, `"B"`, `"C"` (Basel 3.1 CRE20.16-21 / PRA PS1/26 Art. 121). Use `"A_ENHANCED"` when the counterparty satisfies the Art. 121(5) quantitative thresholds (CET1 ratio ≥ 14% and leverage ratio ≥ 5%) — yields a 30% RW (>3m) vs the standard Grade A 40% (>3m) / 20% (≤3m) weights.
`is_investment_grade`	`Boolean`	No	Publicly traded + investment grade → 65% SA RW (Basel 3.1 CRE20.47)

Entity Type: The Single Source of Truth¶

The entity_type field is the authoritative source for determining both SA and IRB exposure classes. Each entity type maps to specific exposure classes for each approach. This design ensures consistent classification across the calculation pipeline.

Valid entity_type values:

Entity Type	SA Exposure Class	IRB Exposure Class	Regulatory Reference
Sovereign Class
`sovereign`	CENTRAL_GOVT_CENTRAL_BANK	CENTRAL_GOVT_CENTRAL_BANK	CRR Art. 112(a)
`central_bank`	CENTRAL_GOVT_CENTRAL_BANK	CENTRAL_GOVT_CENTRAL_BANK	CRR Art. 112(a)
RGLA Class (Regional Governments/Local Authorities)
`rgla_sovereign`	RGLA	CENTRAL_GOVT_CENTRAL_BANK	CRR Art. 115 - has taxing powers/govt guarantee
`rgla_institution`	RGLA	INSTITUTION	CRR Art. 115 - no sovereign equivalence
PSE Class (Public Sector Entities)
`pse_sovereign`	PSE	CENTRAL_GOVT_CENTRAL_BANK	CRR Art. 116 - govt guaranteed
`pse_institution`	PSE	INSTITUTION	CRR Art. 116 - commercial PSE
MDB/International Org Class
`mdb`	MDB	CENTRAL_GOVT_CENTRAL_BANK	CRR Art. 117(1) — non-named MDB; institution table (CRR) / dedicated Table 2B (PS1/26 Art. 117(1)(a))
`mdb_named`	MDB	CENTRAL_GOVT_CENTRAL_BANK	CRR Art. 117(2) / PS1/26 Art. 117(2) — named MDBs on the eligible list (e.g. IBRD, IFC, EIB, EBRD) qualify for 0% RW. Use `mdb` for non-named MDBs that fall back to the institution / Table 2B treatment
`international_org`	MDB	CENTRAL_GOVT_CENTRAL_BANK	CRR Art. 118 — international organisations (e.g. IMF, BIS); 0% RW
Institution Class
`institution`	INSTITUTION	INSTITUTION	CRR Art. 112(d)
`bank`	INSTITUTION	INSTITUTION	CRR Art. 112(d)
`ccp`	INSTITUTION	INSTITUTION	CRR Art. 300-311 (CCP treatment)
`financial_institution`	INSTITUTION	INSTITUTION	CRR Art. 112(d)
Covered Bond Class
`covered_bond`	COVERED_BOND	COVERED_BOND	CRR Art. 129 / PS1/26 Art. 129 — eligible covered bonds. Rated: CQS lookup against Table 6A (CRR) / Table 7 (B31). Unrated: derived from issuing institution's senior unsecured RW per Art. 129(5). B31 adds new due-diligence requirement (Art. 129(4A)) and expands the unrated derivation table from 4 to 7 entries
Corporate Class
`corporate`	CORPORATE	CORPORATE	CRR Art. 112(g)
`company`	CORPORATE	CORPORATE	CRR Art. 112(g)
Retail Class
`individual`	RETAIL_OTHER	RETAIL_OTHER	CRR Art. 112(h)
`retail`	RETAIL_OTHER	RETAIL_OTHER	CRR Art. 112(h)
Specialised Lending Class
`specialised_lending`	SPECIALISED_LENDING	SPECIALISED_LENDING	CRR Art. 147(8)
Equity Class
`equity`	EQUITY	EQUITY	CRR Art. 133
High-Risk Class (Basel 3.1 only — see warning below)
`high_risk`	HIGH_RISK	HIGH_RISK	PS1/26 Art. 128 — generic "items associated with particularly high risk"; 150% RW. Assessment criteria per Art. 128(3): high risk of loss from obligor default; impossible to assess on standard data
`high_risk_venture_capital`	HIGH_RISK	HIGH_RISK	PS1/26 Art. 128 — venture capital exposures held outside the equity exposure class (priority 3 equity takes precedence; this bucket is for non-equity VC exposures only). 150% RW
`high_risk_private_equity`	HIGH_RISK	HIGH_RISK	PS1/26 Art. 128 — private-equity exposures held outside the equity exposure class. 150% RW
`high_risk_speculative_re`	HIGH_RISK	HIGH_RISK	PS1/26 Art. 128 — speculative immovable property financing (e.g. land acquisition with uncertain end use). 150% RW. Distinct from ADC (Art. 124K) which has its own loan-splitter treatment
Other Items Class (CRR Art. 134 / PS1/26 Art. 134)
`other_cash`	OTHER	OTHER	CRR Art. 134(3) / PS1/26 Art. 134(3) — cash in hand and equivalent items; 0% RW
`other_gold`	OTHER	OTHER	CRR Art. 134(4) / PS1/26 Art. 134(4) — gold bullion held in own vaults or on an allocated basis (backed by bullion liabilities); 0% RW
`other_items_in_collection`	OTHER	OTHER	CRR Art. 134(2) / PS1/26 Art. 134(2) — cash items in the process of collection; 20% RW
`other_tangible`	OTHER	OTHER	CRR Art. 134(7) / PS1/26 Art. 134(7) — other tangible assets, prepayments, and accrued income (where the counterparty cannot be identified); 100% RW
`other_residual_lease`	OTHER	OTHER	CRR Art. 134(7) / PS1/26 Art. 134(7) — residual value of leasing exposures (i.e. the portion not captured as a lease receivable on the obligor); 100% RW

Art. 128 (high-risk items) is omitted from current UK CRR

Art. 128 was omitted from UK onshored CRR by SI 2021/1078 (the Capital Requirements Regulation (Amendment) Regulations 2021). The high-risk exposure class is therefore a dead letter under current UK CRR (pre-2027) — exposures tagged with any high_risk* entity_type will fall through to other classes under a CRR-mode run. The class is re-introduced under PRA PS1/26 Art. 128 with effect 1 January 2027. Use the high_risk* entity types only when running in Basel 3.1 mode, or accept that the row will be reclassified under CRR. See SA risk weights spec - High-risk exposures.

Why SA and IRB Classes Can Differ¶

For certain entity types, the regulatory treatment differs between SA and IRB approaches:

RGLA/PSE with sovereign treatment: Under SA, these use dedicated RGLA/PSE risk weight tables. Under IRB, those with government guarantees or taxing powers use the central govt/central bank IRB formula.
RGLA/PSE with institution treatment: Under SA, these use RGLA/PSE tables. Under IRB, commercial PSEs without sovereign backing use the institution IRB formula.
MDB/International Orgs: Under SA, named MDBs (Art. 117(2)) receive 0% RW; other MDBs use institution tables (CRR) or dedicated Table 2B (Basel 3.1). Under IRB, they use the central govt/central bank formula.

Additional Classification Flags¶

Column	Purpose	When Used
`apply_fi_scalar`	Directly controls whether FI scalar (1.25x correlation) applies	Financial sector entities with this flag set to True get FI scalar under IRB (CRR Art. 153(2))
`is_managed_as_retail`	SME managed on pooled retail basis	Can use 75% RW under SA (CRR Art. 123)

Financial Sector Entity (FSE) and FI Scalar¶

The 1.25x correlation multiplier (Art. 153(2)) applies to large financial sector entities (LFSEs) and unregulated financial sector entities. The LFSE total-assets threshold differs by framework:

Framework	LFSE threshold	Citation
CRR	Total assets ≥ EUR 70 billion	CRR Art. 142(1)(4)
Basel 3.1	Total assets ≥ GBP 79 billion	PS1/26 Glossary p. 78 (Note: "corresponds to Article 142(1)(4) of CRR")

How it works in the calculator: The apply_fi_scalar flag on the counterparty record is the sole input — the calculator does not automatically compare total_assets against the framework-specific threshold. The user is responsible for setting apply_fi_scalar = True on counterparties that meet the regulatory LFSE (or unregulated FSE) criteria for the framework in use. The classifier derives requires_fi_scalar directly from this flag with no entity-type gate.

Two distinct thresholds — do not conflate

LFSE total-assets threshold (EUR 70bn CRR / GBP 79bn B31) → 1.25x correlation multiplier (Art. 153(2), both CRR and Basel 3.1). Applies to the asset correlation coefficient R, not to the capital requirement directly.
GBP 440m annual revenue → F-IRB only approach restriction (Art. 147A(1)(e), Basel 3.1 only). Does not affect correlation.

These thresholds serve entirely different purposes and apply to different entity populations. See Key Differences for details.

Example:

import polars as pl

counterparties = pl.DataFrame({
    "counterparty_reference": ["CORP_001", "CORP_002", "SOV_001", "BANK_001", "PSE_001"],
    "counterparty_name": ["Acme Corp Ltd", "Beta Industries PLC", "UK Treasury", "Major Bank PLC", "Local Council"],
    "entity_type": ["corporate", "corporate", "sovereign", "bank", "pse_sovereign"],
    "country_code": ["GB", "GB", "GB", "GB", "GB"],
    "annual_revenue": [25_000_000.0, 500_000_000.0, None, None, None],
    "total_assets": [30_000_000.0, 600_000_000.0, None, 80_000_000_000.0, 500_000_000.0],
    "default_status": [False, False, False, False, False],
    "sector_code": ["62.01", "28.11", None, "64.19", None],
    "apply_fi_scalar": [False, False, False, False, False],
    "is_managed_as_retail": [False, False, False, False, False],
})

Classification Algorithm Summary¶

The classifier (engine/classifier.py) processes counterparties through these steps:

Entity Type Mapping: Maps entity_type to both SA and IRB exposure classes
SME Classification: Checks if annual_revenue < EUR 50m for corporates
Retail Threshold: Aggregates exposures by lending group against retail threshold (EUR 1m CRR / GBP 880k Basel 3.1)
Default Identification: Checks default_status for defaulted treatment
FI Scalar Determination: Identifies large/unregulated FSEs for 1.25x correlation
Approach Assignment: Assigns SA/F-IRB/A-IRB/Slotting based on IRB permissions

See Classification for the complete classification algorithm.

Facility Schema¶

Purpose: Defines committed credit facilities (parent nodes in exposure hierarchy). Facilities represent credit limits; actual drawings are captured in the Loan schema.

File: exposures/facilities.parquet

Column	Type	Required	Description
`facility_reference`	`String`	Yes	Unique identifier for the facility
`product_type`	`String`	No	Product classification
`book_code`	`String`	No	Portfolio/book classification
`counterparty_reference`	`String`	Yes	Link to counterparty
`value_date`	`Date`	No	Facility start date
`maturity_date`	`Date`	No	Final maturity date — used to derive `M` for IRB unless `effective_maturity` is populated
`facility_termination_date`	`Date`	No	Bank's contractual right-to-terminate date — overrides `maturity_date` for IRB `M` derivation when shorter (Art. 162(2)(g))
`currency`	`String`	No	ISO 4217 currency code
`limit`	`Float64`	No	Committed facility limit
`committed`	`Boolean`	No	Whether the facility is a binding commitment to lend (default `True`). When `False`, the facility is treated as unconditionally cancellable: the hierarchy resolver does not generate a `facility_undrawn` exposure row, so no commitment CCF / EAD / RWA is held against the unused headroom. Loans and contingents already mapped to the facility are unaffected and continue to flow through the pipeline as their own exposure rows, with normal counterparty / parent rollup and collateral allocation.
`lgd`	`Float64`	No	Internal LGD estimate (A-IRB)
`lgd_unsecured`	`Float64`	No	A-IRB unsecured LGD estimate, applied to the residual EAD after eligible collateral (CRR Art. 181)
`beel`	`Float64`	No	Best estimate expected loss
`has_sufficient_collateral_data`	`Boolean`	No	A-IRB attestation that the firm has data quality sufficient to recognise own-estimate collateral effects under Art. 181(1)(f). When `False`, the LGD floor framework treats the row as if no collateral data is available
`is_revolving`	`Boolean`	No	Revolving vs term facility (default `False`)
`is_qrre_transactor`	`Boolean`	No	QRRE transactor flag. Drives the Basel 3.1 SA 45% weight (Art. 123(3)(a)) and the IRB 0.05% PD floor (Art. 163(1)(c)). True iff the revolving account has been repaid in full at each scheduled repayment date for the previous 12 months, OR the overdraft has not been drawn for the previous 12 months (PRA Glossary p. 9). Per Art. 154(4), accounts with less than 12 months of repayment history must be flagged False. Assessed upstream — not validated by the calculator. See Transactor Exposure.
`seniority`	`String`	No	`senior` (default) or `subordinated` (affects F-IRB LGD)
`risk_type`	`String`	No	Off-balance sheet risk category (see below)
`underlying_risk_type`	`String`	No	Underlying-exposure risk type used by the CRM substitution path when the facility's own `risk_type` would otherwise mask the protection's true category
`ccf_modelled`	`Float64`	No	A-IRB modelled CCF (0.0-1.5)
`ead_modelled`	`Float64`	No	A-IRB modelled EAD — when populated and approach is A-IRB, replaces the `drawn_amount + ccf x undrawn_amount` derivation (CRR Art. 166(8a))
`is_short_term_trade_lc`	`Boolean`	No	Short-term trade LC for goods movement (Art. 166(9))
`is_payroll_loan`	`Boolean`	No	Payroll-deduction / pension lending — qualifies for the Basel 3.1 35% retail RW (PS1/26 Art. 123(2))
`is_buy_to_let`	`Boolean`	No	Buy-to-let property lending — excluded from SME supporting factor (CRR Art. 501)
`has_one_day_maturity_floor`	`Boolean`	No	Set True for short-term self-liquidating trade exposures and certain capital-markets-driven exposures eligible for the 1-day `M` floor under CRR Art. 162(3) (otherwise the standard 1-year floor applies)
`is_sft`	`Boolean`	No	Securities Financing Transaction — selects the F-IRB 0.5-year repo-style supervisory maturity (`M`) under CRR Art. 162(1) (a fixed value, not a floor; deleted under Basel 3.1)
`effective_maturity`	`Float64`	No	Explicit numeric `M` override (years) per CRR Art. 162(3) / PS1/26. When populated it supersedes the `maturity_date`-derived `M` and bypasses the 1-year floor — firm-owned judgement for short-term carve-outs

Valid product_type values:

Value	Description
`rcf`	Revolving credit facility
`term_loan`	Term loan facility
`mortgage`	Mortgage facility
`overdraft`	Overdraft facility
`credit_card`	Credit card facility
`trade_finance`	Trade finance facility
`guarantee`	Guarantee facility
`project_finance`	Project finance

Valid seniority values:

Value	F-IRB LGD	Description
`senior`	45%	Senior unsecured claims
`subordinated`	75%	Subordinated claims

Valid risk_type values (CRR Art. 111 CCF Categories):

Code	Full Value	SA CCF	F-IRB CCF	Description
`FR`	`full_risk`	100%	100%	Direct credit substitutes, guarantees, acceptances
`MR`	`medium_risk`	50%	75%	NIFs, RUFs, standby LCs, committed undrawn
`MLR`	`medium_low_risk`	20%	75%	Documentary credits, trade finance
`LR`	`low_risk`	0%	0%	Unconditionally cancellable commitments

Note: Under F-IRB (CRR Art. 166(8)), MR and MLR both become 75% CCF.

F-IRB Exception (Art. 166(9)): Short-term letters of credit arising from the movement of goods retain 20% CCF under F-IRB. To flag these exposures, set is_short_term_trade_lc = True for MLR risk type items.

Column	Type	Required	Description
`is_short_term_trade_lc`	`Boolean`	No	True for short-term trade LCs for goods movement (Art. 166(9) - retains 20% CCF under F-IRB)

A-IRB Modelled CCF: For A-IRB exposures, provide the bank's own modelled CCF estimate (0.0 to 1.5) in ccf_modelled. Retail IRB CCFs can exceed 100% due to additional drawdown behaviour. When populated and approach is A-IRB, this value takes precedence over the risk_type lookup.

Example:

from datetime import date
import polars as pl

facilities = pl.DataFrame({
    "facility_reference": ["FAC_001", "FAC_002"],
    "product_type": ["rcf", "term_loan"],
    "book_code": ["CORP_LENDING", "CORP_LENDING"],
    "counterparty_reference": ["CORP_001", "CORP_002"],
    "value_date": [date(2024, 1, 15), date(2023, 6, 1)],
    "maturity_date": [date(2029, 1, 15), date(2028, 6, 1)],
    "currency": ["GBP", "GBP"],
    "limit": [10_000_000.0, 5_000_000.0],
    "committed": [True, True],
    "lgd": [None, None],  # Supervisory LGD used for F-IRB
    "beel": [None, None],
    "is_revolving": [True, False],
    "seniority": ["senior", "senior"],
    "risk_type": ["MR", "MR"],  # Medium risk - committed undrawn
    "ccf_modelled": [None, None],  # No modelled CCF (use regulatory)
})

Loan Schema¶

Purpose: Defines drawn loan exposures (leaf nodes in exposure hierarchy). Loans represent actual credit usage under facilities.

File: exposures/loans.parquet

Column	Type	Required	Description
`loan_reference`	`String`	Yes	Unique identifier for the loan
`product_type`	`String`	No	Product classification
`book_code`	`String`	No	Portfolio/book classification
`counterparty_reference`	`String`	Yes	Link to counterparty
`value_date`	`Date`	No	Loan origination date
`maturity_date`	`Date`	No	Loan maturity date
`currency`	`String`	No	ISO 4217 currency code
`drawn_amount`	`Float64`	No	Outstanding principal balance (default `0.0`)
`interest`	`Float64`	No	Accrued interest (adds to on-balance-sheet EAD)
`lgd`	`Float64`	No	Internal LGD estimate (A-IRB)
`lgd_unsecured`	`Float64`	No	A-IRB unsecured LGD estimate, applied to the residual EAD after eligible collateral (CRR Art. 181)
`beel`	`Float64`	No	Best estimate expected loss
`has_sufficient_collateral_data`	`Boolean`	No	A-IRB collateral data-quality attestation per Art. 181(1)(f). See Facility schema
`seniority`	`String`	No	`senior` (default) or `subordinated` (affects F-IRB LGD)
`is_payroll_loan`	`Boolean`	No	Payroll-deduction / pension lending — qualifies for the Basel 3.1 35% retail RW (PS1/26 Art. 123(2))
`is_buy_to_let`	`Boolean`	No	Buy-to-let property lending — excluded from SME supporting factor (CRR Art. 501)
`has_one_day_maturity_floor`	`Boolean`	No	Eligible for the CRR Art. 162(3) 1-day `M` floor in lieu of the standard 1-year floor
`is_sft`	`Boolean`	No	Securities Financing Transaction — see Facility schema
`effective_maturity`	`Float64`	No	Explicit numeric `M` override (years) per CRR Art. 162(3) / PS1/26. Bypasses the 1-year floor when populated
`netting_agreement_reference`	`String`	No	CRR Art. 195/219 on-balance-sheet netting set. A non-null reference is the sole signal that the loan participates in a netting agreement; exposures net against each other iff they share the same reference — independent of facility or counterparty
`due_diligence_performed`	`Boolean`	No	Basel 3.1 Art. 110A: True if the firm has performed the prescribed due-diligence assessment of the obligor. Required for the SA RW override below to apply. Absence raises diagnostic warning SA004 under B3.1
`due_diligence_override_rw`	`Float64`	No	Basel 3.1 Art. 110A SA RW override (decimal, e.g. `1.50` for 150%). Applied as `max(calculated_rw, override_rw)` — the override can only increase the regulatory RW, never decrease it. CRR-only runs ignore this column

Note: Loans do not have CCF fields (risk_type, ccf_modelled, is_short_term_trade_lc) because CCF only applies to off-balance sheet items. For drawn loans, EAD = drawn_amount + interest directly.

Example:

from datetime import date
import polars as pl

loans = pl.DataFrame({
    "loan_reference": ["LOAN_001", "LOAN_002", "LOAN_003"],
    "product_type": ["rcf_drawing", "term_loan", "term_loan"],
    "book_code": ["CORP_LENDING", "CORP_LENDING", "CORP_LENDING"],
    "counterparty_reference": ["CORP_001", "CORP_001", "CORP_002"],
    "value_date": [date(2024, 3, 1), date(2024, 4, 15), date(2023, 6, 1)],
    "maturity_date": [date(2029, 1, 15), date(2029, 1, 15), date(2028, 6, 1)],
    "currency": ["GBP", "GBP", "GBP"],
    "drawn_amount": [2_000_000.0, 1_500_000.0, 5_000_000.0],
    "interest": [10_000.0, 7_500.0, 25_000.0],  # Accrued interest
    "lgd": [None, None, None],
    "beel": [None, None, None],
    "seniority": ["senior", "senior", "senior"],
    "is_buy_to_let": [False, False, False],
    "netting_agreement_reference": [None, None, None],
})

Contingent Schema¶

Purpose: Defines off-balance sheet commitments that require Credit Conversion Factor (CCF) application.

File: exposures/contingents.parquet

Column	Type	Required	Description
`contingent_reference`	`String`	Yes	Unique identifier
`product_type`	`String`	No	Product classification
`book_code`	`String`	No	Portfolio/book classification
`counterparty_reference`	`String`	Yes	Link to counterparty
`value_date`	`Date`	No	Contract start date
`maturity_date`	`Date`	No	Contract expiry date
`currency`	`String`	No	ISO 4217 currency code
`nominal_amount`	`Float64`	No	Notional/nominal amount (default `0.0`)
`lgd`	`Float64`	No	Internal LGD estimate (A-IRB)
`lgd_unsecured`	`Float64`	No	A-IRB unsecured LGD estimate, applied to the residual EAD after eligible collateral (CRR Art. 181)
`beel`	`Float64`	No	Best estimate expected loss
`has_sufficient_collateral_data`	`Boolean`	No	A-IRB collateral data-quality attestation per Art. 181(1)(f). See Facility schema
`seniority`	`String`	No	`senior` (default) or `subordinated`
`risk_type`	`String`	No	Off-balance sheet risk category (see Facility schema)
`underlying_risk_type`	`String`	No	Underlying-exposure risk type for CRM substitution (see Facility schema)
`ccf_modelled`	`Float64`	No	A-IRB modelled CCF (0.0-1.5)
`ead_modelled`	`Float64`	No	A-IRB modelled EAD — replaces the CCF-derived value when populated under A-IRB (CRR Art. 166(8a))
`is_short_term_trade_lc`	`Boolean`	No	Short-term trade LC for goods movement (Art. 166(9))
`has_one_day_maturity_floor`	`Boolean`	No	Eligible for the CRR Art. 162(3) 1-day `M` floor
`is_sft`	`Boolean`	No	Securities Financing Transaction — see Facility schema
`effective_maturity`	`Float64`	No	Explicit numeric `M` override (years) per CRR Art. 162(3) / PS1/26
`bs_type`	`String`	No	`"ONB"` (on-balance-sheet / drawn) or `"OFB"` (off-balance-sheet / undrawn). Default: `"OFB"`
`due_diligence_performed`	`Boolean`	No	Basel 3.1 Art. 110A due-diligence attestation (see Loan schema)
`due_diligence_override_rw`	`Float64`	No	Basel 3.1 Art. 110A SA RW override (max-only — see Loan schema)

Example:

from datetime import date
import polars as pl

contingents = pl.DataFrame({
    "contingent_reference": ["CONT_001", "CONT_002", "CONT_003"],
    "product_type": ["trade_finance", "guarantee", "import_lc"],
    "book_code": ["TRADE", "GUARANTEE", "TRADE"],
    "counterparty_reference": ["CORP_001", "CORP_002", "CORP_003"],
    "value_date": [date(2024, 1, 1), date(2024, 2, 1), date(2024, 3, 1)],
    "maturity_date": [date(2025, 1, 1), date(2025, 2, 1), date(2024, 6, 1)],
    "currency": ["GBP", "GBP", "GBP"],
    "nominal_amount": [500_000.0, 1_000_000.0, 250_000.0],
    "lgd": [None, None, None],
    "beel": [None, None, None],
    "seniority": ["senior", "senior", "senior"],
    "risk_type": ["FR", "MR", "MLR"],  # FR=100%, MR=50%/75%, MLR=20%/75%
    "ccf_modelled": [None, None, None],  # No modelled CCF
    "is_short_term_trade_lc": [False, False, True],  # Third is Art. 166(9) exception
    "bs_type": ["OFB", "OFB", "OFB"],  # Off-balance sheet (default)
})

Collateral Schema¶

Purpose: Defines collateral/security items used for Credit Risk Mitigation (CRM). Collateral can be linked at counterparty, facility, or loan level.

File: collateral/collateral.parquet

Column	Type	Required	Description
`collateral_reference`	`String`	Yes	Unique identifier
`collateral_type`	`String`	Yes	Type of collateral (see valid values)
`currency`	`String`	No	ISO 4217 currency code
`maturity_date`	`Date`	No	Collateral maturity (if applicable)
`market_value`	`Float64`	Conditional	Current market value (required unless `pledge_percentage` provided)
`nominal_value`	`Float64`	No	Nominal/face value
`pledge_percentage`	`Float64`	Conditional	Fraction of beneficiary EAD (0.5 = 50%). Used when `market_value` not provided. Resolved to absolute `market_value` before haircuts.
`beneficiary_type`	`String`	Yes	Level of allocation
`beneficiary_reference`	`String`	Yes	Reference to counterparty/facility/loan
`issuer_cqs`	`Int8`	No	CQS of issuer (for securities)
`issuer_type`	`String`	No	Issuer type (for haircut lookup)
`residual_maturity_years`	`Float64`	No	Residual maturity in years
`original_maturity_years`	`Float64`	No	Original maturity in years — used for collateral maturity-band differentiation where the haircut depends on issuance maturity rather than residual maturity
`is_eligible_financial_collateral`	`Boolean`	No	Meets SA eligibility (CRR Art 197)
`is_eligible_irb_collateral`	`Boolean`	No	Meets IRB eligibility (CRR Art 199)
`is_airb_model_collateral`	`Boolean`	No	Default `False`. When `True`, the firm asserts the collateral has been used to construct the internal LGD model (CRR Art. 181 / Basel 3.1 Art. 169A). The CRM allocator excludes flagged rows from non-AIRB exposures (no double-counting of the modelled-LGD effect) and routes them only to AIRB-pool exposures whose modelled LGD is preserved. Direct allocation of a flagged row onto a non-AIRB exposure raises CRM006
`is_qualifying_re`	`Boolean`	No	Meets Basel 3.1 qualifying real-estate criteria for the loan-splitter (PS1/26 Art. 124A finished-property / legal-enforceability / valuation / borrower-ability tests). Drives Art. 124F (RRE) / Art. 124H (CRE) eligibility
`is_main_index`	`Boolean`	No	Equity collateral listed on a recognised main index (e.g. FTSE 100) — qualifies for the lower main-index haircut (CRR Art. 197(8))
`valuation_date`	`Date`	No	Date of last valuation
`valuation_type`	`String`	No	`market`, `indexed`, `independent`
`property_type`	`String`	No	`residential` or `commercial` (RE only)
`property_ltv`	`Float64`	No	Regulatory LTV per Art. 124C — must include prior/pari passu charges (Art. 124C(3)) in numerator
`prior_charge_ltv`	`Float64`	No	LTV portion from prior/pari passu charges only (Art. 124C(3)); 0.0 = first charge. Used by Art. 124F(2)/124G(2) junior charge treatment
`is_income_producing`	`Boolean`	No	Material dependency on property cash flows per Art. 124E. `True` = materially dependent (residential: Art. 124G whole-loan; commercial: Art. 124I). `False`/null = not materially dependent (residential: Art. 124F loan-splitting; commercial: Art. 124H). For residential RE, this requires upstream assessment of Art. 124E(1) exceptions (primary residence, three-property limit, social housing, cooperative). For commercial RE, the own-business-use test (Art. 124E(6)). See Art. 124E spec
`is_adc`	`Boolean`	No	Acquisition/Development/Construction
`is_presold`	`Boolean`	No	ADC pre-sold to qualifying buyer
`rental_to_interest_ratio`	`Float64`	No	CRR Art. 126(2)(d): rental income / interest payments ratio. The ≥ 1.5 test gates the 50% preferential CRE risk weight; if absent, CRE collateral is conservatively treated as failing the test under CRR and the loan-splitter leaves the exposure in its original corporate / retail class. Not used under Basel 3.1.
`liquidation_period_days`	`Int32`	No	Liquidation period in business days — used to scale supervisory haircuts under the Comprehensive method when the firm's holding period differs from the supervisory minimum (CRR Art. 224)
`qualifies_for_zero_haircut`	`Boolean`	No	Repo-style transaction qualifying for the zero-haircut carve-out under CRR Art. 224(1) / Art. 227 (core market participants, eligible securities, daily margining)
`insurer_risk_weight`	`Float64`	No	Credit-protection insurer risk weight override — when funded credit protection is provided by an eligible insurer, this RW is substituted for the obligor RW on the protected portion (CRR Art. 235)
`credit_event_reduction`	`Float64`	No	Adjustment factor (decimal, default `0.0`) applied to the protection's covered amount when the contract restricts the set of credit events. Reduces recognised CRM accordingly (CRR Art. 213-216)

Valid collateral_type values:

Value	Description
`cash`	Cash collateral (0% haircut)
`gold`	Gold collateral (CRR 15% / B31 20% haircut)
`bond`	Bond securities — haircut depends on `issuer_type`, `issuer_cqs`, and `residual_maturity_years`
`equity`	Equity securities
`real_estate`	Real estate — use `property_type` for residential/commercial classification
`receivables`	Trade receivables
`other_physical`	Other physical collateral

Valid beneficiary_type values:

Value	Description
`counterparty`	Allocated at counterparty level (expands to all exposures)
`facility`	Allocated at facility level (expands to facility + child loans)
`loan`	Allocated directly to specific loan
`contingent`	Allocated directly to contingent

Example:

from datetime import date
import polars as pl

collateral = pl.DataFrame({
    "collateral_reference": ["COLL_001", "COLL_002"],
    "collateral_type": ["cash", "real_estate"],
    "currency": ["GBP", "GBP"],
    "maturity_date": [None, None],
    "market_value": [1_000_000.0, 500_000.0],
    "nominal_value": [1_000_000.0, None],
    "beneficiary_type": ["counterparty", "loan"],
    "beneficiary_reference": ["CORP_001", "LOAN_003"],
    "issuer_cqs": [None, None],
    "issuer_type": [None, None],
    "residual_maturity_years": [None, None],
    "is_eligible_financial_collateral": [True, False],
    "is_eligible_irb_collateral": [True, True],
    "valuation_date": [date(2024, 12, 31), date(2024, 11, 15)],
    "valuation_type": ["market", "independent"],
    "property_type": [None, "residential"],
    "property_ltv": [None, 0.65],  # Art. 124C: includes prior charges in numerator
    "prior_charge_ltv": [None, 0.0],  # Art. 124C(3): 0.0 = first charge
    "is_income_producing": [None, False],
    "is_adc": [None, False],
    "is_presold": [None, None],
})

Guarantee Schema¶

Purpose: Defines guarantee protection for Credit Risk Mitigation using the substitution approach.

File: guarantee/guarantee.parquet

Column	Type	Required	Description
`guarantee_reference`	`String`	Yes	Unique identifier
`guarantee_type`	`String`	No	Type of guarantee
`guarantor`	`String`	Yes	Guarantor counterparty reference
`currency`	`String`	No	ISO 4217 currency code
`maturity_date`	`Date`	No	Guarantee expiry date
`amount_covered`	`Float64`	Conditional	Amount covered by guarantee (required unless `percentage_covered` provided)
`percentage_covered`	`Float64`	Conditional	Fraction of exposure covered (`1.0` = 100%). Used when `amount_covered` not provided
`beneficiary_type`	`String`	Yes	Level of allocation
`beneficiary_reference`	`String`	Yes	Reference to counterparty/facility/loan
`protection_type`	`String`	No	`"guarantee"` (default) or `"credit_derivative"` — drives CRM treatment under CRR Art. 213 vs Art. 215 (credit-derivative-specific eligibility tests)
`includes_restructuring`	`Boolean`	No	True when the credit derivative includes restructuring as a credit event. Required for full recognition under CRR Art. 216(1)(d); when False the recognised cover is reduced to 60% per Art. 216(1)(e)

Valid guarantee_type values:

Value	Description
`guarantee`	Standard guarantee
`credit_derivative`	Credit derivative protection
`counter_guarantee`	Counter-guarantee

Example:

from datetime import date
import polars as pl

guarantees = pl.DataFrame({
    "guarantee_reference": ["GUAR_001"],
    "guarantee_type": ["guarantee"],
    "guarantor": ["SOV_001"],  # UK Treasury guaranteeing
    "currency": ["GBP"],
    "maturity_date": [date(2030, 12, 31)],
    "amount_covered": [2_000_000.0],
    "percentage_covered": [1.0],
    "beneficiary_type": ["counterparty"],
    "beneficiary_reference": ["CORP_001"],
})

Provision Schema¶

Purpose: Defines IFRS 9 provisions/impairments for EAD reduction and IRB expected loss comparison.

File: provision/provision.parquet

Column	Type	Required	Description
`provision_reference`	`String`	Yes	Unique identifier
`provision_type`	`String`	Yes	`scra` (specific) or `gcra` (general)
`ifrs9_stage`	`Int8`	No	IFRS 9 stage (1, 2, or 3)
`currency`	`String`	Yes	ISO 4217 currency code
`amount`	`Float64`	Yes	Provision amount
`as_of_date`	`Date`	Yes	Provision as-of date
`beneficiary_type`	`String`	Yes	Level of allocation
`beneficiary_reference`	`String`	Yes	Reference to counterparty/facility/loan

Valid provision_type values:

Value	Description	Usage
`scra`	Specific Credit Risk Adjustment	Reduces exposure value; affects defaulted RW
`gcra`	General Credit Risk Adjustment	Reduces exposure value

Valid beneficiary_type values:

Value	Description	Resolution
`loan`	Allocated directly to a specific loan	Matched by `beneficiary_reference` = `loan_reference`
`contingent`	Allocated directly to a contingent	Matched by `beneficiary_reference` = `contingent_reference`
`facility`	Allocated at facility level	Distributed pro-rata across facility's exposures by `ead_gross`
`counterparty`	Allocated at counterparty level	Distributed pro-rata across all counterparty exposures by `ead_gross`

Valid ifrs9_stage values:

Stage	Description	ECL Type
`1`	Performing	12-month ECL
`2`	Performing, significant increase in credit risk	Lifetime ECL
`3`	Non-performing/credit-impaired	Lifetime ECL

Example:

from datetime import date
import polars as pl

provisions = pl.DataFrame({
    "provision_reference": ["PROV_001", "PROV_002"],
    "provision_type": ["scra", "gcra"],
    "ifrs9_stage": [1, 2],
    "currency": ["GBP", "GBP"],
    "amount": [50_000.0, 100_000.0],
    "as_of_date": [date(2024, 12, 31), date(2024, 12, 31)],
    "beneficiary_type": ["loan", "counterparty"],
    "beneficiary_reference": ["LOAN_001", "CORP_002"],
})

Rating Schema¶

Purpose: Defines internal and external credit ratings for risk weight determination.

File: ratings/ratings.parquet

Column	Type	Required	Description
`rating_reference`	`String`	Yes	Unique identifier
`counterparty_reference`	`String`	Yes	Link to counterparty
`rating_type`	`String`	Yes	`internal` or `external`
`rating_agency`	`String`	Yes	Rating source
`rating_value`	`String`	Yes	Rating value (e.g., `AAA`, `Aa1`)
`cqs`	`Int8`	Yes	Credit Quality Step (1-6)
`pd`	`Float64`	No	Probability of Default (internal ratings)
`rating_date`	`Date`	Yes	Rating as-of date
`is_solicited`	`Boolean`	No	Whether rating was solicited
`model_id`	`String`	No	IRB model identifier — links to Model Permissions for per-model approach gating. Flows through rating inheritance pipeline to exposures. Null defaults to SA.
`is_short_term`	`Boolean`	No	`True` flags this row as a dedicated short-term ECAI assessment (PRA PS1/26 Art. 120(2B) Table 4A / Art. 122(3) Table 6A). Defaults to `False` (long-term, counterparty-wide).
`scope_type`	`String`	No	Which exposure the short-term rating attaches to: `facility`, `loan`, or `contingent`. Must be populated when `is_short_term=True`; must be null otherwise.
`scope_id`	`String`	No	Matching `facility_reference` / `loan_reference` / `contingent_reference`. Must be populated when `is_short_term=True`; must be null otherwise.

Short-term ECAI assessments (Art. 120(2B) / Art. 122(3)): Short-term assessments are issue-specific — they attach to a particular exposure rather than to the counterparty as a whole. Populate is_short_term=True together with scope_type and scope_id to override the counterparty-level long-term rating for the targeted exposure. When the scope is facility, the override propagates to all loans and undrawn portions drawn under that facility. The SA engine routes the overridden exposure via Table 4A (institution) or Table 6A (corporate). The producer is responsible for ensuring the underlying exposure satisfies the regulatory maturity test before flagging the rating short-term — the engine does not re-check maturity.

Valid rating_agency values:

Value	Description
`internal`	Internal rating system
`SP`	Standard & Poor's
`MOODYS`	Moody's
`FITCH`	Fitch Ratings
`DBRS`	DBRS Morningstar

CQS Mapping:

CQS	S&P/Fitch	Moody's	Sovereign RW	Institution RW	Corporate RW
1	AAA to AA-	Aaa to Aa3	0%	20%	20%
2	A+ to A-	A1 to A3	20%	30%*	50%
3	BBB+ to BBB-	Baa1 to Baa3	50%	50%	100%
4	BB+ to BB-	Ba1 to Ba3	100%	100%	100%
5	B+ to B-	B1 to B3	100%	100%	150%
6	CCC+ and below	Caa1 and below	150%	150%	150%

*CRR Art. 120 Table 3 assigns CQS 2 institutions 50%. Basel 3.1 ECRA (PRA PS1/26 Art. 120 Table 3) reduces this to 30%.

Example:

from datetime import date
import polars as pl

ratings = pl.DataFrame({
    "rating_reference": ["RAT_001", "RAT_002"],
    "counterparty_reference": ["CORP_001", "SOV_001"],
    "rating_type": ["external", "external"],
    "rating_agency": ["SP", "SP"],
    "rating_value": ["BBB+", "AA"],
    "cqs": [3, 1],
    "pd": [None, None],
    "rating_date": [date(2024, 6, 15), date(2024, 1, 1)],
    "is_solicited": [True, True],
})

FX Rates Schema¶

Purpose: Defines FX (foreign exchange) rates for converting exposure amounts from their original currencies to a reporting currency. This enables consistent RWA calculations across multi-currency portfolios.

File: fx_rates/fx_rates.parquet

Column	Type	Required	Description
`currency_from`	`String`	Yes	Source currency code (ISO 4217)
`currency_to`	`String`	Yes	Target currency code (ISO 4217)
`rate`	`Float64`	Yes	Conversion multiplier: `target_amount = source_amount * rate`

Usage: - Rates should be provided for all currency pairs needed (source → target) - Include identity rates (e.g., GBP→GBP = 1.0) for the target currency - The target currency should match CalculationConfig.base_currency

Converted Fields: - Exposures: drawn_amount, undrawn_amount, nominal_amount - Collateral: market_value, nominal_value - Guarantees: amount_covered - Provisions: amount

Audit Trail: After conversion, the following columns are added: - original_currency - Currency before conversion - original_amount - Amount before conversion (drawn + nominal) - fx_rate_applied - Rate used (null if no conversion needed)

Example:

from datetime import date
import polars as pl

fx_rates = pl.DataFrame({
    "currency_from": ["GBP", "USD", "EUR", "JPY", "CHF"],
    "currency_to": ["GBP", "GBP", "GBP", "GBP", "GBP"],
    "rate": [1.0, 0.79, 0.88, 0.0053, 0.89],
})

Behaviour: - Missing rates: Exposures in currencies without rates retain original values; fx_rate_applied is null - FX disabled: Set apply_fx_conversion=False in CalculationConfig to skip conversion - No FX file: If fx_rates.parquet is not provided, no conversion occurs

Specialised Lending Schema¶

Purpose: Defines specialised lending metadata for slotting approach treatment (CRE33). Keyed by counterparty_reference — all exposures to an SL counterparty inherit the same slotting treatment. This allows a corporate counterparty to have both SL and non-SL exposures.

File: ratings/specialised_lending.parquet

Column	Type	Required	Description
`counterparty_reference`	`String`	Yes	Links to counterparty (all exposures inherit SL treatment)
`sl_type`	`String`	Yes	Type of specialised lending
`project_phase`	`String`	No	Project lifecycle phase — typically `"planning"`, `"construction"`, or `"operational"`. Drives the CRE33.5 < 2.5-year strong/good preferential weights (which only apply during construction / pre-operational phase)
`slotting_category`	`String`	No	Slotting category
`is_hvcre`	`Boolean`	No	High-volatility commercial real estate

Valid sl_type values:

Value	Description
`project_finance`	Project finance (PF)
`object_finance`	Object finance (OF)
`commodities_finance`	Commodities finance (CF)
`ipre`	Income-producing real estate
`hvcre`	High-volatility commercial real estate

Valid slotting_category values:

Category	CRR RW	Description
`strong`	70%	Excellent risk profile
`good`	90% (70% if <2.5yr)	Good risk profile
`satisfactory`	115%	Acceptable risk profile
`weak`	250%	Higher risk profile
`default`	0%	In default (provisions apply)

SFT Input Schemas (FCCM)¶

Purpose: Dedicated input contract for securities financing transactions (SFTs) — repos, reverse repos, securities-borrowing/lending and margin-lending — whose exposure-at-default is computed by the Financial Collateral Comprehensive Method (FCCM) under CRR Art. 220–223. SFTs are a peer of the SA-CCR derivative book, not a sub-mode of it: they have their own input bundle, their own dataloads and their own pipeline stage.

Two unrelated meanings of \"SFT\"

The transaction_type == "sft" / FCCM path documented here (CCR EAD, CRR Art. 220–223) is a completely different concept from the is_sft Boolean carried on the loan / contingent / facility schemas (which selects the F-IRB 0.5-year repo-style supervisory maturity under CRR Art. 162(1)). Same acronym, different concept. (FCCM SFT rows that route to IRB carry their Art. 162 maturity via a dedicated ccr_effective_maturity carrier — never is_sft.) See The two meanings of "SFT" below.

How SFTs flow through the pipeline¶

SFT inputs are loaded into a dedicated bundle and processed by a dedicated stage that runs as a peer of the SA-CCR derivative stage:

sft_trades.parquet (+ optional sft_collateral.parquet)
   → RawSFTBundle  (RawDataBundle.sft)
   → sft_fccm stage  (engine/stages/sft.py)
   → E* = max(0, E·(1+HE) − CVA·(1−HC−HFX))   (CRR Art. 223(5))
   → one synthetic exposure row per netting set
        (risk_type = "CCR_SFT", ccr_method = "fccm_sft", drawn_amount = E*)
   → Classifier → CRM → SA / IRB exposure ladder

The sft_fccm stage sits immediately after ccr_sa_ccr in the literal pipeline registry, so the two regulatory EAD methods are adjacent and visible:

Stage	Input bundle	Regulatory EAD basis	Output `risk_type`
`ccr_sa_ccr`	`RawDataBundle.ccr`	SA-CCR `EAD = α·(RC + PFE)` (CRR Art. 274)	`CCR_DERIVATIVE`
`sft_fccm`	`RawDataBundle.sft`	FCCM `E*` (CRR Art. 220–223 via Art. 271(2))	`CCR_SFT`

The stage no-ops when RawDataBundle.sft is None (a firm with no SFT book is unaffected), and it re-seals its output to the existing ccr_exit edge brand so SFT and derivative rows share the same downstream contract.

Details: see the SFT (FCCM EAD) specification for the E* formula, haircut treatment, and the architectural separation from SA-CCR.

SFT Trade Schema¶

File: ccr/sft_trades.parquet (the single primary SFT dataload; optional)

One row per SFT. The netting-set counterparty is denormalised onto the trade row — FCCM's current scope is single-trade, single-counterparty netting sets (Art. 220(1)(a)), so no separate SFT netting-set table is needed. The three exposure_* columns carry the Art. 223(5) exposure-side volatility-haircut (HE) inputs.

Column	Type	Required	Description
`trade_id`	`String`	Yes	Unique identifier for the SFT
`netting_set_id`	`String`	Yes	Netting-set reference — used for the `ccr__<netting_set_id>` output reference and the collateral join
`counterparty_reference`	`String`	Yes	Counterparty (denormalised from the netting set) — drives the SA institution risk-weight lookup (CRR Art. 120(1) Table 3)
`notional`	`Float64`	Yes	`E` — the exposure amount lent / sold under the SFT (Art. 223(5))
`currency`	`String`	Yes	ISO 4217 currency code of the exposure
`maturity_date`	`Date`	Yes	SFT maturity date
`start_date`	`Date`	Yes	SFT start date
`exposure_collateral_type`	`String`	No	`HE` input — exposure-side security type for the Art. 224 Table 1 haircut lookup. Null when the exposure side is cash / a standard loan (`HE = 0`)
`exposure_security_cqs`	`Int8`	No	`HE` input — exposure-side issuer CQS
`exposure_security_residual_maturity_years`	`Float64`	No	`HE` input — exposure-side residual maturity (years)
`is_margined`	`Boolean`	No (default `False`)	Branch selector (Art. 224(2) final subparagraph). `False` → unmargined branch (today's behaviour); `True` → the margined Art. 285 MPOR branch
`remargining_frequency_days`	`Int16`	No (default `1`)	Dual role: `N_R` for the Art. 226 non-daily revaluation factor on the unmargined branch (1 = daily revaluation, factor = 1.0) / `N` in `MPOR = F + N − 1` on the margined branch
`mpor_floor_category`	`String`	No (default `repo_only`)	MPOR floor `F` selector (margined branch only): `repo_only` → 5 (Art. 285(2)(a)) / `other` → 10 (Art. 285(2)(b)) / `illiquid_or_large` → 20 (Art. 285(3)). Constrained to `VALID_MPOR_FLOOR_CATEGORIES`
`has_margin_dispute_doubling`	`Boolean`	No (default `False`)	`True` doubles `F` for the two quarters following more than two margin disputes (Art. 285(4))
`mpor_days_override`	`Int16`	No (default `null`)	Explicit MPOR in business days; supersedes the `F + N − 1` derivation when set

The five margining columns are inert on the unmargined-daily path

All five margining columns default so that is_margined = False with remargining_frequency_days = 1 reproduces today's E* exactly (the Art. 226 non-daily factor collapses to 1.0 at daily revaluation). They take effect only when is_margined = True (the Art. 285 MPOR branch) or remargining_frequency_days > 1 (non-daily revaluation). See the SFT specification — two mutually-exclusive branches.

These three columns were previously schema-orphaned

exposure_collateral_type, exposure_security_cqs and exposure_security_residual_maturity_years are first-class on SFT_TRADE_SCHEMA. They are not the identically-named columns on LOAN_SCHEMA / CONTINGENTS_SCHEMA (which serve an unrelated lending purpose) — grepping the name across schemas can mislead.

SFT Collateral Schema¶

File: ccr/sft_collateral.parquet (optional — appears only when securities are posted)

Collateral received against an SFT, keyed by netting_set_id (a genuinely different grain — 0..n securities per netting set — so it stays a separate optional file). Feeds the CVA·(1 − HC − HFX) collateral term of the FCCM E* formula. An uncollateralised SFT carries no collateral row, so for the common case this is literally one file.

Column	Type	Required	Description
`sft_collateral_reference`	`String`	Yes	Unique identifier for the collateral row
`netting_set_id`	`String`	Yes	Join key back to the SFT trade's `netting_set_id`
`collateral_type`	`String`	Yes	Collateral security type — `HC` input for the Art. 224 Table 1 lookup
`market_value`	`Float64`	No (default `0.0`)	`CVA` — collateral market value in the `E*` formula; `0.0` when unknown (no collateral credit)
`currency`	`String`	No	Collateral currency. Drives the `HFX` same-currency shortcut (Art. 224 Table 4: `HFX = 0` when collateral and exposure currencies match)
`issuer_cqs`	`Int8`	No	`HC` input — collateral issuer CQS
`residual_maturity_years`	`Float64`	No	`HC` input — collateral residual maturity (years)

`transaction_type` discriminator¶

The CCR/SFT split key is transaction_type on the SA-CCR TRADE_SCHEMA, constrained to VALID_TRANSACTION_TYPES = {"derivative", "sft"} via COLUMN_VALUE_CONSTRAINTS. A bad value (e.g. "SFT", "repo") would otherwise silently mis-route an SFT into the SA-CCR Art. 274 chain (≈£0 EAD); the value constraint surfaces it as a DQ006 data-quality error instead. SFT rows belong in SFT_TRADE_SCHEMA / RawDataBundle.sft, not on the SA-CCR trade frame.

Configuration¶

SFTConfig.method (on CalculationConfig.sft) selects the SFT EAD method per CRR Art. 271(2):

`method`	Meaning	Status
`"fccm"` (default)	Financial Collateral Comprehensive Method (Art. 220–223)	Implemented
`"var"`	VaR method (Art. 221)	Reserved — engine fails loud (`NotImplementedError`)
`"imm"`	Internal Model Method (Art. 283)	Reserved — engine fails loud

The method is exposed as the sft_method factory argument on CalculationConfig.crr() / .basel_3_1().

The two meanings of "SFT"¶

The acronym "SFT" denotes two unrelated concepts that never interact:

	FCCM SFT (this section)	Lending `is_sft`
Carrier	`transaction_type == "sft"` on `SFT_TRADE_SCHEMA`	`is_sft` Boolean on `LOAN` / `CONTINGENT` / `FACILITY` schemas
Concept	Securities financing transaction routed to FCCM CCR EAD	A lending exposure flagged as an SFT for the F-IRB maturity carve-out
Drives	The `sft_fccm` stage: `E* = max(0, E·(1+HE) − CVA·(1−HC−HFX))`	The F-IRB 0.5-year fixed supervisory maturity (`M`) for repo-style transactions, in place of the 2.5-year default
Regulatory basis	CRR Art. 220–223, Art. 271(2)	CRR Art. 162(1) (fixed `M = 0.5y`; deleted under Basel 3.1)
Engine site	`engine/sft/fccm.py`	`engine/irb/transforms.py`

The lending is_sft Boolean is deliberately not renamed as part of the SFT/FCCM separation — that would touch the sealed hierarchy_resolved edge, the IRB transforms and every fixture that sets is_sft — so it is reserved for a standalone future codemod. See the SFT (FCCM EAD) specification.

Equity Exposure Schema¶

Purpose: Defines equity holdings (SA only under Basel 3.1, IRB approaches withdrawn).

Column	Type	Required	Description
`exposure_reference`	`String`	Yes	Unique identifier
`counterparty_reference`	`String`	Yes	Link to counterparty
`equity_type`	`String`	No	Type of equity exposure (default `"other"`)
`currency`	`String`	No	ISO 4217 currency code
`carrying_value`	`Float64`	No	Balance sheet value
`fair_value`	`Float64`	No	Mark-to-market value
`is_speculative`	`Boolean`	No	Higher-risk equity (unlisted + business < 5yr)
`is_exchange_traded`	`Boolean`	No	Listed on recognised exchange
`is_government_supported`	`Boolean`	No	Government-supported programme
`is_significant_investment`	`Boolean`	No	>10% of CET1
`ciu_approach`	`String`	No	CIU treatment selector — `"look_through"` (Art. 132(3)), `"mandate_based"` (Art. 132(4)), or `"fallback"` (1,250% per Art. 132(2)). When null the calculator selects based on data availability and falls back conservatively
`ciu_mandate_rw`	`Float64`	No	Mandate-based RW (decimal) when `ciu_approach = "mandate_based"`. Calculated upstream from the fund's investment mandate
`ciu_third_party_calc`	`Boolean`	No	True when the look-through RW has been calculated by an eligible third party (custodian / depositary / management company) per Art. 132(3) — relaxes the holding-level evidence requirement
`fund_reference`	`String`	No	Join key into the CIU Holdings schema. Required when `ciu_approach = "look_through"` and `ciu_third_party_calc = False`
`fund_nav`	`Float64`	No	Fund NAV — denominator of the look-through RW formula `sum(holding_value_i × rw_i) / fund_nav`. See equity-approach spec

Valid equity_type values:

Value	Risk Weight	Description
`central_bank`	0%	Central bank equity holdings (Art. 133(6))
`listed`	100%	Exchange-traded equities
`exchange_traded`	100%	Listed on recognised exchange
`government_supported`	100% CRR / 250% B31	Government-supported programme
`unlisted`	250%	Unlisted equities
`private_equity`	250%	Private equity investments
`private_equity_diversified`	190%	Diversified private equity portfolio
`speculative`	400%	Higher-risk (unlisted + business < 5yr)
`ciu`	Look-through	Collective investment undertakings
`other`	250%	Other equity exposures

CIU Holdings Schema¶

Purpose: Per-holding constituents of a Collective Investment Undertaking (CIU) used by the look-through approach (Art. 132(3)). Each row represents one underlying exposure inside a fund, joined to the parent equity row via fund_reference. The look-through RW is computed as sum(holding_value_i × rw_i) / fund_nav and applied to the equity exposure's carrying value.

File: No canonical loader path — passed in-memory via DataSourceConfig.ciu_holdings_file (loader source: engine/loader.py:139). Not part of the standard fixture registry, since the CIU look-through is opt-in.

When required: Only when one or more EQUITY_EXPOSURE rows have ciu_approach = "look_through" and ciu_third_party_calc = False. If absent, CIU rows fall through to the mandate-based or punitive 1,250% fallback (Art. 132(2)) — see equity-approach spec.

Column	Type	Required	Description
`fund_reference`	`String`	Yes	Join key to `EQUITY_EXPOSURE.fund_reference` — uniquely identifies the parent fund
`holding_reference`	`String`	Yes	Unique identifier for the holding within the fund
`exposure_class`	`String`	Yes	Exposure class of the underlying holding (`"corporate"`, `"institution"`, `"sovereign"`, etc.) — drives the SA risk weight applied
`cqs`	`Int8`	No	Credit Quality Step of the underlying holding, where applicable
`holding_value`	`Float64`	No	Market value of the holding inside the fund — numerator of the look-through formula

Regulatory references:

CRR Art. 132: CIU treatment under the Standardised Approach
PRA PS1/26 Art. 132(2): Punitive 1,250% fallback when neither look-through nor mandate-based data is available
PRA PS1/26 Art. 132(3): Look-through approach — requires sufficient and frequent information about the underlying exposures
PRA PS1/26 Art. 132(4): Mandate-based approach — uses the fund's investment mandate / prospectus

Example:

import polars as pl

ciu_holdings = pl.DataFrame({
    "fund_reference": ["FUND_001", "FUND_001", "FUND_001"],
    "holding_reference": ["HOLD_001", "HOLD_002", "HOLD_003"],
    "exposure_class": ["corporate", "sovereign", "institution"],
    "cqs": [3, 1, 2],
    "holding_value": [40_000_000.0, 30_000_000.0, 30_000_000.0],
})

# Parent equity row links via fund_reference:
equity_exposures = pl.DataFrame({
    "exposure_reference": ["EQ_001"],
    "counterparty_reference": ["FUND_MGR_001"],
    "equity_type": ["ciu"],
    "ciu_approach": ["look_through"],
    "ciu_third_party_calc": [False],
    "fund_reference": ["FUND_001"],
    "fund_nav": [100_000_000.0],
    "carrying_value": [5_000_000.0],
    # ... other equity fields
})

Facility Mapping Schema¶

Purpose: Defines parent-child relationships between facilities, loans, and contingents.

File: exposures/facility_mapping.parquet

Column	Type	Required	Description
`parent_facility_reference`	`String`	Yes	Parent facility reference
`child_reference`	`String`	Yes	Child facility/loan/contingent reference
`child_type`	`String`	Yes	`facility`, `loan`, or `contingent`

Example:

import polars as pl

facility_mapping = pl.DataFrame({
    "parent_facility_reference": ["FAC_001", "FAC_001", "FAC_001"],
    "child_reference": ["LOAN_001", "LOAN_002", "FAC_001A"],
    "child_type": ["loan", "loan", "facility"],
})

Org Mapping Schema¶

Purpose: Defines organisation hierarchy for rating and turnover inheritance.

File: mapping/org_mapping.parquet

Column	Type	Required	Description
`parent_counterparty_reference`	`String`	Yes	Parent counterparty reference
`child_counterparty_reference`	`String`	Yes	Child counterparty reference

Example:

import polars as pl

org_mapping = pl.DataFrame({
    "parent_counterparty_reference": ["CORP_PARENT", "CORP_PARENT"],
    "child_counterparty_reference": ["CORP_001", "CORP_002"],
})

Lending Mapping Schema¶

Purpose: Defines lending groups for retail threshold aggregation.

File: mapping/lending_mapping.parquet

Column	Type	Required	Description
`parent_counterparty_reference`	`String`	Yes	Lending group lead reference
`child_counterparty_reference`	`String`	Yes	Member counterparty reference

Exposures are aggregated to the group level for retail eligibility (threshold: EUR 1m CRR Art. 123(c) / GBP 880k Basel 3.1 Art. 123(1)(b)(ii)).

Model Permissions Schema¶

Purpose: Defines per-model IRB approach permissions, enabling granular control over which exposures can use FIRB, AIRB, or slotting. When permission_mode=PermissionMode.IRB, model permissions drive all approach routing. When absent in IRB mode, the pipeline falls back to SA for all exposures with a warning.

Why model-level permissions matter: Banks typically have multiple IRB models, each approved for specific exposure classes, geographies, and portfolios. This schema allows the calculator to resolve the correct approach per-exposure based on the model it belongs to, rather than applying a single org-wide permission.

File: config/model_permissions.parquet

Column	Type	Required	Description
`model_id`	`String`	Yes	Unique model identifier (e.g., `"UK_CORP_PD_01"`) — referenced by `model_id` on internal ratings
`exposure_class`	`String`	Yes	ExposureClass value this permission covers (e.g., `"corporate"`, `"institution"`)
`approach`	`String`	Yes	Approved approach: `"foundation_irb"`, `"advanced_irb"`, or `"slotting"`
`country_codes`	`String`	No	Comma-separated ISO country codes where this permission applies. Null = all geographies.
`excluded_book_codes`	`String`	No	Comma-separated book codes excluded from this permission. Null = no exclusions.

Optional columns

When country_codes or excluded_book_codes columns are entirely absent from the input file, they are treated as null for all rows — all geographies are permitted and no book codes are excluded. You only need to include these columns if you want to restrict permissions by geography or exclude specific book codes.

Approach determination logic:

Condition	Result
AIRB permission + `internal_pd` + modelled `lgd`	A-IRB
AIRB permission + `internal_pd` + no `lgd`	SA (falls back unless FIRB permission also exists)
FIRB permission + `internal_pd` + no `lgd`	F-IRB (uses regulatory LGD floors)
Both AIRB + FIRB permissions + `internal_pd` + `lgd`	A-IRB (higher approach wins)
Both AIRB + FIRB permissions + `internal_pd` + no `lgd`	F-IRB (fallback)
Slotting permission + `internal_pd`	Slotting
No `model_id` on internal rating	SA (default)

Example:

import polars as pl

model_permissions = pl.DataFrame({
    "model_id": ["UK_CORP_PD_01", "UK_CORP_PD_01", "EU_INST_01"],
    "exposure_class": ["corporate", "corporate_sme", "institution"],
    "approach": ["advanced_irb", "foundation_irb", "advanced_irb"],
    "country_codes": ["GB", "GB,IE", None],  # None = all geographies
    "excluded_book_codes": [None, "LEGACY_BOOK", None],
})

Data Preparation Checklist¶

Before running the calculator, verify your data meets these requirements:

[ ] All required files present in expected locations
[ ] Column names match schema exactly (case-sensitive)
[ ] Data types match expected types
[ ] All required columns have non-null values
[ ] Reference columns have valid foreign key relationships
[ ] Dates are in YYYY-MM-DD format
[ ] Currency codes are valid ISO 4217
[ ] Country codes are valid ISO 3166-1 alpha-2
[ ] Numeric amounts are non-negative where expected
[ ] PD values are in range [0, 1]
[ ] LGD values are in range [0, 1.25]

See Data Validation Guide for validation functions and troubleshooting.

Next Steps¶

Data Validation Guide - Validation rules and error handling
Intermediate Schemas - Pipeline intermediate data
Output Schemas - Calculation results

Input Data Schemas¶

Quick Reference¶

Counterparty Schema¶

Entity Type: The Single Source of Truth¶

Why SA and IRB Classes Can Differ¶

Additional Classification Flags¶

Financial Sector Entity (FSE) and FI Scalar¶

Classification Algorithm Summary¶

Facility Schema¶

Loan Schema¶

Contingent Schema¶

Collateral Schema¶

Guarantee Schema¶

Provision Schema¶

Rating Schema¶

FX Rates Schema¶

Specialised Lending Schema¶

SFT Input Schemas (FCCM)¶

How SFTs flow through the pipeline¶

SFT Trade Schema¶

SFT Collateral Schema¶

transaction_type discriminator¶

Configuration¶

The two meanings of "SFT"¶

Equity Exposure Schema¶

CIU Holdings Schema¶

Facility Mapping Schema¶

Org Mapping Schema¶

Lending Mapping Schema¶

Model Permissions Schema¶

Data Preparation Checklist¶

Next Steps¶

`transaction_type` discriminator¶