UMR & Funding

Uncleared margin and IM funding analytics for equity total return swap books versus a hedge counterparty.

Overview

The UMR & Funding dashboard shows how initial margin (IM) and variation margin (VM) flow between a client equity TRS book and a London hedge counterparty, and how a schedule-based IM proxy (single-name vs index supervisory factors, with per-name netting) compares to posted IM. It also surfaces MVA-style funding cost using SOFR plus a configurable spread over an average trade life, SaiMM (live equity-delta IM), and OpenGamma batch outputs when public/saimm/latest.json is present.

This is not a production SIMM or UMR Phase 6 engine — it is a transparent, editable workbook-style view for institutional workflows and education.

Open the tool: /regulatory/umr. The broader regulatory hub is at /regulatory.

Description and examples (PDF): description and examples

What the dashboard shows

Area	Content
Client book	One synthetic client TRS row per equity line; optional Client / NettingSet column groups lines into netting sets (posted IM summed per set for the $50mm threshold chart). Schedule IM, UMR utilization, threshold charts
Hedge	Aggregate hedge leg vs London-style CP: notional, IM/VM posted — scaled when you load a new portfolio so funding gaps stay interpretable
Funding	Net IM / VM gaps, waterfall, MVA panel (SOFR, spread, discount), breakeven bps on client notional
Portfolio	Generate a synthetic book from benchmark indices (same CSV conventions as the Broker Margin Optimizer) or paste / upload CSV — see below

Schedule IM proxy

Per-row schedule IM uses supervisory-style factors: 15% for single-name equity underliers and 10% for broad index proxies (e.g. SPY, QQQ, IWM), with optional maturity factor. Cross-name netting within the same underlier applies a simple netting benefit when long and short IM exist for that name.

Posted IM on each row is derived from notional and factor so the dashboard stays consistent with the schedule view. SaiMM (equity delta IM) and OpenGamma / OpenSIMM batch reference material are shown in the SaiMM panel; see the repo opengamma/README.md for JVM setup.

SaiMM (equity delta IM)

The SaiMM stack (Python package src/simm in the repo) implements a proprietary equity delta initial margin (branded SaiMM, not ISDA SIMM), writes CRIF-shaped files, and can optionally run OpenGamma OpenSIMM (OPENSIMM_JAR) or a custom JVM for batch reference output — see src/simm/opengamma/README.md. OpenSIMM 1.0 outputs are from OpenGamma’s deprecated historical VaR (HVaR) model, not current SIMM margin.

On the UMR dashboard (/regulatory/umr), SaiMM recomputes in the browser whenever you load or change the portfolio: same formula as the Python engine (per netting_set_id), with totals, per netting set, and per-name (netted) tables. Ticker buckets come from a generated table (npm run build:saimm-buckets in web/): it merges equity_buckets.csv with every ticker listed under Optimizer index constituents (INDEX_MAP), using the same sector→bucket heuristic for names not in the base CSV. Anything still unknown uses bucket 12.

The Python batch loads equity_buckets_merged.csv when present (same merge). It is still used for CRIF archives, OpenGamma runs, and optional public/saimm/latest.json snapshots. Run with:

python -m src.simm.run_daily --write-web (from the primeRiskReport Python project root, with PYTHONPATH=.).

Aligning batch CSV with the dashboard: set SAIMM_POSITIONS_CSV to positions with the same netting_set_id values as the optional Client / NettingSet column on the UMR CSV.

Portfolio input

The UMR screen reuses the Broker Optimizer portfolio workflow:

Generate — pick an index benchmark, long/short counts, diversification, short % of gross, total MV ($M). Lines are assigned round-robin to four notional client netting sets (NS_CLIENT_1 … NS_CLIENT_4), i.e. four documented counterparties.
Paste or upload CSV — Ticker, Shares, Side (optional: MV in $M, Sector, Country). Same base format as /optimizer.
Optional column: Client, NettingSet, or NS — one netting set per documented counterparty (signed CSA). All lines with the same value belong to that client’s portfolio (max 4 distinct values in this dashboard). Example: four clients × three names = twelve rows with four distinct Client values (e.g. C1…C4). If omitted, every line is grouped under a single netting set (NS_CLIENT_1) — one counterparty.

Each uploaded line is mapped to a synthetic client TRS row; posted IM in the $50mm chart is aggregated per netting set. The hedge block is recalibrated so total notional and dealer IM scale with the new book (illustrative funding gap).

References and PDF

description and examples — PDF (same document linked on the UMR page header).

For regulatory background on uncleared margin and SIMM, see industry materials from ISDA and BCBS/IOSCO; implementation details vary by phase and asset class.

K-Factor (MIFIDPRU) — prudential capital for investment firms (separate from UMR, but often used in the same regulatory workflows).
Broker Margin Optimizer — portfolio CSV format shared with UMR portfolio load.