Shell Energy Trading: Multi-Desk P&L Platform
Real-time profit and loss across six trading desks, multi-commodity, front-to-back — replacing the spreadsheets that the ETRM system couldn't.
Six desks, five commodities, one platform
Shell Energy's European trading operation spans power, natural gas, LNG, carbon, and coal across multiple markets and regulatory jurisdictions. Six-plus trading desks, each with its own P&L requirements, risk views, and regulatory obligations. The existing ETRM system handled trade capture and basic position management. Everything else — P&L calculation, reconciliation, regulatory reporting — was done in spreadsheets.
Spreadsheets running a multi-commodity trading operation
Each desk maintained its own Excel workbooks for P&L. The numbers were credible most of the time — until they weren't. Month-end reconciliation consumed days. Traders waited for yesterday's P&L while making today's decisions. The specific failures:
- No real-time visibility. P&L was a batch process. By the time a number was final, the market had moved.
- No consistency across desks. Each desk calculated P&L differently. Aggregation to the portfolio level was manual and error-prone.
- Regulatory exposure. EMIR and REMIT reporting required data that lived in spreadsheets, not systems. Every reporting cycle was a scramble.
- Unit-of-measure chaos. Gas traded in therms, pence per therm, MWh equivalents. Power in MW and MWh. Emissions in tonnes. Converting between them in Excel invited errors that reached P&L.
- The vendor couldn't solve it. The ETRM system was designed for trade capture, not for the bespoke P&L views each desk needed. Customisation requests took quarters, not weeks.
Three layers of P&L, one source of truth
We designed a multi-layer P&L architecture that gave each audience what they needed without compromising on accuracy or speed.
Multi-layer P&L
- Fast P&L — real-time, streaming, approximate. Traders see their position move with the market. Updated on every tick via Kafka and KSQL.
- Human P&L — near-real-time, enriched with desk-specific adjustments and overrides. What the desk head uses to manage intraday risk.
- Accurate P&L — end-of-day, fully reconciled against the ETRM and settlement systems. What goes to finance and the regulators.
Unit-of-measure standardisation
A unified conversion layer across all gas hubs (TTF, THE, PEG, NBP) and power markets, handling therms, kWh, MWh, and emissions tonnes. Conversions applied once at ingestion, not ad hoc in downstream calculations.
Streaming architecture
Kafka for event sourcing, KSQL for real-time aggregation. Trade events, market data, and position updates flow through a single pipeline. Each P&L layer subscribes to the events it needs at the latency it requires.
Automated regulatory reporting
EMIR and REMIT reports generated directly from the platform's transaction and position data — no manual extraction from spreadsheets, no last-minute scrambles before filing deadlines.
From spreadsheets to production — still running
Replaced Excel P&L across six-plus trading desks
Every desk moved from manual spreadsheet P&L to the platform. No desk-specific workarounds, no parallel runs.
Real-time P&L visibility for the first time
Traders saw their P&L move with the market rather than waiting for yesterday's batch. Desk heads could manage intraday risk with current numbers.
Full front-office to back-office lifecycle
Trade capture through to settlement, regulatory reporting, and finance reconciliation — one system, one data lineage.
Regulatory reporting automated
EMIR and REMIT filings generated from system data. Reporting cycles that used to consume days became routine.
Still in production
The platform continues to run Shell Energy's European trading P&L. It wasn't a proof of concept — it was built to last.
If your vendor can't deliver it, this is a solved problem
If your trading desks are still on spreadsheets for P&L, or your ETRM vendor keeps promising customisation that never arrives — this problem has been solved before. The architecture exists. The implementation patterns are proven. The question is whether to keep waiting for your vendor or build what you actually need.
The same approach — multi-layer P&L, streaming architecture, automated regulatory reporting — applies whether you trade power, gas, carbon, or all of them. The specifics change; the design principles don't.