Energy Trading / Europe / 2020
A real-time risk engine for a regional energy trading desk
The desk was managing exposure on end-of-day positions in a market that moved intraday. We built a multi-factor risk system fed by live market data, giving continuous assessment and a defensible basis for portfolio optimisation.
Disciplines
- Risk systems
- Market data
- Quantitative modelling
Context
The trading company’s risk process was sound in construction and stale in operation: exposure was assessed against positions and prices that no longer held by the time the assessment was read. In a market with the volatility profile of regional energy, that lag was itself a material risk, and it was invisible in the reporting.
Approach
We built the risk engine around a real-time market data path first, on the principle that a simple model on current data outperforms a sophisticated model on yesterday’s. Once that path was established and proven under load, we layered a multi-factor model over it to produce dynamic exposure assessment across the portfolio.
The optimisation layer was deliberately advisory. The desk retained the decision; the system’s role was to make the consequences of a position legible at the moment the position was being considered, not to take the trade.
Outcome
The desk moved from periodic to continuous risk assessment, reducing exposure to intraday market movement and improving portfolio-level return. Because the model’s factors are explicit and inspectable, risk positions can be explained to the firm’s own committee — a requirement that shaped the design as much as any performance target.