How Daily P&L Investigation Works in ETRM

"Why did today's P&L change?" is one of the most frequent questions asked on a trading floor, and answering it well - quickly, and with a clear attribution rather than a shrug - is one of the clearest signals of a mature ETRM and risk process.

The attribution chain

A daily P&L movement is rarely caused by one thing. Experienced ETRM and risk teams walk the same attribution chain every time, in roughly this order:

Trade changes

Start with what actually happened to the book: new trades, amendments, cancellations. Any of these directly change the position being valued, and it's the first and easiest thing to rule in or out.

Market price changes

Next, isolate how much of the movement is explained purely by the market moving against yesterday's position, holding the position itself constant. This is usually the largest single driver of daily P&L movement in an actively traded book, and separating it out clearly is what lets a desk distinguish "the market moved" from "something is wrong."

Position changes

Beyond new trades, positions can shift from settlements rolling off, options expiring, or scheduled physical deliveries reducing an open position. These changes affect P&L independently of both new trading activity and market moves.

FX and fees

For any book with cross-currency exposure, FX rate movement needs to be isolated as its own attribution line - otherwise it gets silently absorbed into "market price changes" and makes that number harder to trust. Transaction fees, broker commissions and similar costs are smaller in magnitude but should still be attributed explicitly rather than left as unexplained residual.

Settlement and valuation

Finally, check whether anything settled during the period - realized P&L crystallizing out of unrealized - and whether the valuation methodology itself changed (a curve rebuild, a change in discounting convention, a new volatility surface). These are less frequent but can produce large, one-time P&L movements that look alarming until identified.

Why the residual matters most

After walking through every line above, there's often a small residual that doesn't map cleanly to any single cause. A well-built attribution process makes this residual small and explainable (rounding, timing differences between systems). A poorly built one leaves a large, unexplained residual - which is usually the real signal that something in the underlying data or logic needs investigation, not the daily P&L number itself.