Daily PnL Attribution in Production

Knowing that a portfolio made $40,000 today is much less useful than knowing why. Daily PnL attribution takes the single number a P&L report shows and splits it into contributions — by position, by factor exposure, by trade — so a portfolio manager can tell the difference between "the strategy worked as designed" and "we got lucky on one name that has nothing to do with our edge." Without attribution, a string of good days and a string of bad days look identical from the outside; attribution is what lets someone say which days were driven by the intended signal and which were driven by something else entirely.

The basic mechanics are simple even when the bookkeeping gets fiddly: for each position, multiply the price change by the position size held during the day, then add financing costs, fees, and any dividend or coupon adjustments. Summing across positions should reconcile to the reported total P&L; if it doesn't, that gap itself is a flag worth chasing, since it often means a trade, a corporate action, or a fee was booked incorrectly somewhere upstream. Beyond the position-level split, attribution is often pushed further into factor terms — how much of the day came from market beta, how much from the specific stock-picking signal, how much from sector tilts the strategy didn't intend to take.

A concrete example: a long-short equity fund is up $50,000 on the day. Position-level attribution shows $45,000 came from a single large short that got squeezed favorably after a competitor's earnings miss, while the other 200 positions netted only $5,000 combined. The headline number looks like a good day for the strategy; the attribution shows it was really one lucky idiosyncratic outcome unrelated to the systematic signal the fund is supposed to be running. A manager who only watches the top-line P&L would miss that the "real" strategy barely broke even.

Attribution done daily, rather than reconstructed after the fact during a quarterly review, is what makes it useful for catching problems early — a factor tilt that's been quietly building for two weeks shows up immediately in the daily breakdown, long before it would show up as a headline drawdown.