Gross and Net Exposure Accounting
Gross and net exposure are supposed to be simple ratios, but a backtest that sizes them off a stale NAV silently reports a leverage number that drifts away from what the book is actually running, especially after a drawdown.
Gross exposure is the sum of the absolute value of every position, divided by net asset value; net exposure is the sum of signed positions, divided by the same. Both definitions look immune to error — they're just a ratio. The part that quietly breaks a backtest is the denominator: NAV changes every day as the book gains or loses money, and a simulation that computes position sizes off a starting capital figure instead of current NAV reports an exposure ratio that drifts from the number the strategy actually intended to run, in a direction that gets worse exactly when it matters most.
Worked example: 160% gross that becomes 200% after a drawdown, silently
A 130/30 long-short book is designed to run at 160% gross (130% long, 30% short) at all times, rebalanced against current NAV. A backtest bug instead sizes every position off the account's starting capital of $100 million, fixed for the whole simulation, rather than the NAV as of each rebalance.
| Intended (against current NAV) | Bug (against starting $100M) | |
|---|---|---|
| Starting NAV | $100M | $100M |
| Gross notional at inception | $160M (160%) | $160M (160%) |
| NAV after a 20% drawdown | $80M | $80M |
| Gross notional held (bug keeps sizing off $100M) | rebalances down to $128M (160% of $80M) | stays near $160M (unchanged) |
| Real gross exposure ratio | 160% | $160M / $80M = 200% |
The bug doesn't announce itself — gross notional in dollar terms looks perfectly stable across the drawdown, which is exactly why it's easy to miss on a quick check. But measured the way a risk desk actually measures it, against current NAV, the book has silently levered up from 160% to 200% at the worst possible time, right after losing a fifth of its capital, because the position-sizing logic never re-based off the number that had actually changed.
Gross and net exposure must be computed against current NAV at every rebalance, not against a fixed starting capital figure — a book that looks stable in dollar terms can be silently levering up as NAV falls beneath it.
This bug is most dangerous in a drawdown precisely because that's when real leverage rises fastest under it, compounding losses at the moment a risk desk would normally be forcing exposure down, not letting it drift up.
The check is simple and worth automating: at every rebalance, recompute gross and net exposure directly from current position values divided by current NAV, and assert the result stays within the intended band — never trust a dollar-notional figure alone as a proxy for exposure.
Related concepts
Practice in interviews
Further reading
- Grinold & Kahn, Active Portfolio Management (ch. on portfolio construction)