Hedging Factor Exposures

Back to the wobbly table: instead of planing down the long leg — neutralisation — you can just jam a wedge underneath the short one. Nothing about the table itself changes; you've added something external that cancels the wobble. Hedging a factor exposure works the same way. Rather than touching a single stock in the underlying portfolio, you add a separate, liquid instrument — an index future, a sector ETF, a swap — sized so its factor exposure exactly offsets the book's unwanted tilt.

Sizing the wedge

The hedge notional needed is the portfolio's dollar exposure to the factor, divided by the hedge instrument's own exposure to that same factor:

$$N_{\text{hedge}} = -\frac{\text{Portfolio exposure}}{\beta_{\text{hedge instrument}}} .$$

In words: figure out how many dollars of the unwanted factor the portfolio is carrying, figure out how much of that same factor one dollar of the hedge instrument delivers (its beta, or its factor loading), and divide — that tells you how many dollars of the hedge to short (the minus sign) to bring the net back to zero. The hedge doesn't need to be a perfect twin of the exposure; it needs the same factor loading, which is why a broad index future can hedge a diversified book's market beta even though the future and the book don't hold a single identical stock.

Worked example 1 — hedging book beta with index futures

A long-only book holds $50m of stocks with a portfolio beta of 1.3 against the S&P 500 — meaning it's expected to move 1.3% for every 1% the index moves. Its dollar beta exposure is $50 \times 1.3 = 65$, i.e. $65m. An E-mini S&P 500 future has, by construction, a beta of 1.0 to the index. The hedge needed is $65 / 1.0 = 65$, i.e. $65m notional of futures, sold short. If the S&P then falls 4%, the book's stocks are expected to fall roughly $1.3 \times 4\% = 5.2\%$, i.e. lose about $2.6m, while the short futures position gains roughly $65 \times 4\% = 2.6$, i.e. $2.6m — the two moves offset, leaving the book's factor-driven P&L close to flat and exposing mostly the stock-specific bets the manager actually wanted to keep.

Worked example 2 — hedging a sector tilt, not just beta

The same book also carries a $20m net overweight to the Energy sector relative to its benchmark, with an Energy sector-factor loading of 1.0 on that $20m. An Energy sector ETF has a factor loading of roughly 0.95 to the same sector factor (it isn't a perfect replica of the model's sector definition). The hedge needed is $20 \times 1.0 / 0.95 \approx 21.1$, i.e. roughly $21.1m short in the ETF — slightly more notional than the raw $20m exposure, precisely because the ETF's own loading to the factor is a little less than 1.0.

The collar above is a useful mental model even though it's option jargon: a hedge is exactly this kind of asymmetric add-on — a separate position layered on top of an existing one, built purely to reshape the combined exposure, without touching the underlying holding itself. Drag the strikes and notice the underlying position never moves; only the added leg does.

What this means in practice

Hedging is the tool of choice when the unwanted exposure is temporary, when there's a liquid instrument available to cancel it cheaply, or when rebuilding the underlying portfolio's own weights would trigger unwanted trading costs or taxes. It's also imperfect by construction — a futures or ETF hedge only cancels the factor exposure it's built to track, so any gap between the hedge's true factor loading and the model's estimate of it (basis risk) leaves a residual the book still carries.