Cross-Hedging With A Proxy

Sometimes the risk you're carrying has no direct hedge at all. There's no listed instrument on a private company's supplier, no futures contract on a thinly-traded regional bank, no liquid options on a mid-cap that trades $2m a day. You still need to reduce the exposure, so you reach for a proxy: something else, liquid, that tends to move with the risk you actually have — even though it isn't the risk you actually have.

The trade you're really making

A cross-hedge is a bet that the correlation between your position and the proxy holds up when it matters. It usually holds up fine on an ordinary day. It's far less reliable on the specific days a hedge exists to protect you — a company-specific blowup, a regulatory action against one bank but not the sector, an earnings miss that hits your name and nothing else. That's the uncomfortable truth about proxy hedges: they tend to work best exactly when you need them least, and to fail exactly when the loss they're supposed to cushion is largest.

This isn't a reason to avoid cross-hedging — an imperfect hedge is still better than no hedge for most unrewarded risks. It's a reason to size the hedge conservatively and never treat "hedged" as a synonym for "safe."

Picking a proxy

The candidates, roughly best to worst for a typical single-name proxy problem:

• A close peer — another company in the same sub-sector, similar size, similar business mix. Highest correlation on normal days, but shares almost none of your name's idiosyncratic risk.
• A sector ETF or index — broader, more liquid, lower correlation to the specific name, but very reliably tradable in size.
• A macro factor (rates, credit spreads, a currency) — use only when the exposure genuinely is macro-driven, e.g. a regional bank whose main risk really is the rate environment more than anything company-specific.

A worked case

You hold $3m of a regional bank with almost no listed options and thin equity volume — maybe $1.5m a day. You can't short the name itself in size without moving it, and there's no futures contract on it. You choose to cross-hedge with a regional-banking sector ETF that has historically moved with about 0.65 correlation to this specific name over the last two years.

You don't hedge the full $3m one-for-one — that would overstate how much protection the ETF actually provides given the imperfect correlation. Instead you size the short at roughly 65 percent of the position's dollar value, about $1.95m, reflecting the historical relationship rather than pretending it's a one-for-one hedge. On an ordinary down day for regional banks, this works close to as intended — the ETF short offsets most of the name's decline. But if this specific bank gets hit with an idiosyncratic problem — a bad loan-loss disclosure, a regulator's public letter — the ETF may barely move while the stock falls 20 percent, and the hedge does almost nothing for the loss that actually mattered. That gap is the basis risk you accepted by using a proxy, and it doesn't disappear just because the hedge ratio was calculated carefully.

Keeping the correlation honest over time

A proxy hedge sized off a two-year correlation estimate is not a number you set once and trust indefinitely. Correlations between a single name and its proxy drift — a bank that used to trade closely with its sector can decorrelate as its business mix changes, a merger changes what a peer even represents, an index reconstitution changes what's inside the ETF you're using. Re-checking the correlation on a regular cadence, and re-sizing the hedge when it moves meaningfully, is the maintenance work that keeps a proxy hedge from quietly becoming stale while looking, on paper, exactly as it did the day it was put on.

Sizing conservatively on purpose

Because a proxy hedge is known in advance to be imperfect, there's a case for deliberately under-sizing it relative to the historical correlation rather than matching it exactly — accepting a slightly larger residual exposure on ordinary days in exchange for not overstating how protected you are. A hedge sized to the full historical beta can create false confidence: the position looks hedged on the risk report, and that comfort is exactly what leads a desk to hold the underlying larger than it otherwise would. A hedge deliberately sized a little light keeps the residual risk visible, which is often more useful than a number that looks tidier but hides how much is still riding on a correlation assumption.