# Why a Hedged Position Still Loses Money: Sizing, Currency, and Funding

URL: https://stackfi.io/market/why-hedged-positions-still-lose-money/
Collection: market
Published: 2026-09-12T00:00:00.000Z
Updated: 2026-09-12T00:00:00.000Z
Description: A delta-neutral trade can be perfectly hedged on paper and still bleed. Three failure modes — share-count sizing, an unhedged currency leg, and funding that inverts — with live data from tokenized equity markets.
Tags: tokenized equities, basis trade, funding rate, hedging, delta neutral, perpetuals
Sources: Hyperliquid info API (xyz HIP-3 perp DEX); Jupiter price API v3; Frankfurter FX reference rates; StackFi tokenized equity basis monitor

---

A hedged position is supposed to be boring. You own the thing, you are short the thing, the price cancels out, and you collect whatever spread or carry drew you in. When people describe these trades they use the word *arbitrage*, and the word does a lot of quiet work: it implies the outcome is known and the only question is size.

Then the position loses money, and the loss has nothing to do with the direction of the asset.

This happens often enough, and for a small enough set of reasons, that the reasons are worth naming. All three below are visible in a trade a lot of people put on during mid-2026 — holding SK Hynix shares in Seoul against a dollar-denominated short somewhere else — and all three transfer directly to tokenized US equities held against perpetual futures, which is the same structure with different plumbing.

## Failure one: sizing by units instead of by value

This is the subtle one, and it is the failure that looks most like competence.

Suppose a company's local share trades at $100, and a depositary receipt or tokenized wrapper represents ten of those shares. The wrapper "should" be worth $1,000. Instead it trades at $1,350, a 35% premium — which is the whole reason you are interested.

The intuitive hedge is one wrapper against ten shares. The units line up. The ratio is the stated ratio. It is also wrong.

| | Short 1 wrapper | Long 10 shares | Net |
|---|---|---|---|
| At entry | $1,350 | $1,000 | **short $350 of exposure** |
| Share rises 10%, premium unchanged | $1,485, loses $135 | $1,100, gains $100 | **−$35** |

You have hedged the *units* and left yourself short $350 of the underlying. The position now has a directional opinion you never intended to express, and it loses money on a rally even when the spread you were trading does not move at all.

Sizing by value fixes it. Short $1,350 of the wrapper against $1,350 of shares — 13.5 shares, not 10. Run the same rally through it and the legs cancel exactly, leaving you exposed to the premium and nothing else.

The general rule: **hedge equal value on each side, never equal units.** Any time a conversion ratio and a price premium both exist, the two are not the same number, and using the ratio as your hedge silently bundles a directional bet into what you believed was a spread trade.

## Failure two: the currency leg nobody hedged

If your short is denominated in dollars and your long is a share listed in another currency, you are short that currency whether you meant to be or not. The dollar price of a foreign share already contains the exchange rate, so a dollar-denominated hedge covers the share *and* the currency, while your long leg covers only the share.

This is not a rounding error. Between 15 July and 11 September 2026 the Korean won went from 1,492 to 1,343 to the dollar. The won gained almost exactly 10% in under two months. Anyone holding Seoul-listed shares against a dollar short, without neutralising the currency, gave back ten percent of position value to a variable that had nothing to do with their thesis.

The fix is mechanical rather than clever: end up with no net position in the foreign currency. If you borrowed local currency to buy local shares, that borrowing has to be extinguished — typically by converting the debt into the currency your hedge settles in. What matters is the end state. **No residual balance and no residual loan in the foreign currency**, because either one is an open FX position sitting underneath a trade you are describing to yourself as market-neutral.

Tokenized equities remove this particular trap, since both legs are already dollar-denominated. They replace it with a different one.

## Failure three: treating funding as a yield

Hold a tokenized share against a short perpetual and the share price cancels. What is left is the funding rate — a payment that flows between longs and shorts, hourly on most venues, and changes sign whenever positioning does.

People quote it annualised, which makes it sound like a deposit rate. It does not behave like one.

StackFi's [basis monitor](/tools/tokenized-equity-basis/) tracks 11 tokenized US equities against their perpetuals. Over the first full week sampled:

- **Every one of the 11 markets paid the short at some point.** Not one held a continuously positive carry.
- Only four stayed positive for more than nine hours in ten.
- One market printed **5.48% annualised** at the moment of reading while having been **negative for 69 of the previous 168 hours**, with a median of 3.84%.
- The widest weekly swing ran from roughly **−312% to +240%** annualised.

That last pair of numbers is the point. A rate quoted per hour and expressed per year turns small hourly moves into enormous annual figures, in both directions. Entering a carry position because the current rate looks attractive is a decision made on the least informative number available. The median, the range and the count of negative hours describe what a position would actually have experienced.

## Failure four: entering once, at full size

This one needs no mechanism, only honesty. Spreads that look extreme are extreme because something is sustaining them, and the thing sustaining them rarely stops on the day you arrive.

A premium that has reached 35% can reach 43% and stay there for a fortnight. If your entire position went on at 35%, the intervening two weeks are spent managing your own nerves rather than the trade, and the temptation to close at the worst moment is strongest precisely when the spread is widest. Entering in tranches costs a little of the edge and buys the ability to still be in the position when it converges.

The related error is entering on someone else's narrative. Convergence trades opened on the strength of a rumour that some structural barrier is about to be removed have a way of outlasting the rumour.

## What this means for tokenized equities specifically

The structure that used to require a prime broker in one country and an exchange account in another is now reachable from a single self-custodial wallet: the tokenized share settles on one chain, the perpetual on another venue, and both are dollar-denominated. That removes the currency trap and most of the operational complexity.

It does not remove the other three. Sizing by value still matters, because tokenized shares do not always represent exactly one share — dividends are reinvested by raising a multiplier, so a token quietly comes to represent 1.004 shares and then 1.008. Funding still inverts. Concentration still hurts.

And it adds one the older version did not have: **depth**. The spot side of these markets is thin. The widest entry edge in our data sat on a pool holding under $400,000, which an order of any size moves against you before the second leg is even open. You can check the current state on the [premium tracker](/tools/tokenized-stock-premium/) and the [basis monitor](/tools/tokenized-equity-basis/).

None of this is a reason not to run the trade. It is a list of the ways a position that is genuinely hedged against price still finds a way to lose, which is worth knowing before rather than after.

## Frequently Asked Questions

### Should you size an arbitrage by share count or position value?

By position value, always. Matching share counts hedges the conversion ratio rather than the exposure, which leaves you net short by the size of the premium. With a wrapper trading at a 35% premium to ten underlying shares, a one-for-ten hedge leaves roughly a quarter of the position unhedged and directional — it loses money when the underlying rallies even if the premium never moves. Short equal dollar value on each leg and the price risk cancels, leaving only the spread you intended to trade.

### Why did my delta-neutral position lose money on currency?

Because a dollar-denominated hedge against a foreign-listed share covers both the share price and the exchange rate, while the share itself covers only the share price. The difference is an open currency position. Between mid-July and mid-September 2026 the won gained about 10% against the dollar, which is 10% of position value for anyone who left that leg open. The fix is to finish with no net balance and no net borrowing in the foreign currency.

### What goes wrong with cash and carry on equities?

Four things, roughly in order of how often they occur. Funding or the carry inverts and the position pays instead of collects. The hedge is sized by units rather than value, leaving hidden directional exposure. A currency leg is left open when the two sides settle in different currencies. And the whole position is entered at once at a level that then gets worse. Execution faults — a stale price feed, a failed leg — belong on the list too, and have cost professional desks more than adverse prices ever have.

### How often does perpetual funding actually turn negative?

More often than the annualised headline suggests. Across 11 tokenized US equity perpetuals over one week, every market spent some hours negative, and only four stayed positive for more than 90% of hours. One market was negative for 69 of 168 hours while still displaying a positive current rate. Check the distribution and the count of negative hours rather than the reading at the moment you happen to look.

### Does one tokenized share always equal one share?

No, and this matters for sizing. Issuers of tokenized equities generally reinvest dividends by raising a multiplier rather than paying cash, so each token gradually comes to represent slightly more than one share. A token at a multiplier of 1.004 represents 1.004 shares, and a hedge sized as though it were exactly one will be marginally wrong in a way that grows with every dividend.