I keep running into people who think shorting is one thing. You have a bearish view, you press the sell button, you win if the price drops. That is true in the same way that driving is one thing whether you are in a parking lot or on a mountain pass. The direction of the wheel is the same. Everything that actually decides whether you keep your money is different. The same short view on the same underlying costs you a different amount, behaves differently when it goes against you, and gets squeezed for completely different reasons depending on whether you put it on in a stock, a future, or a crypto perpetual.
So I want to walk through the mechanics of getting short in each of those three, and end with a way to price the true holding cost so you can compare them on the same footing.
Stocks: you are borrowing something real
Shorting a stock is the only one of the three where you are physically borrowing an asset and owing it back. You want to be short 100 shares, so your broker has to find 100 shares somewhere, lend them to you, and you sell them into the market. At some point you buy them back and return them. The whole thing hinges on the borrow being available, which is why this is the market with the most friction and the most ways to get hurt that have nothing to do with your thesis.
A few things fall out of that borrow. First, the locate. Before you can short, someone has to confirm the shares exist to be lent. For big liquid names this is invisible and free. For small floats, heavily shorted names, or anything the whole world already wants to short, the locate is hard and the borrow gets expensive. That expense is the borrow fee, quoted as an annualized rate, and it can range from a rounding error on a mega-cap to something genuinely painful on a hard-to-borrow name, sometimes tens of percent annualized. You pay it every day you hold, and it can change on you daily because it floats with supply and demand for the borrow.
Second, you are on the hook for dividends. If the company pays a dividend while you are short, you owe it to the person you borrowed from. That is not optional and it is easy to forget when you are modeling the trade.
Third, and this is the one that ends accounts, your borrow can be recalled. The lender can want their shares back, and if your broker cannot find a replacement borrow, you get bought in. You are forced to close at whatever the price is, at a time you did not choose. This is a big part of why short squeezes in single stocks get so violent. It is not only that price rises and margin tightens. It is that the mechanical scarcity of the borrow forces covering, which drives price up, which forces more covering. The instrument itself has a built-in feedback loop.
Futures: the symmetric short
Futures are the clean one. There is no asset to borrow because a future is a contract, an agreement to exchange something at a price on a date. For every long there is a short, by construction. Going short a future is mechanically identical to going long, just with the sign flipped. You post margin, you take the position, and there is no locate, no borrow fee, no recall, no dividend obligation in the stock sense.
The cost structure lives somewhere else. It is baked into the price of the future relative to spot. When a future trades above spot, that spread is roughly the cost of carry, the financing and storage and convenience of holding the thing versus holding the contract. As a short, that curve can work for you or against you, and you have to roll the position before expiry, which means paying the bid-ask and eating whatever the spread between the expiring contract and the next one is. That roll is your real recurring cost, and on a steeply shaped curve it is not small.
Squeezes still happen in futures, but the mechanism is different. There is no borrow to recall, so a futures squeeze is about margin and delivery. As price runs against shorts, margin calls force liquidations, and near expiry, shorts who cannot or will not deliver the physical underlying have to close at any price. The pain is real but it comes from the delivery and margin plumbing, not from a scarce pool of lendable shares.
Perps: you pay rent for the view, or you collect it
Crypto perpetuals took the symmetric structure of a future and deleted the expiry. That solves the roll problem and creates a new one. With no expiry to anchor the contract price to spot, exchanges use a funding rate to keep the perp tracking the underlying. Funding is a small payment exchanged directly between longs and shorts, typically every few hours. When the perp trades above spot, longs pay shorts. When it trades below, shorts pay longs.
What that means for a short is genuinely different from the other two. In a normal market where everyone is bullish and the perp trades rich to spot, you get paid to be short. Funding flows into your position while you hold it. That is the closest thing to a free lunch in any of these markets, and it is why some people run short perp positions mostly to harvest funding rather than to express a view. But it flips. In a hard sell-off where everyone piles short, funding goes negative and now you are the one paying, sometimes at rates that annualize into the hundreds of percent for short stretches. Your holding cost is not fixed and it is not knowable in advance. It is a live auction that reprices every few hours.
Squeezes on perps are their own animal. There is no borrow and no delivery, so a perp squeeze is pure liquidation cascade. Leverage is high, liquidation engines are automatic, and when price moves against a crowded side, the exchange force-closes those positions, which pushes price further, which liquidates the next tier. It can happen in minutes. The tell is usually visible beforehand in funding and open interest, which is the kind of thing worth watching if you run size. I built a lot of Blockcircle around surfacing exactly that, funding, open interest, and positioning across venues, because the setup is legible if you are looking at the right numbers.
Pricing the same view three ways
Here is how I actually compare them. Take one bearish view and one horizon, say you want to be short for roughly a month, and compute the annualized holding cost in each instrument before you decide where to put it on.
- Stock. Borrow fee (annualized, and check whether it is hard-to-borrow) plus any dividends you will owe during the hold, plus your best guess at recall risk. If the name is a crowded short, add a fat premium for the chance you get bought in at the worst moment.
- Future. Estimate the roll cost from the shape of the curve, annualize it, and add the bid-ask you will pay on each roll. If the curve is steep against you, this can quietly be the most expensive of the three.
- Perp. Look at the recent funding rate and annualize it, but treat it as a range, not a number. If funding is positive you may get paid, so the cost can be negative. Then stress it. Ask what happens to your cost if the trade works and the whole market piles into your side and funding flips hard against you.
Once you have those three annualized numbers side by side, the choice of venue stops being a habit and becomes a decision. Sometimes the cheapest place to be bearish is the perp because funding pays you. Sometimes it is the future because there is no borrow to worry about. Sometimes the stock is the only place you can express the specific name, and the borrow fee is just the price of admission. The mistake is assuming the instrument you always use is the cheapest one for this particular short, because more often than I would like, it is not.