It’s the long call/short put pair you already know, minus the expiration date.
On November 5, Kalshi started listing perpetual futures on individual stocks, and it says traders moved more than $300 million the first morning. By Monday there were 58 of them, from Apple and NVIDIA to McDonald’s, plus SPY and QQQ, alongside the US 500 index perp Kalshi launched in October.
If you trade options, you already know this instrument. There’s a decent chance you’ve used it yourself.
The synthetic long that never expires
Buy a call and sell a put at the same strike and expiration, and you have a synthetic long. It moves dollar for dollar with the stock without your owning a share. A perp is that pair with the expiration date removed.
That’s most of it. A perp has delta and nothing else: no strike, no theta, no vega, no premium. Each stock perp contract is 0.02 shares, so one contract of NVIDIA is about $3.60 of exposure, and you size in dollars instead of 100-share blocks.
The more useful comparison is with the call you’d actually buy. By put-call parity, a long call is that same synthetic long with a long put attached. The put is what caps your loss at the premium. A perp doesn’t come with one, and nearly everything below follows from that missing put.
The deposit, and the line underneath it
You don’t pay for a perp. You post a deposit and pick a leverage. At 4x, $250 controls $1,000 of stock, and the deposit comes back when you close, plus or minus the result.
Leverage is capped by name, and the caps are lower than offshore crypto apps have taught people to expect. As of Monday, Apple allowed 5.0x, NVIDIA 4.6x long and 4.1x short, and Tesla 4.3x long and 3.8x short. Sandisk has the lowest cap on the list, at 1.6x. The index perp goes to 14.6x.
The number that matters more is where Kalshi closes you out. A position is liquidated when the deposit falls to a maintenance level, well before it reaches zero. At maximum leverage that line sits about 5.5% below entry on Apple, 5.9% on NVIDIA, and 6.5% on Tesla.
At 2x it’s about 40% away on all three. NVIDIA fell about 4.5% on Monday alone, which gives you a feel for how much room 5.9% is.
This is where the options habit breaks. When a short put goes wrong, the worst case is assignment: you own shares you were willing to own, at a price you picked, and you can wait. When a perp goes wrong, there’s nothing to be assigned. The position is closed and the loss is realized. In a fast market, Kalshi’s disclosure adds, the close can fill well past the trigger and losses can exceed the deposit.
Funding is the carry, paid daily
A synthetic long isn’t free to hold. The cost of financing the stock, less its dividends, is built into the gap between the call and put prices on the day you open it. A perp collects the same carry a little at a time, through funding.
Once a day at 4:00 PM Eastern, Kalshi compares where the perp has been trading with the stock. If the perp ran above it, longs pay shorts. If it ran below, shorts pay longs. Dividends are handled separately, as cash credited to longs and debited from shorts.
Two details matter. Funding is paid by the crowded side, which isn’t always the wrong side. You can be right on direction and still pay every day because everyone else agrees with you. And it can change sign whenever the crowd does. Tesla’s first three prints were negative, positive, then negative again.
Kalshi caps funding at 2% of the position per day. Nothing has come close to that so far, with early prints mostly within about 0.05% either way. But funding is charged on the full position and felt on the deposit, so at 4x every 0.05% is 0.2% of what you put up.
Figure 1 shows what that looked like in the index perp’s first four weeks. The index fell about 3%, and longs paid funding on 15 of 18 trading days anyway. A 5x long on a $1,000 deposit paid $15.59 over that stretch, about 1.6% of the deposit, on top of the price loss. At 10x it was $31.

Watching one burn
To see how that adds up, I put a small long Bitcoin position on Kalshi’s demo exchange in October and left it alone. I wasn’t trading it. The point was to watch the costs land in real time, and the cleanest way to do that was not to touch it.
It was $22 at 2.6x. Bitcoin fell about 7% on the demo exchange while I held it, which cost about $4. Funding, which crypto perps pay three times a day, cost me $16.87, four times as much. On November 1 the position was liquidated with about 95 cents left. It did exactly what I wanted it to, which is the only time I’ve been pleased to lose 96% of anything.
Before anyone reads too much into the $16.87: the demo exchange runs its own order book, separate from the real one, so play money can’t move real prices. Its perp traded rich for two weeks straight, and the longs paid for it.
On the real exchange, over the same two weeks, the same position would have paid about 42 cents in funding. It was the same contract under the same rules, but it was being traded and held differently on the real-money production exchange vs. the practice demo exchange. What it costs to hold a perp depends as much on who else is in that book as on the asset itself.
How often the line gets touched
The perp asks for one decision, the leverage, so it helps to know what each choice means. Options traders already have the tool: probability of touch. Figure 2 estimates it the standard way, taking the options-implied chance of finishing below the liquidation price and doubling it.

At maximum leverage, NVIDIA and Tesla are about 40% and 36% to touch the line within a week, and better than even within a month. Apple is calmer at 15% and 27%. At 3x the one-week odds drop to low single digits, and at 2x they’re negligible. NVIDIA’s November 17 earnings fall inside the one-month window, which is part of why its numbers are elevated. These are one day’s market-implied odds, not forecasts, and they carry the same risk premium put sellers collect, so if anything they run high.
Closures add a wrinkle the figure doesn’t capture. Stock perps trade from Sunday evening to Friday afternoon, not around the clock, and can pause around corporate actions. The stock can move while the perp is shut, and the position reopens wherever the price is.
A few practical details
Fees are 0.12% of the full position for market orders and 0.02% for limit orders, each way. At 4x, a market-order round trip is close to 1% of the deposit. Perps sit in a separate account, through Kalshi Prime for most retail users, and you apply and take a tutorial first. The index perp tracks a MerQube 500-stock index, not the S&P 500.
Where I land
A perp is never a better option. It’s sometimes a better synthetic.
If you use options for what the put gives you, a capped loss and the freedom to sit through a drawdown, this isn’t that tool, and no setting on the ticket turns it into one. If you’ve been building synthetics, using deep in-the-money calls as stock replacement, or shorting through a broker, a perp does the same job in smaller pieces, with nothing to roll and no borrow to find. In exchange you get a daily bill and a hard line underneath. So which would you rather manage: a strike and an expiration date, or a leverage and a funding rate?
