Perpetual Futures: At 10x on Bitcoin, More Than a Third of Correct Calls Get Liquidated First

BitMEX invented the perpetual swap in 2016, and its own documentation is unusually direct about how the running cost works. “Your Mark Value is irrespective of leverage. For example, if you hold 100 XBTUSD contracts, funding is charged/received on the notional value of those contracts, and is not based on how much margin you have assigned to the position.”¹ Funding is quoted against the position; it is paid out of your money. Leverage is the ratio between those two things.

Run the arithmetic. The classic resting funding rate on a BitMEX-style venue is 0.01% per 8 hours. There are 1,095 eight-hour periods in a year, so that is 10.95% a year of notional. At 10x leverage, the notional is ten times the cash you posted, so the same rate costs 109.5% a year of your own capital. At 20x it is 219%. Exchange explainers quote the first number. The second one is the one that leaves your account.

This matters more than it used to, because perpetual futures stopped being an offshore product. Coinbase brought the first CFTC-regulated version to US traders on July 21, 2025, carefully labelled perpetual-style: the nano BTC-PERP and ETH-PERP contracts carry no monthly expiry and up to 10x leverage, but they do expire, in five years.² On May 29, 2026 the CFTC approved KalshiEX’s BTCPERP as a genuinely indefinite perpetual futures contract and issued a policy statement on listing perpetual contracts generally.³ CME sued the Commission on June 18, 2026 to void both. That case is still pending.⁴ Perpetuals account for about 90% of global crypto derivatives volume by Coinbase’s own reckoning, and a retail investor with a US brokerage login can now reach them.

Perpetual futures solve a real problem for hedgers and market makers, and they are a poor instrument for a DIY investor building long-term wealth. The case for that rests on arithmetic rather than disapproval, and the arithmetic is the part almost nobody publishes.

Where perpetual futures came from

The idea is older than crypto. In 1993 Robert Shiller proposed a contract with no delivery date, cash-settled daily against both the change in an asset’s price and an index of its income, so that markets could exist for assets too illiquid to trade directly, such as single-family housing or an individual’s labor income.⁵ The design problem he was solving was the absence of a spot market, not the presence of leverage.

BitMEX commercialized the structure in 2016 with XBTUSD, an inverse perpetual swap offering up to 100x leverage.¹ The product grew into the dominant form of crypto derivative because it removes the operational nuisance of rolling: a dated future has to be closed and reopened every quarter, at a spread, forever, if the exposure you want is indefinite. Kalshi made exactly this argument to the CFTC, calling the perpetual structure “functionally superior to dated alternatives for participants whose economic exposure [to bitcoin] is itself perpetual or indefinite in duration.”³

That argument is sound as far as it goes, and it is worth being precise about what it covers. Perpetuals remove a recurring transaction cost and fragmented liquidity across expiries. In exchange, the carrying cost that a dated contract embeds in its basis becomes a floating rate charged at intervals for as long as you hold.

Funding is the price of never having to roll

A dated future converges to spot because at expiry it must: somebody delivers, or a cash payment is made against the settlement price. That terminal event is what disciplines the price along the way. Remove the expiry and the discipline has to come from somewhere else.

Funding is that mechanism. At fixed intervals, whichever side of the market is trading away from spot pays the other side. The CFTC described the effect in its approval order: the payments impose “a periodic cost on whichever side is ‘trading away’ from the spot price, creating a continuous economic incentive for the price of the contract to converge towards the spot price.”³ When the contract trades above spot, longs pay shorts. When it trades below, shorts pay longs.

The important property for a long-term holder is the direction this usually runs. Retail demand in crypto is persistently long and persistently levered, which pushes the contract above spot and keeps funding positive. Holding a long perpetual through a flat market is a slow transfer of your equity to the people on the other side.

The same funding number means different things at three different venues

“Funding is 0.01%” is not a portable statement. Three venues a US reader can actually encounter implement it three different ways, and the differences change the resting cost of holding.

BitMEX. The funding rate combines an interest component and a premium index, with a dampener:

BitMEX spells out the consequence itself: “if (I - P) is within +/-0.05% then F = P + (I - P) = I. In other words, the Funding Rate will equal the Interest Rate.”¹ With the interest component at its conventional 0.01% per 8 hours, the rate returns to exactly +0.01% for any premium index between −0.04% and +0.06%. Whenever the premium sits inside that band, which covers ordinary market conditions, the design has longs paying roughly 11% a year of notional without anyone choosing that number. Funding occurs at 04:00, 12:00 and 20:00 UTC, and the rate itself is capped at 75% of the gap between initial and maintenance margin. What happens when the premium sits below the band is covered further down, and it is not what most readers of a funding-rate screen assume.

Kalshi. BTCPERP references the CF Benchmarks Bitcoin Real Time Index, trades in units of one ten-thousandth of a bitcoin, and funds three times a day at 12:00 AM, 8:00 AM and 4:00 PM Eastern, clearing through Kalshi Klear.⁶⁷ The rule is a time-weighted average of the 480 one-minute premium candles in each funding period, with no interest component, a clamp at ±2% per interval, and an explicit zero threshold: if the absolute rate comes in below 0.01%, it is set to zero and no payment is made.¹⁹ The same 0.01% number that anchors the rate from below on BitMEX switches it off entirely here.

Coinbase. Perpetual-style futures compute funding hourly, as a time-weighted average of the futures-versus-spot premium sampled over twenty three-minute periods and scaled by a factor of 24, then smoothed against the previous hour at 75% / 25% weights. There is no interest component. Funding accrues hourly but is processed twice a day, once during the midday margin cycle and once at end of day, as “cash adjustments through the clearinghouse” (Nodal Clear), recorded as a separate entry that does “not affect variation margin.”⁸

The difference shows up immediately in the data. Boon Chuan Lim collected the complete public funding history of Kalshi’s perpetuals from launch through July 11, 2026, 1,226 eight-hour observations across 13 contracts, and found the funding rate was exactly zero in 73.0% of periods. When it did activate, the mean absolute rate was 1.88 basis points, barely outside the band. Kalshi’s contract sat within one basis point of its reference index in 64.6% of bitcoin windows.¹⁹

The deadband by itself does not explain that. Lim applies Kalshi’s rule counterfactually to Bybit’s own premium index over the identical window and gets a zero share of 0.0%, because Bybit’s bitcoin perpetual traded at a persistent discount to its index, mean premium −4.7 basis points and never once positive across five weeks. Same asset, same weeks, two venues sitting in different places relative to their own reference prices.

That paper also corrects a reading that is easy to get backwards, and it applies to the BitMEX column above. Bybit’s realized funding over the window averaged only 0.38 basis points in absolute value, which looks like a tightly anchored market. It is arithmetic: the clamped interest term contributes up to +5 basis points against a premium near −4.7, and the two nearly cancel. Lim validates this by reconstructing Bybit’s realized funding from its public premium series to a mean absolute error of 0.10 basis points. So the “longs pay about 11% a year” resting state of a BitMEX-style formula is what you get when the premium sits inside the dampener band. In a sustained drawdown, where the premium sits well below it, realized funding can be near zero while the basis is wide open. Read the premium and the interest term separately.

The clearinghouse row deserves a moment. On a US designated contract market, funding is a cleared cash adjustment on a published schedule, booked separately from variation margin, with a central counterparty standing between the two sides. Offshore, funding hits your margin balance directly at every interval with the exchange as counterparty. That is a meaningful difference in credit risk and in operational transparency, and it is one of the few respects in which the onshore versions are plainly better.

The bottom row is worth noticing too. Coinbase’s contracts are marketed on having no monthly expiry, and its own launch post says they are “long-dated with expiration dates of 5 years.”² For a trader that difference is immaterial. For anyone comparing products, it is the reason Coinbase says perpetual-style and Kalshi says perpetual.

Write all of this down with dates attached, because the specifications move. BitMEX, the venue that invented the instrument, announced on July 23, 2026 that it will shut down entirely at 04:00 UTC on September 23, 2026, with accounts restricted to reduce-only from August 26.⁹

Who receives the funding, and what they are being paid for

Funding is not a fee the venue collects. BitMEX states plainly that it “does not charge any fees on funding; it is exchanged directly peer-to-peer.”¹ Somebody is receiving what you pay, and it is worth knowing what they are being compensated for.

The Bank for International Settlements studied crypto carry over April 2019 to July 2024 and found it averaged about 7% a year.¹⁰ Their explanation is a supply-and-demand story: levered retail demand for long exposure exceeds the arbitrage capital available to take the other side, and the imbalance shows up as a persistent premium. The study also found that high carry predicts liquidations of short futures positions, with a one-standard-deviation increase in carry associated with short liquidations of about 22% of open interest the following month. Two scope caveats matter: the paper studies fixed-date futures rather than perpetuals, and the mechanism it identifies describes which side of the market is crowded rather than a signal you can trade.

The most useful finding for a DIY investor is what happened when the supply constraint eased. When US spot bitcoin ETFs launched, carry compressed by roughly 3 percentage points across exchanges and about 5 percentage points on CME.¹⁰ A large share of what perpetual longs had been paying was compensation for the difficulty of getting spot exposure at all. That difficulty is gone. A retail investor who wants bitcoin exposure can buy a spot ETF in a brokerage account for a few basis points a year, which is the cheap version of the trade that funding used to price.

Leverage sets the distance to your liquidation price

Margin is a performance bond rather than a purchase price. Post $M$ against notional $N$ and your leverage is $L = N/M$. Your equity absorbs the full notional move, so a 1% move against you costs $L$ percent of your capital.

Liquidation happens when equity falls to the maintenance requirement $m$, expressed as a fraction of notional. The adverse move that gets you there is

At 10x with a 0.5% maintenance requirement, that is 9.5%. Bitcoin moves 9.5% in a quiet week. The costs term is the part traders forget: every dollar of funding paid and every basis point of fees comes out of the same equity buffer, so the liquidation price creeps toward you the longer you hold. The calculator below holds the barrier fixed, which makes its estimates optimistic.

Being right about direction and surviving the path are separate bets

Every perpetual futures calculator on the web reports where the liquidation price sits. None of them reports how likely you are to touch it at some point before your horizon.

Those are different questions with different answers, and the second one is computable in closed form. Model the price as geometric Brownian motion, set the barrier at your liquidation level, and first-passage probabilities give you both the chance of being stopped out and the chance of being stopped out on a path that would have finished in your favor. The formulas are in the appendix.

Bitcoin’s realized volatility over the 365 days through July 31, 2026 was 43.6%, computed from daily log returns and annualized by the square root of 365.²⁰ Using that, a 0.5% maintenance requirement, and zero drift, here is what leverage and patience do to a long position.

Read the highlighted row carefully. A trader at 10x who holds for a quarter and turns out to be right about the direction still gets liquidated before the payoff arrives on more than a third of those paths. The call was correct. The position was closed anyway.

This is the structural difference between holding an asset and holding a levered claim on it. A spot holder who is early is simply early; being down 30% is unpleasant and reversible. A perpetual holder who is early is gone, and the position does not reopen when the thesis comes good. Compare the 5x and 20x rows: quadrupling leverage takes the liquidation probability from 9% to 73%, because the barrier moves from 19.5% away to 4.5% away while the volatility stays the same.

Two honest caveats. Zero drift is an assumption, and a trader with a genuine directional edge faces better numbers, which is why the calculator lets you set drift. And geometric Brownian motion understates the risk for bitcoin: real prices gap, jump, and produce the clustered cascades that empty accounts fastest. The true probabilities are worse than the table shows, not better.

Run your own position through it

The lab below computes all three pieces: where the liquidation price sits, what funding drains from your equity while you wait, and how likely the path is to reach the barrier first. Its defaults reproduce the highlighted row above.

What the evidence shows about retail outcomes

The evidence here is thinner than either side of this debate tends to admit, so each finding below carries its limits with it.

The margin-call threshold is low. A study of cryptocurrency perpetual leverage found that keeping the probability of a margin call under 1% on a given day required staying at roughly 3x for long positions and 5x for short positions.¹¹ The paper is single-authored and its sample is eight consecutive days, so treat it as an order-of-magnitude result rather than a precise threshold. It points the same direction as the table above: the leverage that survives routine volatility is far below what venues offer.

Retail derivatives outcomes are mediocre rather than catastrophic. The CFTC studied 36,538 retail accounts trading CME micro equity-index futures between February 2021 and November 2022 and found a median loss of roughly $100 to $200, with the 60th percentile account about breaking even.¹² That study covers equity-index futures rather than crypto, uses end-of-day positions, and estimates rather than observes profit and loss, so it transfers to perpetuals only loosely. Its most transferable finding is behavioral: accounts in the bottom quintile of initial performance “continue to trade as frequently as those in the top quintile.” Losing does not teach people to stop.

You will not see the cascade coming. Work on crypto liquidation cascades finds they behave as discontinuous transitions driven by shocks rather than as critical phenomena with reliable early warning signals.¹³ The practical consequence is that “I will cut it before it gets bad” is not a plan. On October 10 and 11, 2025, CoinGlass reported more than $19 billion of forced closures across roughly 1.6 million accounts in a matter of hours after a macro headline.¹⁴ Both figures come from exchange-reported liquidation feeds, which are throttled under load and count notional rather than trader losses, so read them as a reported lower bound on the scale rather than as a measure of money lost.

One widely repeated statistic is deliberately absent above: the claim that 70% to 90% of retail derivatives traders lose money. It has no traceable primary source, and an argument for care about numbers cannot lean on a number nobody can find.

The same exposure through four instruments

Most comparisons of perpetuals cover spot, dated futures and perpetuals, and stop there. Adding a long call changes the picture, because it is the instrument that gives levered upside without a path-dependent wipeout.

The convexity row is where a long call earns its keep. A perpetual is a linear claim: its second derivative with respect to price is zero, so volatility does not enter its value at all. A long call has positive gamma and positive vega, which means an increase in volatility raises what it is worth. A perpetual holder in the same conditions is closer to liquidation. Both instruments cost money to hold, and the option’s cost is bounded and known when you buy it, while the perpetual’s is floating and unbounded. For a directional bet with a defined horizon, the option is usually the better-behaved wrapper. Black-Scholes and option pricing covers how that premium is set.

On tax: US regulated futures contracts can qualify for Section 1256’s 60/40 treatment with year-end mark-to-market, but the application to perpetual contracts has not been definitively addressed, and offshore perpetuals do not qualify at all. Treat this as a question for a tax professional rather than a settled advantage.

When a perpetual is the right instrument

There are legitimate users, and it is worth naming them, because “retail should mostly avoid this” and “this is a bad product” are different claims and only the first one is defended here.

• Market makers and basis traders. Funding is the revenue line, not the cost line. Somebody has to take the short side of persistent retail long demand, and being paid roughly 7% a year to do it against a hedged spot position is a real business.
• Miners and treasury holders hedging existing exposure. If you already hold the asset and want to reduce exposure without selling, a short perpetual does that continuously and without a roll schedule.
• Anyone whose exposure genuinely is indefinite. This was Kalshi’s argument to the CFTC, and it holds. Rolling a quarterly contract forever costs real money in spreads.
• Capital-efficient hedging inside a larger book. Posting 5% margin instead of 100% notional frees capital for other uses when the rest of the book is managed against that exposure.

The strongest argument on the other side of the retail question came from a CoinDesk opinion piece in July 2026: systemic risk in perpetual markets is a property of venue design, such as leverage caps, index construction and auto-deleveraging rules, rather than of the contract itself.¹⁵ That is correct about systemic risk, and it is compatible with everything above. Better venue design reduces the chance that one trader’s liquidation becomes everyone’s problem. It does not change the arithmetic of paying 109.5% a year on your own capital, and it does not move a barrier that sits 9.5% away. Those are properties of leverage and funding, and they survive any amount of clearing infrastructure.

It is also worth noting how narrowly the CFTC drew its own approval. The order applies to bitcoin and to similarly structured contracts on digital commodities with “deep, active, and continuous spot market trading,” and explicitly does not extend to other asset classes; a footnote observes that perpetual contracts are “likely particularly ill-suited for agricultural products.”³ Kalshi has since been reported to be in talks to extend perpetuals to metals, foreign exchange and energy.¹⁶ Watch whether the convergence argument that justified the bitcoin approval survives the move to assets whose spot markets close at night.

Five questions before you open one

1. What is the funding cost as a percentage of my margin, not my notional? Multiply the annualized notional rate by your leverage. If that number is above your expected return, the position is a losing bet before the price moves.
2. What is my probability of liquidation over my horizon? The chance of touching the barrier at any point, over the period you intend to hold. Anything above 20% means the leverage is choosing the outcome instead of your thesis.
3. Would a long call express the same view? If yes, the option caps your loss at the premium and pays you for volatility instead of punishing you for it.
4. Can I state the venue’s funding formula, index and liquidation rule? If not, you do not know what you are paying or what closes you out. The three venues compared above answer that question three different ways.
5. If this position went to zero tomorrow, what happens to my plan? If the answer is anything other than “nothing,” the position is too large.

Technical appendix: how perpetual futures are priced

A dated future is anchored by cost of carry and terminal convergence:

A perpetual has no $T$, so the second equation is unavailable and funding supplies the anchor instead. Model it as a continuous flow proportional to the gap between the contract price $P_t$ and spot, at intensity $\kappa$. He, Manela, Ross and von Wachter’s Proposition 1 gives the resulting price:

with $\kappa = 1095$ for 8-hour funding. The familiar approximation $S(1 + r/\kappa)$ is the small-$r$ case of this and drops $r'$ entirely. Empirically they find mean absolute deviations from the no-arbitrage level of 60% to 90% a year across cryptocurrencies, declining by roughly 11% a year as arbitrage capital arrived, and a random-maturity arbitrage strategy earning a Sharpe ratio of about 1.8 at retail trading costs against 3.5 for fee-free market makers.¹⁷ Funding arbitrage works, and retail costs take roughly half of it.

Ackerer, Hugonnier and Jermann frame the same object differently: the perpetual price is a risk-neutral expectation of spot at a random stopping time whose distribution is set by funding intensity. Stronger anchoring shortens the effective horizon toward spot.¹⁸ Their results also depend on the payoff convention, so linear, inverse and quanto contracts are genuinely different instruments and the settlement currency changes the answer.

Why volatility prices an option but not a perpetual. For a linear claim $V = a + bS$, the second derivative $\partial^2 V/\partial S^2$ is zero, so the gamma term in the Black-Scholes equation vanishes and volatility drops out of the valuation. A long call has $\Gamma > 0$ and $\nu > 0$. This is the formal version of the convexity row in the comparison table.

Funding is a cash transfer, and theta is a valuation change. Cumulative funding $\sum_j N_j f_j$ leaves your account and arrives in a counterparty’s. Theta is the change in value of a finite-maturity nonlinear claim as time passes. Both make waiting expensive, through different mechanisms, and only one of them can be recovered if volatility rises.

First-passage probabilities. This is what the calculator runs. With $X_t = \ln(S_t/S_0)$, drift $\nu = \mu - \sigma^2/2$, and a liquidation barrier $b = \ln(S_{\text{liq}}/S_0) < 0$ for a long position:

and the joint probability of finishing above the entry price while having been stopped out along the way:

Dividing the second by $P(X_T \ge 0) = \Phi(\nu\sqrt{T}/\sigma)$ gives the share of correct calls that were liquidated first. The short side uses the mirrored maximum-based results. Two assumptions make these numbers optimistic: geometric Brownian motion has thinner tails than bitcoin, and the barrier is held fixed rather than being walked toward you by accumulated funding.

Exceptions. Inverse, quanto and power perpetuals and everlasting options are not linear claims, so a generic “perpetual futures pricing formula” is incomplete without specifying the payoff, funding rule, index, collateral, settlement currency and liquidation rule. BitMEX alone lists all three payout types.¹

Frequently asked questions

Are perpetual futures legal in the US?

Yes, on regulated venues. Coinbase listed CFTC-regulated perpetual-style futures on July 21, 2025, and the CFTC approved KalshiEX’s BTCPERP as a perpetual futures contract on May 29, 2026, alongside a policy statement on listing perpetual contracts. CME sued to void both on June 18, 2026, and that case is pending before Judge Colleen Kollar-Kotelly in the District of Columbia. Offshore venues offering 100x leverage to US persons are a separate matter and generally are not permitted to serve them.

How much does a funding rate cost per year?

Multiply the per-interval rate by the number of intervals in a year: 1,095 for 8-hour funding, 8,760 for hourly. A rate of 0.01% per 8 hours is 10.95% a year of notional. Then multiply by your leverage to get the cost against your own capital, which at 10x is 109.5% a year. Funding floats and can be negative, so this is a run rate at the current level rather than a fixed charge.

What happens if I am right about the direction but get liquidated?

You lose the position and the margin behind it, and you do not participate in the move you correctly predicted. There is no reinstatement. At 10x on bitcoin over 90 days at 43.6% volatility, this happens on roughly 37% of the paths that finish above the entry price.

Do perpetual futures ever expire?

A true perpetual has no expiry, which is the defining feature. Read the contract specification rather than the marketing, though: Coinbase calls its US contracts perpetual-style because they are long-dated with five-year expirations, while Kalshi’s BTCPERP is indefinite. Either way, a position can still end two other ways: liquidation, and the venue closing. BitMEX invented the instrument in 2016 and is shutting down on September 23, 2026, force closing any open positions.

Why do longs usually pay shorts?

Because levered long demand in crypto persistently exceeds the arbitrage capital willing to take the other side, which pushes the contract above spot and makes the premium component positive. On BitMEX-style venues the interest component adds a positive resting level on top of that. The BIS measured the resulting carry at about 7% a year from 2019 to 2024, and found it compressed by 3 to 5 percentage points once US spot bitcoin ETFs made the underlying easy to buy. It is not universal: over five weeks in mid-2026, Bybit’s bitcoin perpetual traded at a persistent discount to its index, which is the configuration in which shorts pay longs.

Is a perpetual future cheaper than an option?

It requires less cash upfront and it has a different risk profile. The option premium is bounded and known at purchase, and the maximum loss is that premium. The perpetual’s funding cost floats, has no upper bound over a long hold, and comes with a liquidation barrier that the option does not have. Comparing them on upfront cost alone leaves out the two things that decide the outcome.

Should a long-term investor use perpetual futures?

For accumulating an asset over years, no. Spot or a spot ETF gives the same directional exposure with no funding, no barrier and no path dependence, and the ETF launch itself removed much of what funding was historically compensating. Perpetuals earn their place for hedging existing exposure, for market making, and for defined-horizon positions sized so that a total loss changes nothing about the plan.

What leverage is actually survivable?

Lower than venues offer. The published research suggests roughly 3x long and 5x short to keep daily margin-call probability under 1%, and the table above shows 5x over 30 days carrying about a 9% liquidation probability against 73% at 20x. Venues advertising 100x are selling access, not a recommendation.

Key takeaways

• Convert funding to a percentage of your margin. The quoted rate is against notional. At 10x, 0.01% per 8 hours is 109.5% a year of the money you actually put up.
• Read the venue’s formula before quoting its rate. BitMEX’s design makes 0.01% a floor, Kalshi makes it a switch that turns funding off, and Coinbase computes hourly and clears the payment twice a day. In its first five weeks Kalshi’s funding was exactly zero 73% of the time.
• A small realized funding rate does not mean a tight basis. On a BitMEX-style formula a positive interest term and a negative premium can cancel, leaving near-zero funding while the contract trades well below its index. Look at the premium and the interest term separately.
• Liquidation probability is the number that matters. Where the liquidation price sits tells you almost nothing. Over 90 days at 10x, more than a third of correct directional calls are closed out before they pay.
• Leverage compresses the barrier faster than it grows the position. Going from 5x to 20x quadruples notional and takes liquidation probability from 9% to 73% over 30 days.
• A long call expresses a levered directional view with a floor. Bounded loss, positive gamma and vega, and no liquidation. Most retail perpetual positions are options trades wearing the wrong wrapper.
• Spot exposure got cheap, which is what funding used to price. Carry compressed by 3 to 5 percentage points when US spot bitcoin ETFs launched. DIY investors now have access to the cheap version of the trade.
• Regulatory approval is narrow and contested. The CFTC limited its reasoning to digital commodities with continuous spot markets, and CME’s challenge is unresolved.

Related guides

• Personal Leverage for margin mechanics and maintenance requirements on an ordinary brokerage account.
• Options Risk and Reward on the payoff shapes a long call gives you that a linear contract cannot.
• The Black-Scholes Equation for where the option premium comes from and why volatility enters it.
• Futures Yield Explained for roll yield and basis on dated contracts, the cost perpetuals replace.
• Should You Invest in Cryptocurrency? for the unlevered version of the allocation question.
• Compensated vs. Uncompensated Risk for whether a risk is one you get paid to bear.
• Position Sizing on how large a speculative position should be when you are right.

Sources and method

1. BitMEX, Perpetual Contracts Guide. Source of the funding formula F = P + clamp(I - P, 0.05%, -0.05%), the funding times (04:00, 12:00, 20:00 UTC), the statement that funding is charged on notional “irrespective of leverage,” the 75% funding-rate caps, the peer-to-peer nature of funding, and the inverse, linear and quanto contract lists. Read August 1, 2026.
2. Andrew Sears, “Perpetual futures have arrived in the U.S.” Coinbase blog, July 21, 2025. Source of the launch date, the nano BTC-PERP and ETH-PERP contracts traded through Coinbase Financial Markets, the up-to-10x leverage figure, fees from 0.02% per contract, the 90%-of-global-volume figure, and the statement that these contracts are “long-dated with expiration dates of 5 years” rather than genuinely perpetual. The nano bitcoin contract represents 1/100th of a bitcoin.
3. Commodity Futures Trading Commission, Order Approving KalshiEX LLC BTCPERP Futures Contract (issued May 29, 2026; contract submitted May 28, 2026). Source of the CF Benchmarks BRTI reference index, the 1/10,000 bitcoin unit, the convergence-mechanism discussion, the limitation of the analysis to digital commodities with deep and continuous spot markets, and footnote 25 on agricultural products. See also the Policy Statement Concerning the Listing of Perpetual Contracts, adopted May 29, 2026 and published in the Federal Register on June 3, 2026.
4. Chicago Mercantile Exchange Inc. v. Selig, No. 1:26-cv-02157 (D.D.C.), filed June 18, 2026, Judge Colleen Kollar-Kotelly. An Administrative Procedure Act challenge to the approval order, with the CFTC and Michael S. Selig as defendants. Docket checked via CourtListener on August 1, 2026: the complaint is the only substantive entry and no ruling has issued.
5. Shiller, Robert J. “Measuring Asset Values for Cash Settlement in Derivative Markets: Hedonic Repeated Measures Indices and Perpetual Futures.” Journal of Finance 48, no. 3 (1993): 911–931.
6. Kalshi, What is the funding rate for perpetual futures? Source of the three daily funding cycles at 12:00 AM, 8:00 AM and 4:00 PM ET. Read August 1, 2026.
7. Kalshi Help Center, Perpetual Futures collection, including the contract list and CF Benchmarks reference indices. Kalshi’s documentation gives the interest-rate component as 0% and the per-cycle cap as ±2.00%. Kalshi publishes no headline maximum leverage figure and none is quoted here; secondary sources conflict on it.
8. Coinbase Help, US Perpetual-Style Futures Funding Rate Mechanism. Source of the hourly premium TWAP over twenty 3-minute periods, the division by 24, the 75% / 25% smoothing weights, and the twice-daily cash adjustment through Nodal Clear recorded separately from variation margin. Read August 1, 2026 (this page returns HTTP 403 to automated fetchers and was read in a browser).
9. BitMEX, BitMEX Exchange to Sunset on 23 September at 04:00 UTC (announced July 23, 2026). Reduce-only restrictions from 04:00 UTC on August 26, 2026.
10. Bank for International Settlements, Working Paper No. 1087 (revised October 2025), on cryptocurrency carry over April 2019 to July 2024. The paper studies fixed-date futures rather than perpetual contracts, which is why its findings are scoped as such above.
11. Zhivkov, Peter. Study of leverage and margin-call probability in cryptocurrency perpetual futures, Journal of Risk and Financial Management 14, no. 2, article 346. Single-authored, with a sample of eight consecutive days.
12. Commodity Futures Trading Commission, study of 36,538 retail accounts trading CME micro equity-index futures, February 2021 to November 2022. End-of-day positions with estimated rather than observed profit and loss, and equity-index rather than crypto exposure.
13. Working paper on liquidation cascades in cryptocurrency derivatives markets (arXiv 2607.27070), finding cascades behave as shock-driven discontinuous transitions rather than critical phenomena with reliable early warning signals. No journal reference; cite as a working paper.
14. Liquidation figures for October 10–11, 2025 are CoinGlass data, republished by CoinGecko, Forbes and CoinDesk: more than $19 billion of forced closures across roughly 1.6 million accounts. Exchange liquidation feeds are throttled during stress and report notional rather than trader losses, so both figures are best read as a reported lower bound on scale.
15. CoinDesk opinion, July 29, 2026, arguing that systemic risk in perpetual markets is a function of venue design rather than of the contract structure.
16. Reuters, July 9, 2026, reporting $16.1 billion of Kalshi perpetual futures volume since launch and talks with regulators to extend perpetuals to metals, foreign exchange and energy. Bloomberg reported $5.5 billion in the first two weeks (June 16, 2026) and CNBC reported $1 billion in the first week (June 9, 2026). Trading opened June 3, 2026 with a launch cohort of four contracts (BTC, ETH, SOL, XRP), reaching sixteen listed perpetuals by July 11, 2026 in successive waves; some coverage dates the launch to May, which is when the CFTC approval issued. Set the cumulative volume figure against the size of the book: at the July 11, 2026 snapshot, bitcoin open interest was 693,877 contracts, about $4.45 million of notional, at a quoted spread of roughly 2 basis points.¹⁹
17. He, Songrun, Asaf Manela, Omri Ross, and Victor von Wachter. “Fundamentals of Perpetual Futures.” Working paper, first draft December 2022, revised August 2024; arXiv:2212.06888v6, SSRN 4301150. Proposition 1 gives F = κ/(κ − (r − r’)) · S; the widely quoted S(1 + r/κ) is its small-r approximation.
18. Ackerer, Damien, Julien Hugonnier, and Urban Jermann. “Perpetual futures pricing.” Mathematical Finance, in press (doi:10.1111/mafi.70018); working paper arXiv:2310.11771. The perpetual price as a risk-neutral expectation at a random effective maturity set by funding intensity.
19. Lim, Boon Chuan. “The Perpetual That Rarely Pays: Funding Deadbands and Basis Discipline in the First CFTC-Regulated Perpetual Futures Market.” SSRN 7098201, version dated July 11, 2026. Source of Kalshi’s disclosed funding rule (480 one-minute premium candles, no interest component, 0.01% zero threshold, ±2% clamp, Kalshi Klear clearing, funding pauses if the CF Benchmarks feed is unavailable), the 73.0% zero share across 1,226 observations and 13 contracts from June 3 to July 11, 2026, the 0.0% counterfactual zero share when the same rule is applied to Bybit’s premium index, the −4.7bp mean Bybit premium, the decomposition of Bybit’s 0.38bp mean absolute realized funding, the listing waves, and the July 11, 2026 open-interest snapshot. Caveats: the author is an independent researcher, the paper is a self-described “first look and pre-registered analysis plan” rather than peer-reviewed work, the sample is five weeks, and the price-discovery analysis that would separate its two competing explanations is registered but not yet executed. The author discloses using AI tools for code and drafting while taking responsibility for the claims. The funding-rule facts it reports match Kalshi’s own contract specifications and the CFTC approval order independently.
20. Volatility input and computed probabilities. Bitcoin’s 43.6% annualized volatility is computed here from 365 daily BTC-USD closing prices through July 31, 2026, taking log returns and annualizing the standard deviation by the square root of 365. Trailing two-year volatility on the same method is 45.8% and trailing 90-day is 35.1%. All liquidation probabilities in this guide come from the first-passage formulas in the appendix, evaluated with the same code the calculator runs.

Editor’s note

Educational content, not investment advice, and not a recommendation to trade any instrument described here. The probability figures assume geometric Brownian motion with constant volatility, zero drift unless stated, and a liquidation barrier that does not move as funding accumulates. All three assumptions make the results optimistic relative to how bitcoin actually trades. Venue specifications, funding formulas and leverage limits change frequently and are quoted with the dates they were read; contract-level parameters live in each venue’s rulebook and contract specifications, and this guide states only what those documents support. Kalshi’s general education pages describe a two-component funding rate in the industry-standard form, while its BTCPERP contract specifications set the interest component to zero and add the 0.01% threshold, so read the specifications rather than the explainer. The 73% zero-funding figure comes from a five-week, non-peer-reviewed first look and should be treated as a first observation of a young market rather than a settled property of it. Tax treatment of perpetual contracts under Section 1256 is unsettled and is a question for a tax professional. Citations verified against primary sources on August 1, 2026; the CME litigation docket should be rechecked before relying on its status.