The Shadow Engine of Markets: Understanding Derivatives, Perpetuals, Futures, and Options

What moves the market isn't always what you see.

The chart, the candle and the price on the screen are the visible end of a longer chain. Underneath every tick sits a lattice of contracts, promises and leverage, and that lattice pushes on the market it appears to follow. Futures, perpetual swaps and exchange-traded funds are usually filed under one heading, instruments that take their value from something else, and the filing undersells them. Each one feeds back into the thing it's supposed to track. A futures curve pulls today's price towards tomorrow's expectation. A perpetual swap ties itself to spot with a payment that, under stress, turns into forced selling. A fund that only wanted to hold bitcoin ends up buying it on a schedule set by other people's orders.

That's the shadow engine: a second layer of markets whose pressure can set the course of the first. You don't need to trade any of it for it to matter to you, because its flows reach the spot price either way.

This page is meant to be come back to. It runs through futures and the curve they draw, perpetuals and the three pieces that hold them to spot (funding, the mark price and the liquidation engine), exchange-traded funds and the arbitrage that makes them buy and sell the underlying, and a short pointer on options. Then it sets the gears running together, works through 12 and 13 March 2020 as a case study, and ends with where these readings fail and a checklist to run before trusting a derivative's price. The contract arithmetic, inverse against linear, lives in Read the Spec Sheet, and trading a price without owning the thing behind it lives in Beyond Leverage. This page is about the machinery that connects the contracts to the market underneath.

#Futures: The Original Shadow

A futures contract is an agreement to buy or sell something at a later date, at a price fixed today, made on an exchange that stands between the two sides. The idea is old. Osaka's Dojima rice market was authorised by the shogunate in 1730 to trade rice certificates for future delivery, with membership and clearing, and the Chicago Board of Trade opened in 1848 and standardised its "to arrive" contracts over the following decades. Farmers locked in a price for a harvest that didn't exist yet. Merchants locked in supply. Speculators stood between them and took the risk neither wanted.

That simplicity hid a separation that changed markets for good. Futures split the trade in risk from the trade in goods. A wheat crop could change hands many times on paper before it left the field, and today far more oil changes hands as contracts than ever moves as barrels. The contract became a market of its own, with its own price, and that price started leaning on the price of the thing itself.

#Margin, Settlement and Expiry

Four mechanics do most of the work.

• Margin. Neither side pays the full price up front. Each posts initial margin, a fraction of the contract's value, and must keep a smaller maintenance margin behind the position. That's where the leverage comes from, and where the danger does.
• Daily settlement. The exchange marks every open position to the day's settlement price and moves cash between the accounts of winners and losers, every day. A loss is paid as it happens, which is why a futures position can be closed out long before anyone is proved right about the delivery date.
• Expiry. Every contract has a last day. At expiry it settles by delivering the underlying (physical settlement) or by paying the difference in cash (cash settlement). Most positions never get there; they're closed or rolled first.
• The roll. A trader who wants to stay exposed past expiry closes the expiring contract and opens the next one. Because large holders roll on a known calendar, the roll is a predictable flow, and markets lean on it.

#The Curve: Contango, Backwardation and Basis

List every expiry of the same future side by side and you get a curve. The gap between a future's price and the spot price is the basis:

and traders usually quote it annualised, as a yield, so contracts with different expiries can be compared:

When later expiries sit above spot, the curve is in contango. That's the normal state for anything that costs money to hold, because the buyer of a future gets the asset later and keeps their cash in the meantime, so the seller wants paying for the wait. When later expiries sit below spot, the curve is in backwardation: the market will pay more to have the thing now than later, usually because it's scarce now.

The basis is also a trade. Buy spot, sell the future, hold both to expiry, and you collect the basis whichever way the price moves, because the two legs converge at settlement. That's the cash-and-carry trade, and its existence is what keeps contango from running too wide: whenever the basis pays more than money costs, someone with capital buys spot to collect it, and that buying is pressure on spot. It's the first gear of the engine, and the plainest one. A curve that pays a high basis pulls capital into spot purchases; a curve that inverts tells you the cash market is short of the thing right now.

#Expiry and Convergence

As expiry approaches, the future and spot have to meet, because at settlement they're the same price by definition. The last days of a contract are therefore a period when arbitrage forces the two together, and the hedges, rolls and unwinds around settlement push on spot as they go. In crypto, CME's futures add one more wrinkle: they trade on CME's calendar while spot trades every hour of every day, so the futures chart carries gaps where the spot chart doesn't. Those gaps, and why they so often fill, are covered in Daily vs Weekly Market Closes.

#Perpetuals: The Endless Bet

A perpetual swap is a future with the expiry removed. It's margined like a future and marked to market like a future, and it never settles. Robert Shiller proposed the idea in a 1993 paper in the Journal of Finance, as a way to make cash-settled markets in assets whose prices are hard to observe, like houses. Crypto took it somewhere he hadn't pictured. BitMEX launched XBTUSD on 13 May 2016, and every major venue copied it. A decade on, the product outlived the exchange that popularised it: BitMEX settled XBTUSD and shut down entirely on 23 September 2026, while perpetuals carry most of crypto's trading. I traded BitMEX's books in its maker-rebate years, which is the story in A Small Edge, Carefully Kept.

Take away expiry and you take away the one thing that forces a future back to spot. A perpetual needs a different tether, and it has three pieces: funding, the mark price, and the liquidation engine.

#Funding

Funding is a payment between the two sides of a perpetual, made at fixed intervals, sized on the position's notional value. When the perpetual trades above spot, longs pay shorts. When it trades below, shorts pay longs. The payment moves from one trader to another without passing through the exchange's own pocket, and it makes the crowded side pay the other side to stay.

On most venues the rate is built from two parts, a premium and an interest rate. The premium index measures how far the perpetual's order book sits from the spot index, averaged over the interval. The interest rate is a fixed number standing for the difference between borrowing dollars and borrowing the coin. Binance's documentation gives the formula most venues share:

with set to a day, which is per eight-hour interval. The clamp is what gives funding its shape. While the premium sits anywhere between and , the clamp cancels it exactly and funding comes out at the interest rate, , which is why funding sits at that number for days at a time in quiet markets. Outside the band, funding follows the premium, offset by .

Two consequences matter more than the formula. The first is cost. every eight hours sounds like nothing, and it's about a year on the notional, paid by longs to shorts even in a perfectly calm market. The second is direction. Funding pays whoever leans against the crowd, so when one side gets crowded the payment grows, and a trader holding the other side of a perpetual, hedged with spot, is paid to stay there. That hedge is the cash-and-carry trade again, rebuilt for an instrument with no expiry, and it's how funding reaches spot: the trade that collects funding buys or sells spot to stay neutral.

How to read funding as data (what a funding print can and can't tell you about positioning, why the same rate means different things on different venues) belongs to a later post on reading open interest, funding and basis. Here it's enough to know what the payment is and which way it runs.

#Index, Mark and Last

A perpetual has three prices, and confusing them is a common way to be surprised by a liquidation.

• The last price is the most recent trade on the venue's own book: what the chart shows.
• The index price is a weighted average of the asset's price on several spot venues.
• The mark price is the index adjusted for the funding that's due. In BitMEX's version, fair price marking, which other venues copied in some form:

Margin, unrealised profit and liquidations are all calculated on the mark price rather than the last price. The design protects traders from a single wick on a thin book. If someone sweeps the perpetual's own book for a moment, the mark price barely moves, because it's anchored to spot venues elsewhere. The cost is that your liquidation can arrive while the chart you're watching hasn't touched your level, and it can fail to arrive while the chart has.

#The Liquidation Engine

Every leveraged position has a liquidation price, where the losses have used up all but the maintenance margin, and behind it a bankruptcy price, where the margin is gone entirely. When the mark price reaches the liquidation price, the exchange takes the position over and closes it in the order book. On a 20× position that's a move of under 5%, and the forced close is a market order: a long that's liquidated is a sale, at whatever the book will pay.

What happens next depends on the fill. If the engine closes the position better than the bankruptcy price, the surplus goes into an insurance fund. If it closes worse, the fund covers the shortfall. And if the fund can't cover it, the exchange falls back on auto-deleveraging (ADL): it closes profitable positions on the other side, ranked by profit and leverage, at the bankrupt position's price. BitMEX's own explanation, written after March 2020, lays out the chain.

This chain is where perpetuals turn from a tether into an accelerant. Liquidations are market orders. Market orders move price. Price moving through the next cluster of liquidation levels creates the next batch of market orders. The structure that keeps a perpetual close to spot in quiet conditions is the same structure that, in a fast market, sells into a falling book without asking anyone. Liquidation levels cluster where leverage clusters, and they're the stop pools of The Depths of the Market, placed by an exchange's rules instead of by traders' hands.

#Where the Volume Sits

In crypto, perpetuals often trade far more than the spot markets they track. Leverage concentrates activity there, and a move frequently starts where the size is and reaches spot through arbitrage, the effect behind The Long Shadow of Markets. That's a tendency rather than a law, and sometimes spot leads, as it can when fund flows arrive in the US morning. But when you ask where a move came from, the derivatives venues are the first place to look, and Turn the Candles Off covers why one venue's volume can't tell you that on its own.

#ETFs: The Derivative in Disguise

Strictly speaking, an exchange-traded fund is a fund whose shares trade on a stock exchange, and a spot bitcoin fund holds the coins themselves, which by most definitions puts it outside the derivatives. It belongs in this post because of what its plumbing does to the underlying, which is the same thing a derivative does: it turns trading in one market into buying and selling in another.

The plumbing is creation and redemption. The fund doesn't sell shares to the public directly. Authorised participants (APs), usually large market-making firms, create new shares by delivering the underlying, or cash to buy it, to the fund, and they redeem shares by handing them back for the underlying. When buying pushes the fund's shares above the value of what it holds, an AP buys the underlying, creates shares and sells them at the premium. When selling pushes the shares below, an AP buys the cheap shares, redeems them and sells the underlying. That arbitrage keeps the shares close to the fund's net asset value, and every turn of it is a trade in the spot market.

The SEC approved eleven spot bitcoin ETFs on 10 January 2024, and they began trading the next day. At launch, creations and redemptions had to be in cash, so the fund itself bought and sold bitcoin to match them. The SEC permitted in-kind creations and redemptions on 29 July 2025, moving that trading to the APs. Either way the effect on spot is the same in direction: inflows into the fund become purchases of bitcoin, and outflows become sales, on a schedule set by the stock market's hours rather than crypto's.

Passive flows deepen the effect. Money that buys a fund on a monthly schedule, whatever the price, buys the underlying on the same schedule, and in equity index funds it buys every member of the index at once. That raises correlation between assets that share little but an index label, and it means a stock or a coin can move on flows into a wrapper that has nothing to do with its own news.

#Options: A Pointer

Options are the fourth instrument in the old version of this post, and the one I'm going to keep brief. An option is the right, without the obligation, to buy (a call) or sell (a put) at a set price before a set date, bought for a premium. The dealers who sell options usually hedge them in the underlying, and that hedging can push the underlying: a dealer short calls buys more of the asset as it rises, and a dealer short puts sells more as it falls, so dealer positioning can make a move run further than demand alone would. When dealers are long the options instead, the same hedging leans against the move and damps it.

That's as far as this series goes for now. I don't trade options, so the mechanics of hedging, the Greeks and how to read positioning in an options market are left to people who do it every day.

#The Shadow Engine in Motion

Each instrument pushes on spot through its own mechanism. Together they interlock, so a push in one becomes a push in the others.

• Futures into spot. A rich basis pulls capital into cash-and-carry trades, which buy spot and sell futures; as expiry nears, convergence forces the two together.
• Perpetuals into spot. Funding is collected by traders who hedge in spot, so a crowded perpetual drags spot towards it; under stress, liquidations sell into falling books and arbitrage carries the selling to the spot venues.
• Funds into spot. Creations and redemptions turn share flows into purchases and sales of the underlying, on the stock market's clock.
• Spot into everything. The index that marks every perpetual is built from spot prices, so a sharp move on the spot venues moves every mark price at once, and with it every liquidation level.

The result is a loop. Positioning in the derivatives alters flows into spot, spot moves the mark prices, mark prices trigger the next round of positioning changes, and round it goes. Much of what looks like the market discovering a price is the loop working through its own mechanics. This is why markets sometimes move with uncanny violence at levels that look arbitrary on a chart: the level wasn't arbitrary to the engine underneath.

#Reference

#Case Study: 12 and 13 March 2020

On 12 March 2020, the day after the World Health Organization declared COVID-19 a pandemic, bitcoin fell about 40% in a day. Binance's USDT-margined perpetual opened the day at 7,938.39 and traded as low as 3,621.81 in the early hours of 13 March, a fall of 54%. Coinbase's spot market fell from 7,938.05 to 3,858, a fall of 51%. Every instrument in this post was involved, and the data shows the engine working.

The volume sat in the derivative. Across 12 and 13 March, the perpetual traded 1.72 million BTC against Coinbase's 244 thousand, about seven times as much, counted in the same unit. Its daily volume had averaged about 158 thousand BTC through February; on 12 March it was 785 thousand, and on 13 March 935 thousand.

The derivative traded below spot when the selling was forced. On 11 March, a quiet day, the perpetual closed each hour within a few dollars of Coinbase, 51 cents above it on average. Then the gap opened:

A perpetual closing an hour 8% below the spot market it's tethered to is the liquidation engine at work: forced market sells arriving faster than arbitrage could carry them to spot.

Funding flipped. Binance's eight-hour funding prints had sat at the default for days. On 12 March at 16:00 UTC the print went negative, , and at 08:00 UTC on 13 March it reached , thirty times the default in the other direction, with the crowded short side now paying longs. It was still negative at the first print on 16 March. The prints are in Binance's public funding history, frozen with the figure's data.

And one of the biggest perpetual venues went quiet. BitMEX, then among the largest, reported its requests delayed between 02:16 and 02:40 UTC on 13 March, and later attributed it to a DDoS attack. The shaded hour in the figure is that window. The Block counted a record $1.6 billion in BitMEX liquidations on the day, and its insurance fund lost 1,627 BTC, about 4.6% of the fund. Price turned during and after the outage. Some traders, Sam Bankman-Fried among them, argued at the time that the outage took BitMEX's liquidation selling off the market and let price recover, and the Celsius bankruptcy estate later sued BitMEX over its March 2020 liquidations. BitMEX's account is the attack. Those are allegations and counter-arguments; the hour itself is in the data.

Hindsight proves less than it seems to here as well. The figure shows where the size was and when the gap opened; it can't show that any one participant intended what happened. What it does show is the structure in this post behaving as designed: leverage concentrated in a perpetual, liquidations arriving as market orders, a derivative trading below the market it follows, and a funding payment swinging to pay anyone willing to take the other side.

#Where It Fails

• Funding isn't a forecast. Deeply positive funding says longs are crowded, which often precedes a flush, and it can stay positive for weeks while price climbs. Reading funding well takes more than the sign, which is a later post's subject.
• The basis can stay wide. Cash-and-carry needs capital, custody and a venue you trust to pay you at expiry. When those are scarce, the basis can stay rich far longer than the arithmetic says it should.
• Mark price only protects you from one book. If the spot venues behind the index fall together, the mark falls with them, and the protection against a single wick doesn't help.
• ADL can close a winning trade. A profitable, leveraged position on the right side of a crash can be closed by auto-deleveraging at a price you didn't choose. It's rare, and it's in the rules.
• Reported liquidations undercount. The record day on the books, about $19 billion on 10 and 11 October 2025, was assembled from feeds that don't report every liquidation, and CoinShares argued the true figure was considerably larger. Totals are a floor.
• Derivatives don't always lead. Fund flows in the US morning, a large spot seller, or a venue-specific event can move spot first. "The derivative led" is a reading to test on each move, with timestamps.

#Checklist

1. Which instrument is this, and does it expire? If it does, when, and when does the roll start?
2. What index marks it, and which spot venues are in that index?
3. Where is your liquidation price on the mark price, and how far is that from the last price on your chart?
4. What's the funding interval and the current rate, and what does holding the position cost a year at that rate?
5. How big is the venue's insurance fund, and what is its ADL policy?
6. Where does the volume in this asset actually sit: spot, perpetuals or futures, and on which venue?
7. Which way is the basis or funding paying, and who is being paid to lean against the crowd?

Since this was published: Read the Spec Sheet works through the contract arithmetic, inverse against linear, that this page leaves out; Beyond Leverage covers what a derivative is for besides leverage, ratio trading above all; and Turn the Candles Off covers why spot and derivatives volume are different populations.

#Conclusion: Reading the Shadows

Derivatives are the hidden gears of the machine: an unseen lattice of contracts that leans on the price you see, sometimes gently, sometimes all at once. A curve pays someone to buy spot. A funding payment pays someone to lean against a crowd. A liquidation engine sells into a falling book without asking anyone. A fund buys the coins its shareholders asked for, on a schedule the coins don't keep.

Reading markets, then, means listening for the engine beneath the candles: where funding bleeds, where the basis pays, where liquidation levels sit, and where passive flows pass themselves off as conviction. The engine runs beneath the floorboards. Price is the sound of its gears turning.

Series note: Part 5 of Understanding Market Mechanics. Previous: Part 4 · The Script and the Rhythm. Next: Part 6 · The Hidden Hand of the Market. The practical companion is Applied Market Mechanics.

---

Understanding Market Mechanics describes how markets work, published purely as education. Nothing in it is financial advice or a recommendation to buy or sell anything, and I'm not a financial adviser. Every case study is drawn from the public record and has already resolved, and nothing in a part should be read as a statement of what I hold.