Chapter 18 · Why Perpetuals Became a Core Crypto Product
What anchors a contract that never settles to its spot price?
- Capabilities practiced
- Financial LiteracyOnchain Literacy
- Hands on
- Record 24 hours of funding rates and open interest for a major coin, and mark when they move in the same direction.
- AI Lab
- Have AI simulate a liquidation cascade for a given leverage and price drop. Recalculate the liquidation price yourself.
A Real-World Problem
Open two screens on the same exchange.
One shows a major coin's spot price; the other its perpetual price. They differ by a fraction of a percent, with the gap changing throughout the day.
Traditional futures also differ from spot, but have a reason to converge: a settlement date. As expiry approaches, prices draw together, and must match at settlement or create free arbitrage. Expiry is the tether pulling futures back to spot.
Cut that tether. A perpetual never expires and never requires delivery of real coins. It is an accounting entry on an exchange or in a contract.
Why does a contract that never settles not drift away?
It could. Enough bullish buyers might push it 5%, 20%, or 100% above spot without ever needing delivery to test that price.
In practice it generally stays near spot, briefly diverges, and returns.
What keeps it there?
The answer matters beyond exchange design: perpetuals are among crypto's largest products by trading volume and its main concentration of leverage. Much of Chapter 21's market-cycle fuel accumulates here.
Thought Experiment
Remove the exchange and leave two people.
A village board quotes a commodity at 100. A expects a rise; B expects a fall. They sign a paper: no delivery, only payments matching changes in the public quote. At 110, B pays A ten; at 90, A pays B ten.
The paper has no expiry.
More people join and transfer these papers, which develop their own price. A problem appears:
If bullish demand pushes the paper to 102 while the board stays at 100, what happens?
Nothing. Nobody must deliver commodities on a settlement day, so 102 can persist. The contract loses its connection to its reference.
The village considers three solutions.
First: add expiry. Settlement at the public quote in a month restores convergence, but creates ordinary futures. Long-term holders must repeatedly close and reopen, paying each time.
Second: allow delivery. Someone buys the commodity and uses the paper to demand a 102 sale. This restores the link but requires someone to hold real inventory, eliminating the no-delivery benefit.
Third: charge the crowded side periodically. Every eight hours, if the paper trades above the board, long holders pay short holders in proportion to position size. Greater divergence means larger payments.
This does not force closure or require commodities. It simply prices the choice to stand on the crowded side.
Longs become costly, while shorts receive income. People gain a reason to short; capital moves toward the scarce side and pulls the contract near its reference.
Futures converge through one-time settlement; perpetuals through recurring payments. One imposes a hard constraint, the other ongoing economic pressure.
That is the first central idea. Add the second.
The village changes the rules: posting only 5% lets you sign a paper ten times as large. The dealer advances the remaining 95%, but closes the position as soon as your contribution is exhausted.
This change transforms the market.
Your Decision
You design the market. The contract has stayed 2% above spot all day. What do you do?
What Happens
The fourth option is the funding rate:
| Contract versus spot | Rate sign | Payment direction | Pressure |
|---|---|---|---|
| Premium | Positive | Longs pay shorts | Longs cost more; shorts receive incentives |
| Discount | Negative | Shorts pay longs | Shorts cost more; longs receive incentives |
| Near parity | Near a baseline | Small payments | Little pressure |
Three commonly misunderstood points:
- Funding generally does not go to the exchange, though some venues take a share. It transfers between traders: holding cost, or holding income.
- Payments recur. Eight-hour intervals are common, with hourly designs too. Shorter intervals tighten the anchor with smaller individual payments.
- The sign can reverse. Today's payer may receive tomorrow; treating funding as fixed misses this.
Annualize it to understand scale:
Payment = position notional × current funding rate
0.01% every 8 hours → 0.03% daily → approximately 11% annually
0.10% every 8 hours → 0.30% daily → approximately 110% annuallyThe second has occurred in real markets. A correctly positioned long can still lose after triple-digit annualized holding costs. This mirrors Chapter 17: it asked who pays yield; here ask whom I pay this cost to. The answer is the trader opposite you.
Now add leverage to examine the second property of this market.
Leverage determines the adverse movement you can withstand. Assume a 0.5% maintenance margin:
| Leverage | Margin as share of notional | Approximate adverse move to liquidation | Intuition |
|---|---|---|---|
| 2× | 50% | 49.5% | Nearly halving |
| 5× | 20% | 19.5% | A major correction |
| 10× | 10% | 9.5% | An ordinary intraday move |
| 20× | 5% | 4.5% | One news event |
| 50× | 2% | 1.5% | One minute |
| 100× | 1% | 0.5% | An order-book fluctuation |
Derive rather than memorize:
Notional = margin × leverage
Liquidation distance ≈ (1 ÷ leverage) − maintenance margin ratioActual maintenance requirements increase with position-size tiers. This table is simplified; check your venue's parameters.
Chapter 16 showed the same structure. A 10× long resembles a 90% LTV collateralized position: contribute 10%, borrow 90%, buy the same asset. In the simulator, leverage = 1 ÷ (1 − L): 50% gives 2×, 80% gives 5×, and 90% gives 10×.
Liquidation threshold: 80% . At this LTV, a price fall of about 37.5% triggers liquidation. In a rapid decline, liquidations can push the price down further.
The simulator fixes its liquidation threshold at 80%, whereas perpetual maintenance sits closer to full collateral value, so perpetuals tolerate a little more at the same borrowing ratio. But perpetuals permit 95% and 99% borrowing equivalents beyond the simulator's range, leaving much thinner cushions overall.
Move borrowing above 90% and change price. At high leverage, health falls off a cliff, shared by more than your own position.
- Price falls
- High-leverage longs reach liquidation
- The engine sells at market
- Depth is consumed and prices fall further
- The next group reaches liquidation
Forced liquidation is a market sale into Chapter 15's finite depth. Positions share the same price and trigger one another.
"My 3× position is safe" works only in isolation. If large positions use 20×, a 5% decline starts their dominoes; their selling can push the market down to your threshold.
Build a Model
Three prices
A perpetual market has three simultaneous prices:
| Price | Definition | Purpose |
|---|---|---|
| Index price | Weighted prices from multiple spot venues | Funding reference and mark input |
| Mark price | Index plus smoothing or premium adjustments | Profit accounting and liquidation |
| Last price | Latest trade in this contract | Display and order matching |
Liquidation uses mark price, not last price. A thin contract book can produce a manipulated spike that liquidates many positions. This happened in early markets and motivated the shift to marks.
The index's sources and authority are oracle questions. T21 covers engineering details; Chapter 24 classifies the risk.
Funding components
Funding ≈ interest component + premium component, bounded by caps
Premium ≈ (contract price − index price) ÷ index price,
averaged over a periodInspect three parameters:
- Settlement interval: eight hours or one? This affects anchor tightness.
- Caps: maximum and minimum rates. Narrow caps weaken anchoring in extremes, limiting the pressure available.
- Charging basis: notional or margin? They differ by leverage.
Two margin modes
| Isolated | Cross | |
|---|---|---|
| Maximum exposed funds | Position margin | Entire account balance |
| One liquidation | Other positions unaffected | Can endanger the account |
| Liquidation threshold | Predetermined | Moves with other positions |
Cross-margin liquidation prices move with other positions' profit and loss. Calculating once at entry does not make that value constant.
Open interest: the leverage fuel gauge
Open interest is outstanding contract notional. Volume is flow and can be inflated by repeated trading; open interest is stock, showing leverage still present now. Combine it with funding:
| Open interest | Funding | Structure |
|---|---|---|
| Rising | Positive and increasing | New long leverage with growing holding cost |
| Rising | Negative and increasing in magnitude | New short leverage |
| Falling | Returning quickly toward zero | Leverage closing or liquidating |
| Falling | Remaining extreme | One side liquidating without full clearing |
These describe structure, not direction. They indicate how much must move if price moves, rather than forecasting price. This chapter teaches market fragility, not trading signals.
This table prepares directly for Chapter 21. Open interest is its cycle amplifier; liquidation cascades are the same feedback loop operating across the whole market.
What It Is Called
Direct asset exchange that gives actual ownership.
Spot anchors index prices, funding, and realizable collateral sales. Spot depth determines whether derivatives can unwind under extreme conditions, applying Chapter 15.
A no-expiry contract for differences anchored by funding rather than delivery.
It serves long-term directional exposure without repeated rolls, particularly valuable in continuously open crypto markets. Its costs are ongoing funding with changing signs and potentially high leverage.
Periodic payments between longs and shorts based on contract-index divergence.
Inspect interval, caps, and charging basis. Caps limit the anchor's strength when divergence becomes extreme.
It is not a trading fee. The payer and recipient are opposing traders.
Capital posted for a leveraged position.
Initial margin sets position size; maintenance margin sets liquidation. Also distinguish isolated from cross: cross liquidation prices move with other positions, rather than remaining constant.
Outstanding contract volume.
It measures leverage stock, not direction or activity. Check venue coverage and coin versus dollar denomination; price changes can make these series move oppositely. Chapter 23 formalizes these checks.
Forced market closure when margin falls below maintenance requirements.
It resembles Chapter 16, but leverage can be much higher. Deficits first hit the insurance fund, then potentially automatic deleveraging, forcibly reducing profitable opposing positions.
Even correct, profitable positions can be partially closed by the rules. This is a distinctive counterparty risk.
Hands-on
No position or transfer is required. This is public-data recording. If you later want to inspect a position interface, use a testnet or a fully expendable minimum amount. The output is a chart and three conclusions, not profit.
Choose the asset and venue, then specify definitions.
Use a major asset to reduce noise. Apply Chapter 23's habit and record:
Asset: ...
Venue coverage: one venue or weighted aggregation? How are weights defined?
OI denomination: coins or dollars?
Funding interval: hours per period? Settled rate or predicted rate?Write these before collecting data, or tomorrow your own table may be ambiguous.
Record one row hourly for 24 hours.
| Time, UTC | Mark price | Current funding | Open interest | 24-hour volume |
|---|
Fill sleeping hours from historical data, marking them as backfilled: backfilled and contemporaneous observations differ in reliability.
Plot price, funding, and open interest. Give funding its own axis and show zero, distinguishing payments by longs above it from payments by shorts below.
Mark joint changes:
- Rising price, positive increasing funding, rising open interest
- Falling price, rapidly falling open interest, funding retreating from extremes
These represent leverage accumulation and clearing. A day often contains one or two such stretches.
Investigate the sharpest hour. Compare liquidation size and direction with the chart's turning point.
A match reveals a link in the loop: falling open interest can reflect forced market sales.
Write three sentences:
- Funding was positive for X hours and negative for Y, indicating the crowded side was primarily [longs/shorts].
- The largest open-interest decline occurred at [time], with a Z% price [rise/fall] and liquidations of [amount or unavailable].
- Annualizing the current rate gives [value], corresponding to [long/short] annualized holding cost of [value].
Completing these finishes the Lab. If the third is unclear, recheck whether fees apply to notional or margin.
AI Lab
Request a numerical simulation that can be reproduced line by line, rather than an essay about cascades.
Simulate a liquidation cascade using exactly these inputs:
- Initial asset price: 100
- Three groups of longs:
A: 5× leverage, 100 million notional
B: 10× leverage, 300 million notional
C: 20× leverage, 200 million notional
- Maintenance margin: 0.5% for all
- Depth: each 50 million sold lowers price by 1%
- Initial trigger: a 30 million market sale
For every round, give:
1. Price reached
2. Groups hitting liquidation and calculated thresholds
3. Total liquidated notional
4. Price after those sales under the depth assumption
5. Whether another round triggers
Finally:
- Total price decline
- Difference if each 200 million sold produces a 1% decline
- Any negative-equity positions and who absorbs deficits
Show formulas for every number. For unspecified parameters, write
"needs specification" rather than inventing values.Then recalculate the first round by hand:
5× liquidation price = 100 × (1 − (1 ÷ 5 − 0.005)) = 100 × 0.805 = 80.5
10× liquidation price = 100 × (1 − (1 ÷ 10 − 0.005)) = 100 × 0.905 = 90.5
20× liquidation price = 100 × (1 − (1 ÷ 20 − 0.005)) = 100 × 0.955 = 95.5These become your reference. Models often silently alter the critical assumption rather than miscalculate, especially depth. Assuming sales do not move price ends the simulation after one round and produces a false no-cascade conclusion.
The real output: the same leverage structure can produce completely different outcomes under different depth assumptions. That explains why one market experiences a correction and another collapse.
Verify these points yourself after using AI
- Does liquidation use mark price rather than last trade?
- Does it include maintenance margin instead of only inverse leverage?
- Does it assume infinite depth? Without price impact, the cascade disappears.
- Does it assume full liquidation although many venues use partial liquidation?
- Does funding consumption of margin appear for long holding periods?
- Can you reproduce every first-round formula by hand?
- Does it identify the deficit bearer: insurance, automatic deleveraging, or bad debt?
- For generic claims, ask which venue and parameter they refer to.
Real-World Case
Thin books let modest market orders create spikes that triggered liquidation; forced orders extended the spike. Prices later recovered, but closed positions did not.
Markets shifted toward weighted multi-venue spot indices and smoothed marks: liquidation references moved from a manipulable thin market to a thicker one.
This is T21's oracle problem: thin price sources cost less to manipulate.
Funding can remain positive for weeks or months and repeatedly reach caps. Longs are crowded and continually pay shorts, sometimes at striking annualized rates.
Positive funding does not forecast falling prices, nor negative funding rising prices. It describes crowding rather than direction, indicating forced movement if price reverses.
Different centralized and on-chain venues reference the same spot prices. Liquidations on one affect spot; spot changes other indices and marks; those trigger more liquidations hitting the same books.
Different venues do not diversify a shared price feedback loop. This is Chapter 21's reflexivity on a timescale of minutes.
Gaps can cause liquidation beyond available margin, creating a deficit. Insurance pays first; insufficient insurance triggers automatic deleveraging, closing profitable opposing positions according to profit and leverage rankings.
Being right about direction does not guarantee realizing all profit. This mechanism risk lives in rules, not prices. Chapter 24 classifies it.
Following macroeconomic news, leveraged positions collapsed across markets overnight. Some internal reference prices temporarily diverged and triggered inappropriate liquidations; others exhausted insurance and deleveraged profitable positions.
Several mechanisms in this chapter appeared together. See Hot Topic: Liquidation Cascades.
Change One Variable
Calm markets barely notice. Extremes reveal the limit: the anchor reaches its maximum force while divergence keeps growing.
Only arbitrage remains, precisely when arbitrage capital is most strained. Caps define anchor strength and deserve inspection, despite commonly being ignored.
Individual liquidation distance shrinks from 4.5% to 0.5%. More importantly, market structure changes: the same margin supports fivefold notional and thin-cushion positions accumulate. A 0.5% move can trigger what previously required 5%.
Higher maximum leverage changes the disturbance needed to destabilize the market. Read leverage limits with open interest.
Thick books hide the difference. Thin books create a calculable attack: estimate liquidation thresholds and the cost of reaching them, then compare with forced-selling profits.
A parameter change turns liquidation from a response to volatility into something attackers can manufacture. Chapter 24 examines correct-code economic failures.
Three weaknesses reinforce one another.
Thin indices lower manipulation cost. Shallow spot depth magnifies liquidation impact. Arbitrageurs avoid thin hedge markets, weakening the final anchor.
Perpetual safety depends on underlying spot depth, beyond contract code. Identical code and parameters can produce radically different risks with a different asset.
Takeaway Questions
The product serves long-term directional exposure in a nonstop market without repeated rolls or asset custody.
Ask what inconvenience any derivative removes and for whom. Leverage alone is an amplifier, not a complete user need.
Opposing traders pay and receive funding.
It is generally not protocol revenue, is not a trading fee, and can reverse direction. Annualize funding to understand real position cost, a quick calculation absent from many analyses.
There are two liquidity layers: the contract book and underlying spot depth.
The first sets entry and exit slippage; the second enables extreme unwinds and arbitrage. Many inspect only the first. Thick derivative books over thin spot markets deserve scrutiny.
Three layers:
Normally, traders bear their calculable liquidation risk. After negative equity, insurance absorbs deficits. After insurance exhaustion, profitable opposing traders face automatic deleveraging.
Correct positions can still be closed by rules. Inspect insurance size and deleveraging procedures.
Chapter Check
Funding periodically charges the premium side, with larger divergence increasing payment.
It replaces one-time hard convergence with continuous economic pressure.
Nonstop markets avoid repeated rolling interruptions, and arbitrageurs receive income while waiting, reducing uncertainty about an indefinite convergence period.
Thin contract books make last price easy to move.
Using it transfers liquidation authority to a manipulable number. If reaching a threshold costs less than the liquidation opportunity it creates, an attack is economic.
Marks use multi-venue weighted spot indices and smoothing, moving the reference to a thicker market and increasing manipulation cost.
Therefore inspect index price sources first. Thin sources weaken this protection.
In isolation, roughly 33% liquidation distance is substantial.
It assumes other positions cannot affect yours. A market full of 20× longs can liquidate after 5%, consume depth, trigger 10× longs, and continue falling.
Others' leverage becomes your price. Inspect market-wide open interest and leverage distribution as well as your own threshold.
No. Avoid that misunderstanding.
It indicates crowded longs, whose holders currently pay for exposure. Prices can rise for weeks while funding remains positive.
It reveals how much forced movement a reversal could create. Combined with open interest, it describes fragility rather than direction.
This chapter teaches fragility, not a trading strategy.
Check:
- Spot index sources and weights, the price foundation.
- Mark derivation, smoothing and its window.
- Funding interval, caps, and charging basis, defining anchor strength.
- Leverage limits and maintenance tiers, defining instability thresholds.
- Full or partial liquidation, fees, and executors.
- Insurance size and deleveraging ranking, determining access to profits under stress.
- Underlying spot depth, without which the earlier safeguards cannot ensure an orderly unwind.
Move from price foundations to anchoring, leverage, and failure response. Many begin at item 4, while items 1 and 7 determine the risk's scale.
Key idea
Funding rate anchors perpetual futures; leverage is its feedback loop.