Research
A Unified Framework and Comparative Study of Decentralized Finance Derivatives Protocols
DeFi derivatives protocols let users create and trade on-chain instruments whose value derives from cryptoassets, indices, or other reference variables, yet academic research on them is thin and…

- arXiv
- 2512.19113
- Published
- 2025-12-22
- Authors
- Luca Pennella, Pietro Saggese, Fabio Pinelli, Letterio Galletta
AI summary
Overview
Research area: Decentralized Finance (DeFi), specifically derivatives protocols (perpetuals, options, synthetics) built on blockchains such as Ethereum, Solana, Arbitrum, and others; the paper sits at the intersection of financial engineering, blockchain systems, and game theory (arXiv category cs.GT).
Technical level: Intermediate — the paper uses formal notation (tuples, payoff functions, margin-ratio equations), but every construct is defined and illustrated with protocol-level examples, so a reader with basic derivatives knowledge and some blockchain familiarity can follow it.
Scope: A systematic, cross-protocol analysis of 15 leading DeFi derivatives protocols, organized into a protocol-agnostic framework that links instrument specifications, market-state variables, and protocol mechanisms, complemented by numerical simulations.
What This Paper Is About
DeFi derivatives protocols let users create and trade on-chain instruments whose value derives from cryptoassets, indices, or other reference variables, yet academic research on them is thin and fragmented compared with work on lending protocols and decentralized exchanges. The authors build a common analytical language so that heterogeneous implementations of perpetuals, options, and synthetic assets can be described and compared side by side. They then use that language to catalogue how real protocols differ in collateral rules, margin requirements, leverage, and fee mechanics.
Key Contributions
-
A formal characterization of instrument classes. Each derivative class is represented as a tripartite tuple (X, Y^market, Z^prot), separating the instrument specification (underlying, collateral, leverage, direction, entry or strike price, expiry) from the market state (index price, mark price, underlying price) and from the protocol-level mechanism (funding function, transaction fee rate, initial and maintenance margin). The paper derives economic quantities from this structure — notional exposure, signed position size, unrealized PnL, account equity, and margin ratio — including the equations N₀ = L₀·C₀, uPnL(t) = q·(P_mark(t) − P_e), and MR(t) = E(t)/N(t).
-
A taxonomy of agents, components, and operational dynamics. The framework abstracts from implementation details to identify the recurring actors and protocol components involved in issuing, trading, collateralizing, maintaining, and settling decentralized derivative positions.
-
A cross-protocol systematization of real implementations. For a sample of top protocols by Total Value Locked (TVL), the paper tabulates supported underlying assets and collateral types, maintenance-margin rules, maximum leverage, and fee structures, mapping each observed variation back to a parameter of the formal framework.
-
A simulation environment. The authors build Monte Carlo simulations intended to operationalize parts of the formal representation and explore how underlying price changes, volatility, protocol-specific fees, and leverage affect position outcomes.
Main Findings
-
The sector is small in research terms but large and rebounding in capital terms. The authors note that derivatives protocol growth contracted after mid-2022 but that TVL has been rising rapidly again since 2024, driven mainly by newer platforms such as Solana and Arbitrum, with new entrants challenging Ethereum-based incumbents.
-
Perpetuals dominate the sample's TVL. On 31 December 2024 the largest protocol in the analyzed set was Jupiter (Perpetuals, Solana) at 1720.88 million USD, with a TVL change of +1864.36% during 2024, followed by Drift Trade (802.55m, +595.48%), GMX (Arbitrum, 696.06m, −29.07%), dYdX (431.56m, +23.69%), and Hyperliquid (398.65m, +151.84%). The largest synthetics protocol listed is Synthetix (Ethereum, 381.33m, −50.23%), and the largest options protocol listed is Derive (Ethereum, 82.44m, +2202.17%).
-
Collateral is conservative; underlying coverage is not. Protocols restrict acceptable collateral to highly liquid, widely recognized assets while allowing much broader underlying universes. dYdX supports 671 underlyings spanning all categories including forex and real-world assets, but takes only one collateral asset (USDC); GMX and Hyperliquid list 49 and 45 assets respectively and allow nearly all of them as margin (49 and 45 collateral assets); Drift supports 49 underlyings but only 2 collateral assets; Jupiter lists 3 underlyings and 5 collateral assets, with stables as collateral only.
-
Leverage and margin rules vary by more than an order of magnitude. Maximum leverage ranges from 3× to above 100×. Jupiter offers 100× on Solana and 150× on Ethereum and Wrapped Bitcoin; Drift offers 101× on SOL, ETH and BTC with maintenance margin from 3% to 16.67% by asset liquidity; GMX offers 100× on major tokens and 50× on others, with its maintenance-margin policy not disclosed publicly; dYdX offers 50× on BTC and ETH, 20× on SOL, 5–10× on other tokens, with maintenance margins between 0.05% and 10%; Hyperliquid uses 3–40× leverage with maintenance margin equal to half the initial margin, yielding 16.7% to 1.25%.
-
Funding mechanisms are heterogeneous realizations of the same abstraction. Most protocols use periodic funding linked to the deviation between a protocol-internal mark price and an external index price, but the concrete designs differ: Drift charges one twenty-fourth per hour of the gap between a 1-hour EWMA mid-price and the oracle reference; dYdX uses a premium based on max(0, Bid − Index) − max(0, Index − Ask) divided by Index; Hyperliquid uses an average premium index plus clamp(interest rate − premium, −0.0005, 0.0005); GMX combines funding and borrow into a single hourly charge; Jupiter uses a borrow fee only, with no funding fee, computed as utilization ratio × hourly borrow rate × position size.
-
One-off trading costs are consistently proportional but structurally different. Jupiter charges a flat 0.06% of notional per open or close plus a price-impact fee; Drift charges 0.03–0.10% base with a 0.75–5.0% insurance/borrow surcharge; GMX charges 0.04–0.06% discounted by volume tier plus price impact; dYdX charges 0.025–0.05% with volume tiers; Hyperliquid uses a maker–taker schedule of 0–0.045%.
-
Everlasting options are positioned as an intermediate instrument class. Because they replace a maturity date with funding payments and margin constraints, the paper argues they bridge the static payoff structure of expiring options and the continuous-time exposure and funding dynamics of perpetual contracts. Expiring options, in contrast, are described as not relying on intertemporal funding or margin dynamics in their simplest, fully collateralized form.
-
Simulation results are not reported in the available text. The excerpt ends within Section 5.2.2, before the Monte Carlo section (Section 7), so the specific numerical outcomes of the simulations, including any quantified effects of price paths, volatility, fees, and leverage on profitability or liquidation risk, are not reported here.
-
Two candidate protocols were excluded during screening. Outcome Finance was excluded due to inaccessibility and Opyn because its latest dApp was only available on testnet, illustrating the reproducibility constraints of studying fast-moving DeFi systems.
Methodology in Plain English
The authors chose protocols using a data aggregator rather than a hand-picked list. They used DeFiLlama, which tracks metrics such as TVL and sorts protocols into categories, and took the top five by TVL in each of three categories: Derivatives (mostly perpetuals), Synthetics, and Options. They excluded two adjacent categories, Risk Curators and Insurance, as out of scope, and then manually checked that each candidate had an accessible website, trading application, and documentation, dropping any that failed.
For each of the 15 remaining protocols they performed a manual, source-based review: official whitepapers and developer documentation, the live dApps and websites, smart contract source code where available through Etherscan, GitHub, or protocol explorers, and community channels such as official Discord servers and technical articles. An appendix table lists all sources used.
They then worked inductively. After cleaning, filtering, comparing, and harmonizing this heterogeneous material, they abstracted away from implementation details to find structural elements that recur across protocols. That process produced the three-part organization of every instrument class — instrument specification, market state, and protocol mechanism — from which the main economic quantities are derived. Finally, they built a Monte Carlo simulation environment to exercise selected parts of the formal model under stylized assumptions, varying underlying prices, volatility, fees, and leverage.
Why This Matters
Impact on research. Prior work on crypto derivatives has concentrated on centralized exchanges or on single protocol types; the authors state that, to their knowledge, no earlier work builds a unified cross-protocol conceptualization, maps actors and operational flows across designs, or complements this with a simulation framework quantifying how parameters and market conditions affect profitability and liquidation risk. This gives future empirical and theoretical work a shared vocabulary and a reusable formal scaffold, in the way that systematizations already exist for lending protocols, automated market makers, yield aggregators, and staking protocols.
Real-world applications.
- Trader and risk-manager comparison: the leverage, margin, and fee tables let users compare the actual cost and liquidation exposure of equivalent positions on Jupiter, Drift, GMX, dYdX, or Hyperliquid before committing capital.
- Protocol design and auditing: the tuple-based decomposition gives designers and auditors a checklist of the parameters (funding function, transaction fee rate, initial and maintenance margin) that determine a position's solvency path.
- Regulatory and comparative analysis: a protocol-agnostic vocabulary supports describing decentralized instruments in terms familiar from traditional finance forwards, futures, options, and swaps, which helps in classifying these products.
- Data aggregation and taxonomy work: the paper documents how aggregator categories (Derivatives, Synthetics, Options) map onto economically distinct instrument classes, and why that mapping is imperfect.
Industry relevance. The sector's capital is concentrated in a handful of protocols whose TVL swung dramatically within a single year — from GammaSwap's +34860.89% and Derive's +2202.17% to Taiga's −85.64% and Synthetix's −50.23% — while competitive pressure has shifted from Ethereum toward Solana, Arbitrum, and other chains. Understanding which design choices (collateral policy, leverage caps, funding formulas) differ across these venues is directly relevant to users choosing where to trade and to builders competing on mechanism design.
Future Directions
-
Empirically validating the framework against on-chain data. The paper's comparison is built from documentation and interfaces; testing whether the predicted margin and liquidation dynamics match realized on-chain liquidations across the 15 protocols is a natural next step.
-
Extending and quantifying the simulation study. The simulation environment is designed to vary underlying prices, volatility, protocol fees, and leverage; open questions include how sensitive position outcomes are to each parameter and whether specific funding or margin designs systematically reduce liquidation risk relative to others.
-
Broadening the protocol sample and tracking its churn. The selection is a TVL-based snapshot of top-five protocols per category, and the authors explicitly note the difficulty of covering all protocols in a fast-moving ecosystem. Two protocols were already excluded for inaccessibility, so periodic re-sampling as new chains and entrants rise would test the framework's durability.
-
Formal treatment of everlasting options and other hybrid instruments. The paper flags everlasting options as an intermediate class between expiring options and perpetuals and indicates that a full treatment is developed further than the excerpt shows, leaving room for work on how funding-based option designs should be priced and risk-managed.
Target Audience
Researchers and graduate students in blockchain, DeFi, and financial technology who need a structured entry point into decentralized derivatives; protocol designers and smart-contract auditors who want a parameter-level comparison of funding, margin, and collateral mechanisms; market-risk and quantitative analysts moving from centralized to on-chain derivatives; and data providers, regulators, or industry analysts who need a precise vocabulary for distinguishing perpetuals, options, and synthetic assets in DeFi.
Authors’ abstract
Decentralized Finance (DeFi) applications introduce novel financial instruments replicating and extending traditional ones through blockchain-based smart contracts. Among these applications, DeFi derivatives protocols enable the creation and trading of decentralized derivative instruments whose value depends on underlying cryptoassets, indices, or other reference variables. Despite their growing significance, however, they remain relatively understudied compared to other…
The full abstract on arXiv