如何實現 Crypto 的基準利率？

Original Author: @BlazingKevin_, Blockbooster Researcher

1. Crypto Has No "Base Rate"

Leverage and financing in the crypto world—trillions of dollars in leveraged positions, collateralized loans, and yield products—operate without a unified base rate curve.

According to BitMEX's Q1 2026 derivatives report, in the nascent "traditional asset perpetual" track alone, weekly single-quarter trading volume surged from approximately $525.8 million at the end of 2025 to $30.7 billion by mid-March 2026, a quarterly increase of about 5,756%. Its monthly volume skyrocketed from $7.9 billion in November 2025 to $199.1 billion in March 2026, growing roughly 25-fold in five months.

Based on DefiLlama's 30-day snapshot, Hyperliquid processed approximately $172.63 billion in perpetual trading volume, with open interest of about $9.13 billion. In Q1 2026, commodity-type perpetuals accounted for roughly 30% of Hyperliquid's open interest, primarily driven by demand for 24/7 crude oil trading.

Regarding the "traditional asset perpetual" track, Binance listed TradFi perpetual contracts on January 8, 2026, launching first with Gold (XAUUSDT) and Silver (XAGUSDT). Binance captured approximately 62.7% of the TradFi perpetual market share with this first-mover advantage, followed by Hyperliquid at 29.7%.

Hyperliquid's index data for these traditional asset perpetuals comes from a partnership with S&P Global. This collaboration—linking crypto perpetuals directly to traditional indices—is now attracting regulatory scrutiny from the U.S. CFTC.

Meanwhile, Ethena's USDe market cap fluctuated between approximately $4.5 billion and $5.9 billion in early June 2026.

Each of these products reports a "rate" or "yield"—perpetual contracts have funding rates, lending protocols have lending APRs, sUSDe has a staking yield, and tokenized treasury bonds have coupons—but crypto still lacks its own SOFR. There is no widely accepted benchmark curve that can serve as an anchor for pricing. Every exchange and protocol becomes a miniature funding market, quoting its own price, yet none share a public, credible frame of reference with each other.

2. What Qualifies as a Crypto "Base Rate"?

First, let's look at three different sets of rate comparisons:

• Set One: Base financing rate vs. product yield vs. derivative implied rate. sUSDe's APY is a product yield—a return for holders; the perpetual funding rate is a derivative implied rate—a fee paid between long and short positions to maintain the perpetual's price anchor to the spot market. The base financing rate should be a public reference that countless other products can quote and use for pricing. Product yields and derivative implied rates are not benchmarks—they are "downstream" of the benchmark, the result of various premiums and structures layered on top of the base rate.
• Set Two: Overnight rate vs. term rate. Perpetual funding rates settle every 1 or 8 hours, making them essentially an overnight rate—reflecting only the cost of capital from "now until the next settlement point," without a term structure. It cannot tell you the price difference between "borrowing for 30 days" and "borrowing for 90 days." This is just as SOFR itself is an overnight rate, relying on futures markets to piece together a Term SOFR with a term structure. A rate without a term structure cannot support any medium-to-long-term fixed-income market.
• Set Three: Real lending rate vs. algorithmic/implied rate. Real bilateral lending transactions (e.g., Bitfinex's margin financing pool, where real lenders and borrowers match) and algorithmic utilization pricing (e.g., Aave, where rates are automatically calculated from pool utilization via a formula) are two fundamentally different price generation mechanisms. The former is voted on by market participants with real money; the latter is a curve written into the protocol's code by its designers.

From these three distinctions, we can derive the criteria a "qualified benchmark" should meet:

Based on real transactions, sufficiently broad and deep underlying market (difficult for a single participant to manipulate), independent governance (no conflict of interest between the administrator and the market being priced), and ideally possessing a term structure (to support medium-to-long-term pricing).

(SOFR's underlying market is real overnight repo transactions collateralized by U.S. Treasuries, with average daily volume "often exceeding $1 trillion." This is the real trading volume of overnight repos, distinct from the notional amount of futures supporting Term SOFR.)

Applying SOFR's logic to crypto reveals structural isomorphism. The Bank for International Settlements, in its research, likens on-chain collateralized lending markets to "crypto-native money markets," whose operating mechanism resembles traditional tri-party repos—over-collateralization, mark-to-market settlement, and overnight rolling. Since on-chain lending is structurally a form of repo-style secured financing, using the design of SOFR (a benchmark built on real repo transactions) to evaluate a crypto benchmark constitutes a proper isomorphic reference.

3. What are SOFR's Characteristics? Why was LIBOR Discontinued?

LIBOR (London Interbank Offered Rate) was once the cornerstone of global finance. At its peak, approximately $300 trillion in financial contracts (including interest rate swaps, mortgages, student loans, corporate bonds, etc.) relied on LIBOR across five currency zones. However, LIBOR had a fatal design flaw: it was not based on real transactions but on daily "self-reported" borrowing cost estimates from a small panel of banks.

This flaw was fully exposed after the 2008 financial crisis. Regulatory investigations found that traders at several major global banks had systematically manipulated LIBOR quotes to benefit their own derivatives positions.

The manipulation scandal directly led to LIBOR's abolition.

Its replacement, SOFR (Secured Overnight Financing Rate), was almost a "reverse-engineering" of every LIBOR shortcoming: it doesn't use self-reported estimates but is based on real transactions in the overnight repo market collateralized by U.S. Treasuries; it takes the volume-weighted median of transactions across three repo markets (tri-party repo, GCF repo, and bilateral repo cleared through FICC's DVP service), offering a broad, deep, and manipulation-resistant scope; and it is administered by the New York Fed, adhering to IOSCO benchmark principles, with no conflict of interest between the administrator and the market being priced.

However, SOFR has an "inherent weakness": it is an overnight rate, lacking a term structure. The market needs not just "today's overnight cost" but also "expected funding costs for the next three months" to price medium-to-long-term loans. Thus, CME launched CME Term SOFR—a forward-looking rate covering four tenors: 1-month, 3-month, 6-month, and 12-month.

It uses trading data from SOFR futures to infer the market's expectation of the future SOFR path, thereby "constructing" a forward-looking term curve. (The representative notional amount of SOFR futures used to construct Term SOFR was approximately $2.3 trillion per day in Q4 2023.)

4. Some Candidate Rates for Discussion

There are many candidates in the market commonly referred to as "rates" or "yields." Let's break them down one by one, discussing why some are clearly unsuitable as benchmark rates and which ones have room for evolution.

A key thread running through all these analyses is—"who has the right to decide": market weighting, algorithmic utilization, or governance settings?

4.1 Perpetual Funding Rate (Hyperliquid / Binance)

The perpetual funding rate is the implicit price of leverage, driven by the basis between the spot and perpetual prices: essentially an overnight rate, lacking a term structure.

When the spot market for a TradFi underlying asset is closed (e.g., stocks or precious metals on weekends), exchanges cannot obtain a real spot price to calculate the funding rate. Binance's approach is to freeze the index price at the last spot price and use an EWMA mark price with a ±3% cap. Hyperliquid also switches to EWMA during weekends, setting volatility caps per instrument. During market closure, the perpetual price is essentially anchored to a predicted value, not a real transaction price. When the market reopens and the real price gaps beyond this cap, limit-up/limit-down situations can occur. Thus, prices during closed hours are forecasts, not real, arbitrageable anchors.

On May 29, 2026, the U.S. CFTC approved KalshiEX's Bitcoin perpetual contract (BTCPERP), the first truly perpetual regulated Bitcoin perpetual within the U.S., simultaneously releasing a policy statement on perpetual contracts, staff guidance on 24/7 trading clearing, and a no-action position for Coinbase offering perpetuals via Deribit. This is significant because a regulated, centrally cleared perpetual means its funding rate and basis are generated in a compliant, clearing-guaranteed environment—a potential candidate for a future "crypto SOFR." This event, alongside the CFTC's scrutiny of Hyperliquid's S&P Global index partnership, signals that "regulation is closing in on crypto benchmarks."

4.2 Bitfinex Margin Financing + FRR

This is crypto's native dollar term funding market.

Here's the mechanism: Bitfinex operates a peer-to-peer margin financing market where lenders lend funds to margin traders to earn interest. The key design feature is that financing terms range from 2 to 120 days (commonly 2, 7, or 30 days), and both the rate and term must match during matching. This means Bitfinex's financing pool naturally forms a real lending curve from short to long tenors: 30-day money and 120-day money have different prices, matched by real supply and demand. This is one of the very few real lending markets in the crypto world that inherently possesses a term structure.

The FRR (Flash Return Rate) is this market's reference rate: the FRR is the volume-weighted average rate of all active fixed-rate financing, updated hourly. Essentially, it is "Bitfinex's version of a benchmark reference rate"—an index reflecting the current average cost of borrowing in the market. Lenders can choose to lend at FRR, automatically tracking the market.

Bitfinex charges approximately a 15% fee on lending income (18% for hidden orders); the minimum order amount is $150. The FRR is quoted as a daily rate, annualized from the daily rate: Bitfinex USD's FRR is around 0.0136%/day, annualized to about 5.1%—on par with candidates like tokenized treasuries, Aave, and SSR.

The critical issue is its volatility: USD lending rates historically swing violently between roughly 3%–20% APR, strongly correlated with leverage demand.

This daily rate curve spans different tenors from 2 to 120 days, constituting a native dollar financing curve with a real term structure in crypto.

Bitfinex and Tether share the same parent company, iFinex, with overlapping management. This gives Bitfinex the most abundant USDT liquidity in the entire crypto world—one reason its financing market is so deep. However, it also concentrates counterparty risk and stablecoin issuer risk within the same complex. Borrowing Bitfinex's money, using Bitfinex's matching, denominated in Tether, with the same parent company potentially acting as a backstop in extreme scenarios—this is a highly self-contained structure.

While Bitfinex's financing market is the oldest and deepest native dollar term financing market in crypto, its absolute scale (the size of the financing pool and daily matching volume) is still much smaller compared to the trillions of dollars in trading volume seen in the aforementioned perpetual markets.

Comparing FRR to LIBOR and SOFR: on the dimension of "based on real transactions," FRR is actually cleaner than LIBOR. FRR is calculated from real, executed fixed-rate financing weighted by size, reflecting true market behavior. However, FRR originates from a single exchange's order book (concentration), is operated by the same parent company iFinex that controls the largest stablecoin, Tether (conflict of interest), and this operator also acts as the lender of last resort for this market (further concentration and conflict). Therefore, in the dimensions of concentration and conflict of interest, FRR hits precisely the issues SOFR sought to eliminate.

4.3 DeFi Lending Rates (Aave / Morpho)

This is the epitome of algorithmic utilization pricing: rates are not determined by bilateral matching but are automatically calculated from pool utilization via a preset formula—the higher the utilization, the higher the rate. It floats in real-time with borrowing demand.

Aave's USDC deposit rate on the mainnet fluctuates with utilization between roughly 3.5%–6%; Morpho's curated USDC vaults, after curator fees, yield about 5%–7%.

4.4 MakerDAO / Sky Savings Rate (DAI's DSR / USDS's SSR)

This is a "policy-like rate" directly set by protocol governance. DAI's DSR (Dai Savings Rate) and USDS's SSR (Sky Savings Rate) are widely referenced, functionally resembling a central bank's policy rate—they are not determined by market matching or algorithmic utilization triggers but by Sky's governance vote.

The contrast between DSR/SSR's governance-setting, FRR's market weighting, and Aave's algorithmic utilization represents three fundamentally different rate generation mechanisms.

Governance-setting vs. market weighting vs. algorithmic utilization—each mechanism has its own credibility issues and manipulation risks. A mature market's benchmark should ideally come from the least manipulable mechanism (market-weighted real transactions of sufficient breadth and depth). Numerically, SSR was lowered by governance from 4.75% at the end of April 2026 to around 3.6%–3.75% in early June (the "governance-set" mechanism moving in line with the Fed's path); USDS circulation is approximately $11 billion.

4.5 Tokenized Treasury Yields (BUIDL / BENJI, etc.)

This is the roughly 4–5% "risk-free leg," a candidate qualified to be a "crypto risk-free benchmark." BlackRock's BUIDL, Franklin Templeton's BENJI, and others bring the coupon yield of U.S. Treasuries on-chain. Numerically, major tokenized treasury tokens (BUIDL, USDY, USDM, USYC, etc.) paid approximately 4.1%–4.7% APY in April 2026, closely tracking the 3-month U.S. Treasury yield. Its yield can almost directly correspond to the traditional risk-free rate.

The secondary market pricing of this "risk-free leg" in tokenized treasuries is very tight—taking Ondo's tokenized treasury as an example, between February and April 2026, its transaction price median deviation was only about 2 basis points, with 95% of transactions falling within 5 basis points. This shows that when the underlying asset is sufficiently standard and risk-free, on-chain price discovery can be very precise. In contrast, the "prices" of high-risk instruments like perpetuals during closed market hours are full of predictive elements—the lower the risk, the more real the price; the higher the risk, the more the pricing resembles guesswork.

4.6 Ethena sUSDe

This is a securitized product combining the perpetual funding rate and collateral yield. Its APY is highly dependent on the funding rate level in the perpetual market, making it essentially a repackaging of implicit rates, not a benchmark itself.

Putting these seven candidates together: each measures different things (leverage sentiment, real lending, algorithmic utilization, governance policy, risk-free coupons, institutional arbitrage), each carries different risks (liquidation, counterparty, smart contract, governance, credit), and each is priced by different entities.

None simultaneously meets the three conditions of "broad scope +