Uniswap创始人：当股票、国债全面上链，AMM将如何重构全球市场？

Original Author: Hayden Adams, Founder @Uniswap

Original Translation: TechFlow

TechFlow Overview: Uniswap founder Hayden Adams argues that as traditional assets like stocks become tokenized, "correlation trading pairs" such as NVDA/SPY will reduce inventory and hedging costs, giving passive AMMs an opportunity to enter the core markets long dominated by traditional market makers. Crypto KOL Cody, drawing on historical backtesting and real LP experience, further analyzes the feasibility of this approach in the current on-chain market.

Below is the full text:

This is my first blog post since 2019!

Correlation Trading Pairs: How AMMs Win the Biggest Market

I have been working at the frontier of DeFi for 9 years. It is a fascinating field with infinite depth and the potential to transform capital markets.

I have always believed in the enormous potential of AMMs, but over the past decade, one big question has haunted me: Can this new market structure truly become the core engine of all financial markets?

After years of evolution and growth, a path toward AMM global dominance is becoming increasingly clear. And the best way to explain it starts in 1976.

Tokenization Changes Who Makes Markets

Index funds celebrate their 50th anniversary this month. When Jack Bogle launched his in 1976, he hoped to raise $150 million. He raised only $11.3 million. Competitors called it "Bogle's Folly" and printed posters claiming index funds were "un-American." They believed a fund that made no decisions could never beat professionals who made decisions for a living. Today, most U.S. fund assets are held in passive vehicles.

I have been thinking about this recently because the "folly moment" for tokenization is coming to an end. The SEC has approved Nasdaq and NYSE to trade tokenized stocks. The DTCC, which settles nearly every U.S. security, conducted live trials of tokenized settlement in July. Almost all of this activity has been described as the same thing: tokenization as an infrastructure upgrade.

The same market, faster, cheaper, and running 24/7. All of this is true, but I believe the "upgrade" framework obscures a bigger story. Tokenization makes markets programmable: it changes which markets exist, who makes markets, and what counterparties those markets trade against.

In 2018, I built Uniswap. Anyone can deposit two assets into a shared pool, earn fees on every trade, while prices adjust along a curve with buying and selling activity. Uniswap has operated autonomously since day one, settling over $4.6 trillion in volume and helping drive decentralized exchanges from less than 1% of centralized spot volume to over 20%.

As AMMs like Uniswap have grown, their liquidity has formed a pattern that most of the financial world has not yet noticed: correlation trading pairs.

The Easiest Market to Win

To win everything, you must first win something. AMMs first found product-market fit in long-tail markets, where most assets cannot attract professional market maker attention at all. On Uniswap, anyone can create a market in a single transaction, and issuers and early supporters become the first LPs.

Stablecoin pairs followed: on pairs like USDC/USDT, a good passive strategy is close enough to optimal, and lower capital costs are enough to close the gap. That is why professional trading firms today no longer bother making markets for these stablecoin swaps: they have been competed out of the market by passive AMMs.

High Profits, No Competition

Traditional financial markets belong entirely to market-making firms. They bundle capital, trading strategies, execution technology, settlement, and distribution into a vertically integrated business. The evolution of this architecture has its logic: assets exist in isolated systems, settlement is slow, and every function requires someone to perform it, so one firm takes on all functions.

At sufficient scale, all of these fixed costs pay for themselves. Citadel Securities handles about 25% of U.S. equity volume and generated a record $12.2 billion in net trading revenue last year on roughly $21 billion of trading capital.

Most people view these numbers as proof that the system works well. I view them as complacency.

Breaking the Bundle

Blockchain creates competition at every layer, breaking the bundle. Execution happens through code. Custody and settlement become shared services that anyone can plug into. What once required proprietary infrastructure is now open-source software.

In an AMM, capital is the scarce input, and the advantage belongs to whoever can hold inventory at the lowest cost. Trading firms need high returns to justify their overhead, so LPs willing to accept lower returns can undercut them. Most market makers hedge all price exposure, and hedging costs money, so investors who already hold the asset can take on that exposure for free. And asset issuers have negative capital costs, because they typically have to pay professional market makers to make markets for their newly issued assets.

In short, DeFi and AMMs have lowered the barrier to market-making, opening the field to many new participants. Their advantage can come from a variety of sources: lower capital costs, a willingness to hold inventory exposure that professional firms typically hedge away, or even being the issuer themselves.

But it all depends on one question: Can automated strategies perform well enough to make this work?

Liquidity Follows Correlation

Recently, I was on a call with one of the largest institutions in finance. They asked me what the most common benchmark trading pairs in DeFi are. I explained that Ethereum-based assets tend to trade against ETH, Solana assets against SOL, stablecoins pair with each other, and there are a handful of high-liquidity pairs bridging between these clusters.

No one designed this pattern. It emerged naturally, partly because LPs perform best when the assets they hold move in sync. Correlation means lower inventory risk for liquidity providers, which deepens liquidity. As assets become tokenized, the world's biggest markets will reorganize in the same way.

They cannot do this today. Traditional markets settle in dollars out of necessity. Assets exist in isolated systems, and fiat rails like SWIFT and Fedwire are the glue holding everything together. But blockchain is a far more expressive glue. Tokenize the assets, and they share a settlement layer, so any asset can trade directly against any other asset.

NVDA/USD can become NVDA/SPY, with SPY/USD serving as the bridge back to dollars. Oil companies can trade against oil ETFs or tokenized oil. Private credit can trade against tokenized Treasury funds. Tokenization also makes possible markets across different asset types that are extremely difficult, or even infeasible, in traditional financial infrastructure.

Delta Neutrality Is an Inefficiency

Traditional market-making firms typically try to remain "delta neutral" — trader jargon for being denominated in dollars and wanting to minimize any non-dollar risk. When making markets for a volatile asset, they spend money to reduce their non-dollar risk (i.e., hedging), usually through options. This is one of the more costly components of traditional market-making.

Pairing assets into low-volatility "correlation trading pairs," connected by a few high-volatility "bridge pairs," brings numerous efficiency gains, but the most important one is this: if the person making markets actually wants to hold the underlying asset, market-making becomes cheaper and more efficient.

And the higher the correlation between the pair, the smaller the gap between today's passive AMM strategies and the most sophisticated active strategies, making it easier to "undercut" with lower inventory costs.

To give a concrete example: if someone is long Nvidia, you might also be long SPY, and the efficiency gap between a passive AMM and an active strategy on NVDA/SPY is far smaller than on NVDA/USD.

Interconnected Liquidity

If stocks instead trade against SPY, then every trade that starts or ends in dollars will route through the same pair: SPY/USD. These bridge pairs will still require professional capabilities, but they will be far fewer in number, and they will carry so much flow that professional attention is justified.

DeFi has already proven this. ETH/USDC is one of the deepest on-chain markets because every cluster routes through it. Passive LPs provide correlation pairs, while active LPs compete on bridge pairs.

Investors can still buy and sell everything in dollars because cross-pool routing is automatic. Liquidity will concentrate where risk is lowest, not where legacy plumbing requires it to sit. This pushes the deepest markets toward correlation pairs — the very territory where AMMs are already strongest.

Correlation RWA Pairs Already Exist

On-chain correlation liquidity starts with crypto-native assets. But the first correlation markets for tokenized stocks already exist: ten tokenized stocks trade against SPY in Uniswap pools on Robinhood Chain.

In the first 12 days, these pools completed over $33 million in volume from more than 11,000 traders, with a significant portion occurring during U.S. market close hours. Some trades go directly from one stock to another, bypassing dollars entirely.

It is worth noting that we are also beginning to see memecoins paired with "related" stocks: Elon meme coins against Tesla stock, hot dog meme coins against Costco stock. It is unclear how much they actually correlate in price, but I suppose "vibes" count as a form of correlation.

AMMs Will Win

Correlation trading pairs are only part of the puzzle. Another part is AMM design and customization.

Uniswap v4 hooks enable comprehensive market customization that can significantly improve LP returns, such as our recently released DualPool hook, which puts passive AMM capital to work earning lending yields when it is not being used for swaps.

Despite Uniswap's ~$4.6 trillion in volume, I believe AMMs are still in their infancy, and there are many other paths that will enhance their competitiveness. There are many other promising approaches to improving LP returns being built internally at Labs and externally by our partners and ecosystem. More to come!

The argument against index funds in 1976 was: a fund that makes no decisions can never beat professionals who make decisions for a living. Fifty years later, funds that make no decisions beat about 90% of professionals. More importantly, index funds democratized investing and improved the lives of ordinary people. I believe passive liquidity will win with a similar strategy and have an even greater impact — by dramatically lowering the barrier to creating and participating in markets.

Crypto KOL Perspective

Hayden Adams's article has also sparked discussion in the Chinese community. Crypto KOL Cody (@Cody_DeFi) subsequently provided further analysis of the "correlation trading pair" concept from the perspectives of actual market-making and historical backtesting. In my opinion, this is a fairly good interpretation and analysis of the Uniswap founder's original design thinking. The original post is also shared below:

It feels like nobody in the Chinese community is paying attention to the Uniswap founder's magnum opus, "How to Use AMMs to Replace Traditional Market Makers for On-Chain Stocks." Has no one in DeFi noticed?

The founder @haydenzadams actually highlighted a core point: if you are willing to actively hold an asset and intentionally take a long exposure to it, your market-making cost is far lower than that of a traditional market maker, because you do not need to hedge against price fluctuations. Traditional market makers do — they maintain so-called delta neutral positions, which carry significant hedging costs.

Due to the mechanics of AMMs, they can loosely be viewed as an asymmetric fee-based grid trading system. If you hold a long position, such as NVDA, and want to use an AMM to make markets in the NVDA-USDC pair, you face the risk of selling into strength. As NVDA's price rises, you accumulate more USDC and less NVDA. These losses are usually called "impermanent loss."

To address this selling-into-strength risk, Hayden introduces the concept of "paired AMM market-making." Simply put, instead of deploying an NVDA/USDC AMM pair, an ordinary person can deploy a correlated pair like SPY/NVDA. This reduces impermanent loss while maintaining a long exposure to a class of correlated assets and earning fees.

This idea stems from his observation of the mainstream deep liquidity pairs in crypto: ETH and SOL serve as primary quote assets for tokens within their respective ecosystem, just like the cover image of this article depicts.

Finally, he argues that this strategy of market-making for those willing to hold long spot positions will ultimately take over an increasing share of tokenized stock market-making strategies, just as after Uniswap's AMM emerged, on-chain spot volume relative to traditional CEXs has gone from 0 to around 20% over the past few years.

Returning to my personal view, what I agree with most is that AMMs indeed offer an opportunity for ordinary people who are willing to hold inventory to make markets, because if you are willing to hold a certain asset, you can indeed conduct a market-making strategy without hedging.

Beyond that, Hayden proposes a third advantage: the benefit of being an LP in a paired stock token pair is greater than simply holding the stock. On this point, I ran a simple backtest with NVDA and SPY. Based on historical data over the past three years, the approximate impermanent loss at the end of the period was -10.8% of the principal. Without reinvesting fees, you would need roughly 11.5% per year in fees to make up for this loss at the end of the period.

Looking at the current on-chain liquidity and fee levels, that is difficult, because tokenized stocks generate little to no fees outside of trading hours, and there are very few SPY/NVDA pairs. Even where they exist, the APY is very low. This can only be resolved over time.

Finally, compared with the paired LP strategy, what I do on-chain myself is strategies like USDC/NVDA and USDC/CRCL. The core idea is to use a fee-based grid framework combined with traditional financial valuation models to run LP strategies, converting traditional holding-based returns into a hybrid return of holding + fees + grid-based high-sell-low-buy.

This return model is still being validated in live trading, and the cycle may be relatively long. However, I believe AMM market-making strategies truly fill a gap in current tokenized stock market-making. The core theory remains — if you want to hold a position, your market-making cost can be incurred without hedging. But in actual execution, you must account for impermanent loss, which is a blank space still waiting to be explored.