Hook
July 2024. The average price of the top 10 L2 tokens dropped 8.2% month-over-month. June’s decline was 3.5%. The deceleration is accelerating.
This is not a correction. It’s a structural repricing of liquidity risk.
I’ve been watching this decay since May. The Dencun upgrade in March created a temporary demand spike—gas fees dropped, TVL spiked, and everyone cheered. But the underlying supply dynamics were already toxic. The July data confirms what I saw in the order books: smart money is rotating out of L2 tokens into BTC and ETH.
We don’t trade narratives. We trade liquidity. And right now, L2 liquidity is evaporating faster than a Turkish lira trade in 2018.
Context
Layer-2 scaling solutions (Arbitrum, Optimism, Base, zkSync, StarkNet, etc.) were supposed to be the future of Ethereum. Lower fees, higher throughput, same security. The hype cycle peaked in 2022-2023. Total value locked across L2s hit $40B in early 2024. But the token prices have been in a downtrend since March.
Why? The same reason China’s new-home prices are falling: supply is overwhelming demand, and the demand that exists is fake—subsidized by incentives.
Let’s break down the numbers.
Supply-side (L2 tokens): - Circulating supply of the top 5 L2 tokens has increased 35% since January 2024 due to vesting unlocks and inflation rewards. - Hidden supply: tokens allocated to teams, investors, and ecosystem funds that are not yet in circulation but will be dumped over the next 12 months. This is the equivalent of China’s undeveloped land parcels—a time bomb.
Demand-side: - Daily active addresses on L2s peaked in March at 2.5 million. By July, that number dropped to 1.6 million. - TVL in L2 protocols fell from $40B to $28B—a 30% decline. - Fee revenue (the real measure of economic activity) is down 55% from March highs.
The market is pricing in a future where L2s become commodity infrastructure with thin margins. The token prices are reflecting that reality faster than the VCs are willing to admit.
Core: The Supply-Demand Death Spiral
I’ve been analyzing this through the same framework I used for the 2022 Terra collapse. The mechanics are similar—just slower.
1. Inventory (Unlocks) Are Out of Control
The total fully diluted valuation (FDV) of the top 10 L2 tokens is $150B. The current market cap is $18B. That’s an 8x dilution factor. Every month, another $1.2B worth of tokens are unlocked—some to team members who have been waiting for lockup expiry, some to VCs who need to show returns to their LPs.
In July, unlocks accelerated. Why? Because many projects started their unlock schedules in Q2 2024. The supply pressure is not linear—it’s front-loaded.
Compare this to China’s housing: the “visible inventory” (completed homes) is 20 months, but the “hidden inventory” (land already sold but not yet built) is another 30 months. L2 tokens have the same problem. The hidden supply (team tokens, investor tokens, future emissions) is 2-3x the current circulating supply.
2. Demand Is Fading—And It’s Fake
The demand for L2 tokens is almost entirely driven by yield farming and airdrop speculation. Real economic activity (transaction fees, DeFi lending volumes) is a small fraction.
In July, the average transaction fee on Optimism dropped to $0.02. That’s great for users, but it means the protocol earns almost nothing. The token’s value proposition becomes: “we are a speculative vehicle that happens to process transactions.”
I saw this exact pattern in 2020 DeFi farming. I ran a $200K position into $850K by riding yield curves. But when the incentives stopped, so did the users. The real APR was negative—you were just subsidizing the protocol’s TVL.
Same here. The yield on L2 tokens is the rent you pay for holding someone else’s risk.
3. The Policy Pulse Faded
In March, the Dencun upgrade (EIP-4844) reduced L2 data posting costs by 90%. That was the “policy pulse” analogous to China’s May 2024 housing stimulus. It created a temporary demand spike. But the effect lasted only 6 weeks. By May, monthly active users on L2s were already declining.
Smart money doesn’t wait for confirmation. They sold into the spike. Retail got caught holding the bag.
4. The Real Price Is Worse Than the Index
The official price indices (CoinGecko, CoinMarketCap) are based on exchange prices. But they don’t capture the over-the-counter (OTC) market where large blocks of tokens trade at discounts. In July, I saw OTC bids for ARB tokens at 15% below market price. That’s a sign of distressed selling.
Just like China’s official new-home price index—which is higher than the real transaction price because of quality mix adjustments—the crypto indices smooth out the pain. If you look at the actual realized prices (on-chain average cost basis), the current market price is below the average buy price of most holders.
5. The Contrarian Angle: Retail Holds, Smart Money Exits
On-chain data shows that addresses holding less than $10K of L2 tokens increased their holdings by 5% in July. Meanwhile, addresses holding more than $1M decreased their holdings by 12%.
Retail is buying the dip. Smart money is distributing.
Why? Because retail believes the narrative: “L2s are the future, scaling will bring mass adoption.” But the data shows that without incentives, most L2 users are bots. The real adoption is happening on Ethereum L1 and Bitcoin.
I learned this lesson in 2017 when I shorted ICO tokens. The narrative is a sales tool, not a valuation model. The technology might be revolutionary, but the token is a security. And right now, the security is overpriced relative to its cash flows.
6. The Systemic Risk: Liquidations
Many L2 tokens are used as collateral in DeFi lending protocols. As prices fall, collateral values drop, triggering margin calls. This creates a downward spiral. In July, the total liquidations on Aave and Compound for L2 tokens reached $50M—a 300% increase from June.
If the price drops another 20%, the next wave of liquidations will be 3x larger. This is the same dynamic that killed Terra in 2022. The death spiral is not a linear process. It accelerates.
I reverse-engineered the Terra collapse in 2022. I saw the exact same decay rates in the oracle data. The L2 market is not there yet, but the trajectory is similar. The only difference is that L2 tokens have a more distributed ownership, so the crash might be slower. But the direction is the same.
Contrarian: The Bear Case Most People Miss
Conventional wisdom says: “L2s are undervalued because they have high FDV, but once the unlocks are absorbed, the price will recover.”
That’s a trap.
Here’s what most analysts miss: the demand side is not just low—it’s structurally declining. The fee revenue is not enough to support the current valuation. Even if all tokens are unlocked, the price will not recover if the fee revenue doesn’t grow.
Think about it: if Arbitrum earns $5M in fees per month, and its FDV is $10B, that’s a P/E ratio of 166x. That’s more expensive than Nvidia. And the fee revenue is declining.
Smart money doesn’t hold assets with declining earnings. They sell.
Additionally, the rise of alternative L1s (Solana, Sui, Aptos) is eating L2 market share. Solana’s daily active addresses are 3x higher than Ethereum’s top L2. The narrative that “L2s are the only scaling solution” is being challenged by monolithic chains that offer similar performance without the complexity.
In 2025, I led the development of an AI trading agent. We quickly learned that human intuition is still superior for setting initial parameters. The same applies to L2s: the technology is impressive, but the market’s judgment is that the token economics are broken.
Takeaway: Actionable Levels
The L2 token market is pricing in a structural decline. The key levels to watch: - For ARB: $0.50 (realized cap support). If it breaks, the next stop is $0.30. - For OP: $1.00 (2023 low). Breaking below that would confirm a death spiral. - For L2 sector overall: total market cap of $15B (current $18B). If it drops below $15B, the next wave of liquidations will hit.
We don’t trade narratives. We trade liquidity. And right now, liquidity is flowing out of L2s faster than the VCs can sell.
The question is not whether the technology will survive. It’s whether the token holders will.
And based on the data, I’d say the answer is no.