How Rosetta Beat Every Vault After $100M in Volume
Most DeFi users pick a vault and forget it.
It's the obvious play. Find a decent APY, deposit, and move on with your life.
Here's the problem: yield rates don't sit still.
That 8% vault you deposited into last week? It was 4% yesterday. The 5% vault you ignored? It spiked to 15% while you were sleeping.
I watched this happen for months. The same pattern, over and over. Users leaving yield on the table—not because they're lazy, but because tracking five vaults across different dashboards, comparing rates that change every block, and manually rebalancing at 3am isn't realistic.
So we built something that does it automatically.
72 days and $100 million later, the results are in.
The Numbers
On-chain. Verifiable. Publicly auditable.
How We Beat Every Vault
We went head-to-head with the major USDT0 vaults on Hyperliquid:
| Vault | Their APY | Rosetta Premium |
|---|---|---|
| Rosetta | 9.01% | Baseline |
| MEV Capital | 8.49% | +6% |
| Hyperithm | 7.43% | +21% |
| Felix Frontier | 6.24% | +44% |
| Gauntlet | 5.07% | +78% |
| Felix | 4.07% | +121% |
These are solid protocols run by smart teams.
We still outperformed all of them.
The reason isn't magic. It's math.
Block-Level Precision
(This Is Where It Gets Technical)
Most DeFi data infrastructure is event-based. State only updates when transactions occur.
But interest accrues every block.
A lending market with no transactions still has:
- Borrowers owing more
- Lenders entitled to more
- Utilization shifting
- Rates adjusting
Event-based systems miss all of this. They're showing you what the rate was, not what it is.
Rosetta treats each block as the atomic unit of truth.
Every block, our indexer:
- Accrues interest forward—computes what borrowers owe and lenders earn right now
- Recalculates utilization based on accrued interest
- Runs each protocol's interest rate model to determine current rates
- Updates vault positions and price-per-share
- Stores snapshots at block granularity
On HyperEVM with ~1 second block times, this means near real-time accuracy.
The yield you see is the yield you get.
The Yield Router Architecture
Here's how the system actually works:
Smart Accounts (User-Controlled)
You operate through a sub-account you control. Rosetta orchestrates deposits and rebalances within it, but withdrawals can only go back to your main wallet.
This is enforced at the contract level. Even if our systems were compromised, funds could only return to you.
On-Chain Policies
You define which vaults and assets the router can access. These rules are stored on-chain and enforced programmatically. Changing them requires your signature.
Execution Engine
Monitors yields via the indexer. Evaluates rebalancing opportunities. Executes allocations through your smart account.
When a better opportunity appears, we move—automatically.
What Actually Happened: Month by Month
November 2025: Launch
Markets were volatile. MEV Capital spiked to 50%+ on November 14th. We caught it. Hyperithm surged days later. We rotated within minutes.
December 2025: Compression
Markets cooled. Yields compressed everywhere. Our average: 7.36% APY. Still #1. Still 16% above the next best vault.
January 2026: Consistency
Three months. Three different market conditions. Same result.
The Math For Your Portfolio
If you're manually farming yields on Hyperliquid:
Best single vault (MEV Capital): 8.49% average
Rosetta: 9.01% average
Your premium for doing nothing: 6%
On a $100,000 position over a year: ~$500 in additional yield.
On $1M: $5,000+.
For depositing once instead of checking rates daily.
Why Manual Yield Farming Fails
Three reasons:
1. Time
The best rate at 9am isn't the best rate at 9pm. Markets move. If you're not watching, you're losing.
2. Gas
Manual rebalancing costs money. Time it wrong, and transaction fees eat your gains.
3. Speed
Even if you spot the opportunity—can you act fast enough? Most can't.
Rosetta eliminates all three.
You set policies. We execute within them. Gas is sponsored—the APY you see is the APY you earn.
The Bottom Line
DeFi promised efficient markets.
For yield optimization, it delivered fragmented, manual work.
$100M in volume. 9.01% APY. Consistently outperformed every vault.
That's 72 days.
If you're still manually chasing yield across five different vaults, ask yourself:
Is that really the best use of your time?
Past performance doesn't guarantee future results. DeFi involves risk. Verify on-chain.