Yearn Finance is a yield aggregator: users deposit a token into a vault, and the vault deploys it into one or more strategies that earn yield elsewhere in DeFi, compounding returns and spreading gas costs across every depositor. Building something similar is less about novel math and more about strategy risk management, accounting and operational discipline.
How a Yearn-style vault works
A vault is a pooled account with a share token. When you deposit 1,000 USDC, you receive shares priced at the vault's current share value. As strategies earn yield, the total assets per share rise, so redeeming the same shares later returns more USDC. Nothing is distributed as separate reward tokens; gains show up in the share price.
Yearn's v3 vaults implement ERC-4626, the tokenized vault standard. That standard defines deposit, mint, withdraw, redeem and conversion functions, which means any wallet, lending market or aggregator that understands ERC-4626 can integrate a vault without custom code. If you build a yield product today, using ERC-4626 is close to mandatory.
Strategies and allocation
The vault itself does not farm. It lends capital to strategies, each a separate contract with a single job: supply USDC to a lending market, provide liquidity to a stable pool and stake the LP tokens, or loop a liquid-staking token. In v3, strategies can themselves be ERC-4626 "tokenized strategies," and a vault's allocator decides how much debt each strategy receives. This separation lets a team add or retire strategies without migrating depositors.
Harvests and reporting
Periodically, a keeper calls a report or harvest function. The strategy claims reward tokens, swaps them back into the underlying asset, and reports profit or loss to the vault. To prevent someone from depositing right before a harvest and capturing yield they did not earn, profits are unlocked gradually over a set period. Performance and management fees are taken at reporting time, usually by minting shares to the fee recipient.
Where yield protocols get hurt
Yearn has been around since 2020 and has had incidents, including an exploit of a DAI vault in 2021 and a 2023 attack on a misconfigured legacy contract. The broader category shows recurring failure modes:
- Price manipulation in strategies. A strategy that values its position using a spot pool price can be tricked with a flash loan into reporting a false profit or loss.
- Share inflation on empty vaults. The first depositor can donate assets to inflate the share price and round later depositors down to zero. Mitigations include virtual shares and offsets, or seeding the vault at deployment.
- Underlying protocol failure. Your strategy inherits every risk of the protocols it touches: depegs, oracle failures, governance attacks, bridge hacks.
- Swap slippage on harvests. Selling reward tokens through a thin pool leaks value to MEV bots. Use private submission or aggregator routing with tight bounds.
- Stale legacy contracts. Old vaults with user funds remain attack surface long after the team moves on. Have a deprecation plan.
Design decisions you will face
| Decision | Options | Trade-off |
|---|---|---|
| Vault standard | ERC-4626 vs custom | 4626 gives instant composability; custom rarely justifies the integration cost |
| Strategy count per vault | Single vs multi-strategy | Single is easier to audit; multi diversifies but needs an allocator and debt limits |
| Allocator | Governance, a role-based manager, or automated | Automated chasing of the highest APY can pile into the riskiest venue |
| Withdrawal liquidity | Idle buffer vs instant unwind | Buffers drag returns; instant unwinds can fail or slip under stress |
| Keepers | Own bots vs a keeper network | Own bots are simpler; networks remove a single point of failure |
Building it, step by step
- Define the risk mandate. Which assets, which protocols, maximum exposure per protocol, and what losses the vault can absorb. Publish it.
- Start from audited base code. Yearn's v3 vault and tokenized-strategy code is open source; check its license and build strategies on top of it rather than writing vault accounting from scratch.
- Write and fork-test each strategy. Test against mainnet forks through deposit, harvest, loss and emergency exit. Include depeg and oracle-failure scenarios.
- Set roles carefully. Separate who can add strategies, who can change debt and who can shut down. Use a multisig with a timelock for anything that can move funds.
- Audit, then launch with caps. Deposit limits that rise over time limit the blast radius of any bug that slips through.
- Monitor continuously. Alerts on share price drops, strategy losses, unusual withdrawals and changes in the protocols you depend on.
Most of the code you need overlaps with yield farming and staking platforms; the difference is the automated allocation layer. If your strategies lend into markets, read how Aave's pool and liquidation design works, because liquidations and utilization spikes directly affect your ability to withdraw.
Frequently asked questions
What is the difference between a yield aggregator and a yield farm?
A farm pays rewards for providing liquidity to its own protocol. An aggregator moves deposited capital into other protocols' opportunities and compounds the results for its users.
Do I need my own token?
No. Yearn's YFI is a governance token; the vaults work without it. Launch the vaults first, and only add a token if governance genuinely needs decentralizing.
Are vault returns guaranteed?
No. Displayed APYs are usually trailing figures. Strategies can lose money, and a vault can realize losses that reduce the share price.
Is ERC-4626 safe by itself?
It is an interface, not a security guarantee. Rounding direction, inflation attacks and how totalAssets is calculated are still your responsibility.