Sommelier x Ymax | March 2026
We backtested stablecoin yield across 5 chains (Ethereum, Base, Arbitrum, Avalanche, Plasma) and multiple asset types (USDC, USDT, sUSDe, Pendle SY) from January 2024 to March 2026, covering 34 pools across 11 protocols including Pendle's underlying yield markets. Three findings emerge:
1. No single chain consistently offers the best yield. The optimal chain rotates quarterly. Ethereum earned 14.35% annualized (best single-chain), Arbitrum earned 13.02%, Base only 9.23%. The gap between best and worst major chain choices is 512 basis points.
2. Multi-asset allocation is as important as multi-chain. Three asset types each dominated different periods: USDC (57% of days), Ethena sUSDe (25%), and Pendle underlying yield (18%). A USDC-only strategy misses 43% of the yield opportunity.
3. The yield frontier is shifting toward structured products. Pendle SY pools dominated Q4 2025 and Q1 2026, displacing both traditional lending (Aave/Compound) and Ethena as the best yield source. This trend is accelerating.
A cross-chain, multi-asset strategy that rebalances weekly captures 14.24% annualized after bridge costs -- within 12bps of the best single-chain return, but without requiring foresight about which chain, protocol, or asset type will dominate.
- Period: January 1, 2024 - March 27, 2026 (817 days)
- Data source: DefiLlama Yields API (free, no auth, daily granularity)
- Initial capital: $100,000
- Assets: USDC, USDT, sUSDe (Ethena), sUSDS (Sky/Maker), Pendle SY (underlying yield), USDe, USDT0
- Protocols: Aave V3, Compound V3, Fluid, SparkLend, Euler V2, Dolomite, Sky, Ethena, Pendle, Moonwell, Spark
- Chains: Ethereum (15 pools), Base (5 pools), Arbitrum (10 pools), Avalanche (2 pools), Plasma (5 pools)
- Compounding: Daily (
value *= (1 + bestDailyApy / 365)) - Rebalancing: Weekly (every 7 days, move to best yield across all chains/assets; between rebalances, stay on current chain's best pool)
- Bridge cost assumption: 5 basis points per cross-chain move
For each day, we identified the best available stablecoin yield on each chain (across all protocols and asset types with data), then simulated three strategies.
| Chain | Annualized APY | Final Value ($100K) | vs Best |
|---|---|---|---|
| Ethereum | 14.35% | $135,019 | -- |
| Arbitrum | 13.02% | $131,521 | -133 bps |
| Base | 9.23% | $121,854 | -512 bps |
| Avalanche | 5.38% | $112,435 | -897 bps |
| Plasma | 3.65% | $101,796 | -1070 bps |
Ethereum takes the top single-chain spot driven by Ethena sUSDe and Pendle SY yields. Arbitrum is second (Dolomite USDC in 2024). Base, Avalanche, and Plasma trail significantly.
| Strategy | Annualized APY | Final Value | Chain Switches | vs Best Single-Chain |
|---|---|---|---|---|
| Cross-chain (no costs) | 15.60% | $138,327 | 53 | +125 bps |
| Cross-chain (5bps costs) | 14.24% | $134,708 | 53 | -12 bps |
Cross-chain with costs is within 12bps of the best single-chain -- but the best single-chain is only knowable in hindsight.
| Quarter | Best Chain | Avg APY | Top Protocol | Best Asset | What Drove It |
|---|---|---|---|---|---|
| Q1 2024 | Arbitrum | 24.94% | Dolomite | USDC | Arbitrum incentive season |
| Q2 2024 | Ethereum | 20.06% | Ethena | sUSDe | Ethena yield spike |
| Q3 2024 | Ethereum | 10.84% | Fluid | USDC | Fluid mainnet launch |
| Q4 2024 | Ethereum | 18.54% | Ethena | sUSDe | Second Ethena spike |
| Q1 2025 | Arbitrum | 12.74% | Dolomite | USDC | Dolomite resurgence |
| Q2 2025 | Arbitrum | 7.12% | Dolomite | USDC | Quiet period |
| Q3 2025 | Arbitrum | 9.75% | Dolomite | USDC | Moderate yields |
| Q4 2025 | Ethereum | 10.75% | Pendle | Pendle | Fixed-rate premium expansion |
| Q1 2026 | Ethereum | 13.56% | Pendle | Pendle | Pendle becomes yield leader |
5 lead changes across 9 quarters. Three distinct eras: Dolomite/Arbitrum (Q1 2024), Ethena/Ethereum (Q2+Q4 2024), and Pendle/Ethereum (Q4 2025+). The yield frontier continuously shifts to new protocols and asset types.
| Asset Type | Days as Best | Share | When It Mattered |
|---|---|---|---|
| USDC | 466 | 57.0% | Most of 2024-2025 -- the baseline |
| sUSDe | 203 | 24.8% | Q2 2024, Q4 2024 (Ethena yield spikes >20%) |
| Pendle | 145 | 17.7% | Q4 2025, Q1 2026 (structured yield premium) |
| USDT | 3 | 0.4% | Brief moments |
Three distinct asset types dominated three distinct periods. USDC is the workhorse, but Ethena sUSDe captured Q2 and Q4 2024 (20%+ yields), and Pendle took over in Q4 2025 through Q1 2026 as fixed-rate premiums expanded. A USDC-only strategy misses 43% of the yield opportunity.
The emergence of Pendle as the current yield leader is particularly notable -- it signals that the market is shifting toward structured/fixed-rate products, and any yield management system needs to adapt to new asset types as they emerge.
| Protocol | Chain | Days as Best | Share |
|---|---|---|---|
| Ethena | Ethereum | 203 | 24.8% |
| Dolomite | Arbitrum | 201 | 24.6% |
| Pendle | Ethereum | 144 | 17.6% |
| Fluid | Ethereum | 89 | 10.9% |
| Moonwell | Base | 60 | 7.3% |
| SparkLend | Ethereum | 31 | 3.8% |
| Fluid | Arbitrum | 31 | 3.8% |
| Aave V3 | Avalanche | 13 | 1.6% |
| Others | Various | 45 | 5.6% |
Key observations:
- Ethena (24.8%), Dolomite (24.6%), and Pendle (17.6%) are the big three -- each dominated a different era
- Pendle is the newest entrant and current leader, taking over in Q4 2025. Its share will likely grow.
- Fluid was the most versatile protocol, appearing as best on 3 chains (10.9% + 3.8% + 1.0% = 15.7%)
- Aave V3 (the largest protocol by TVL) was rarely the best yield -- only 1.3% of days across all chains. Size and best yield are inversely correlated.
- Avalanche contributed 13 days (1.6%) via Aave V3 -- meaningful at the margin
- Plasma did not contribute best-yield days in this period (too new)
- 17 distinct protocol/chain combos contributed, showing a deeply fragmented yield landscape
-
Eliminates chain selection risk. The gap between Ethereum (14.35%) and Base (9.23%) is 512 basis points. Between Ethereum and Avalanche, it's 897 bps. Cross-chain allocation neutralizes this risk.
-
Captures asset rotation. Three asset types each dominated different periods: USDC (57%), sUSDe (25%), Pendle (18%). A USDC-only strategy misses 43% of yield opportunity.
-
The yield frontier continuously shifts. In 27 months, three distinct yield eras emerged: Dolomite/USDC on Arbitrum (Q1 2024), Ethena sUSDe on Ethereum (Q2-Q4 2024), and Pendle SY on Ethereum (Q4 2025+). Each era was unforeseeable from the previous one. An automated system monitoring all protocols, assets, and chains is the only way to consistently capture the yield frontier.
-
Structured products are the future. Pendle went from zero to the current yield leader in under 6 months. As fixed-rate and structured yield products mature, they will increasingly outperform traditional variable-rate lending. A system that can only access Aave/Compound is already falling behind.
-
Current conditions understate future cross-chain value. Plasma and Avalanche barely registered. With Tempo (Stripe-backed), Arc, and renewed Solana competition entering, the number of yield-competitive chains will increase. Each new competitive chain adds rebalancing opportunities.
-
New asset types will keep emerging. sUSDe didn't exist before Feb 2024. Pendle SY pools didn't exist before late 2025. The next yield primitive will create temporary spikes that only a multi-asset, multi-chain system can capture.
- Cross-chain yield management with confinement -- agents rebalance across chains within Ymax's delegation system, without custodying user funds
- Multi-asset support -- not locked into USDC; can rotate into sUSDe, sUSDS, or whatever yields best
- Direct lending pool access -- bypass curator vaults and their fees
- Protocol-agnostic -- supports any protocol on any chain; as the backtest shows, the best protocol rotates constantly
- Not a guarantee of future performance
- Bridge costs are simplified. We model a flat 5bps per cross-chain move. In practice: CCTP (USDC native bridging) is roughly free but requires the position to be in USDC; moving from sUSDe or Pendle positions to another chain requires swapping to USDC first (DEX slippage + gas), then CCTP, then potentially swapping back. Some routes (e.g., to/from Plasma, Avalanche) may not have cheap bridge solutions. A production system would need route-dependent, asset-dependent cost modeling.
- Assumes perfect execution (no slippage, no failed txns, no depegs)
- Weekly rebalancing is a simplification; real strategies may adjust frequency
- sUSDe, Pendle SY, and sUSDS carry different risk profiles than USDC -- the analysis treats all stablecoins as equivalent for yield comparison
- Pendle SY yields represent the underlying variable rate, not the PT fixed rate
All yield data sourced from DefiLlama Yields API (yields.llama.fi). Free, no authentication required, daily granularity. Pool IDs are stable UUIDs tied to specific protocol/chain/asset combinations.
npx tsx scripts/cross-chain-backtest.tsRaw results saved to scripts/output/backtest-results-v4.json.
| Protocol | Asset | Chain | Data From | Data To | Days |
|---|---|---|---|---|---|
| Aave V3 | USDC | Ethereum | 2024-01 | 2026-03 | 817 |
| Compound V3 | USDC | Ethereum | 2024-01 | 2026-03 | 815 |
| Fluid | USDC | Ethereum | 2024-05 | 2026-03 | 667 |
| Euler V2 | USDC | Ethereum | 2025-11 | 2026-03 | 141 |
| Dolomite | USDC | Ethereum | 2025-07 | 2026-03 | 269 |
| SparkLend | USDC | Ethereum | 2024-01 | 2026-03 | 815 |
| Aave V3 | USDT | Ethereum | 2024-01 | 2026-03 | 817 |
| Sky | sUSDS | Ethereum | 2025-02 | 2026-03 | 400 |
| Ethena | sUSDe | Ethereum | 2024-02 | 2026-03 | 771 |
| Pendle | sUSDe-SY | Ethereum | 2026-01 | 2026-03 | 78 |
| Pendle | srUSDe-SY | Ethereum | 2025-12 | 2026-03 | 87 |
| Pendle | coreUSDC-SY | Ethereum | 2025-10 | 2026-03 | 138 |
| Pendle | jrUSDe-SY | Ethereum | 2025-12 | 2026-03 | 87 |
| Pendle | reUSDe-SY | Ethereum | 2025-12 | 2026-03 | 103 |
| Aave V3 | USDC | Base | 2024-03 | 2026-03 | 748 |
| Compound V3 | USDC | Base | 2024-03 | 2026-03 | 729 |
| Fluid | USDC | Base | 2024-10 | 2026-03 | 529 |
| Moonwell | USDC | Base | 2024-01 | 2026-03 | 769 |
| Aave V3 | USDT | Base | 2024-01 | 2025-03 | 450 |
| Aave V3 | USDC | Arbitrum | 2024-01 | 2026-03 | 817 |
| Compound V3 | USDC | Arbitrum | 2024-01 | 2026-03 | 815 |
| Fluid | USDC | Arbitrum | 2024-10 | 2026-03 | 529 |
| Euler V2 | USDC | Arbitrum | 2026-02 | 2026-03 | 41 |
| Dolomite | USDC | Arbitrum | 2024-01 | 2026-03 | 751 |
| Sky | sUSDS | Arbitrum | 2026-02 | 2026-03 | 43 |
| Pendle | sUSDai-SY | Arbitrum | 2026-02 | 2026-03 | 43 |
| Aave V3 | USDC | Avalanche | 2024-01 | 2026-03 | 665 |
| Spark | USDC | Avalanche | 2025-12 | 2026-03 | 91 |
| Aave V3 | USDT0 | Plasma | 2025-09 | 2026-03 | 181 |
| Aave V3 | USDe | Plasma | 2025-09 | 2026-03 | 181 |
| Aave V3 | sUSDe | Plasma | 2025-09 | 2026-03 | 181 |
| Pendle | sUSDe-SY | Plasma | 2026-03 | 2026-03 | 22 |
| Pendle | USDT0-SY | Plasma | 2026-03 | 2026-03 | 22 |