DeFi vs TradFi Yields: What I Actually Found
January 13, 2026
“DeFi pays more than TradFi.”
I’d heard this a hundred times but never actually checked for myself.
So I built a dashboard from scratch to find out.
What I Compared
For DeFi, I looked at Aave, Compound, and Morpho. Largest protocols. Most battle-tested on Ethereum. Stuck to USDC and USDT, the closest thing to a cash equivalent.
Also threw in Morpho’s Steakhouse vault since it’s designed for institutions. Curious if a more managed approach looked any different.
For the TradFi benchmark, I went with 3-Month T-Bills and the Fed Funds rate.
Why T-Bills over Fed Funds as the default? Duration felt more comparable. You’re not lending overnight in DeFi. You’re parking liquidity that sits there.
Pulled 18 months of data to capture both the Fed’s hiking cycle and the cuts that followed. Enough to see how DeFi yields respond across different macro environments.
My Hypothesis
DeFi should pay a consistent “risk premium” over risk-free rates. Maybe 100-200 bps to compensate for smart contract risk.
That’s what I expected to find.
That’s not what the data showed.
What the Data Actually Showed
Right now the spread is basically zero.
Aave USDC yields 3.60%. T-Bills pay 3.55%. That’s 5 basis points.
Not a premium. Noise.
Morpho’s Steakhouse vault does a bit better at 3.93%, about 35 bps over Aave with lower volatility. The curated approach seems to smooth things out.
But still not the juicy premium I expected.
So I dug into the historical data. That’s when things got weird.
December 2024: Aave USDC spiked to 41.4% APY. T-Bills were sitting at 4.2%.
That’s a 37 percentage point spread.
And this wasn’t the only instance. July 2023 briefly touched 56%.
This wasn’t a stable risk premium. This was something else entirely.
What Made It Click
The volatility didn’t make sense until I added utilization data.
I built a separate chart tracking borrow utilization for Aave and Compound. Once I could see utilization plotted alongside yields, the pattern clicked.
DeFi lending protocols use interest rate curves with “kink” points. Aave’s kink sits around 90% utilization.
Below the kink, rates rise gradually. Above the kink, rates spike aggressively.
The mechanism is intentional. When utilization gets too high, liquidity dries up. The rate spike incentivizes new deposits and discourages borrowing. It protects the protocol from a liquidity crunch.
That 41% yield in December 2024? Utilization was pushing past the kink during a crypto rally. Borrowing demand was through the roof. Lenders who happened to be in the pool got paid handsomely.
Once I understood the kink mechanism, everything made sense.
DeFi yields aren’t a function of smart contract risk. They’re a function of borrowing demand at any given moment.
The Insight
“DeFi pays more” isn’t wrong. It’s just incomplete.
You’re not earning a steady premium for holding protocol risk. You’re earning irregularly. You get paid when borrowing demand spikes and utilization crosses the kink.
It’s not fixed income. It’s optionality on leverage demand.
For passive allocators looking for a T-Bill replacement? Probably not worth it. A 35 bps premium doesn’t compensate for smart contract, oracle, and governance risk.
For active managers who can monitor utilization in real-time? Maybe there’s something here. December 2024 offered a 37 percentage point spread that lasted for a bit. If you can time those windows, the math changes.
Where I Might Be Wrong
I’m only looking at supply-side yields. Not accounting for gas costs, withdrawal timing, or tail risks like exploits.
18 months might not capture a full market cycle. There are protocols I didn’t include that might tell a different story.
Still forming my views on this. Would love to hear what I’m missing.
Dashboard
Full dashboard here: defi-trad-yields.vercel.app
Tracks yields, utilization, and TVL across protocols vs Fed Funds and T-Bills.