The longer I spend providing liquidity, the less I understand how LP fees got marketed as "passive income."
Most people see a juicy fee APR and call it a day. The problem is the APR isn't fighting reality. Reality is IL. Reality is LVR. Reality is waking up after a volatile week and realizing the fees barely covered what arbitrageurs extracted from the pool.
I've seen plenty of positions that looked great on paper and still got smoked by simply holding the assets.
Fixed-fee AMMs never made much sense to me. Markets can go from completely dead to absolute chaos in a few hours, yet the pool keeps charging the same fee as if nothing changed. Risk moves. Pricing doesn't.
Feels backwards.
When volatility starts ripping, LPs should be getting paid more for taking that risk. When things calm down, tighter fees make sense because traders get better execution and flow stays in the pool. The goal isn't squeezing every last basis point out of traders. It's stopping LPs from constantly donating value to MEV and arbitrage bots.
What interests me isn't dynamic fees by themselves, it's whether they're adjusting early enough.
Once volatility hits, the damage is already happening. The interesting question is whether models can spot changing conditions before the pool gets picked apart. If fees can move ahead of volatility instead of reacting afterward, the economics start looking very different for LPers.
That's where OpenGradient caught my attention.
Not because crypto suddenly discovered another AI narrative. We've had enough of those.
If models are influencing fee levels, liquidity allocation, risk parameters, basically anything that affects where capital flows, I want to know what happened under the hood. "Trust us" isn't a serious answer when real money is involved.
DeFi produces an absurd amount of data every second. Most of it gets ignored.