This is a due-diligence guide, not a yield chase or momentum list. The point is to give Radar readers a repeatable way to compare protocols before surface growth, points farming, or TVL momentum starts doing too much of the talking.
Borrow demand can make a protocol look healthier than it really is. For Radar, the better question is whether the protocol still looks coherent when incentives shrink, volatility rises, and liquidations have to clear through real market depth.
Core Comparison Criteria
- Collateral quality: The cleaner the collateral mix, the easier it is to understand how the lending market behaves during a drawdown.
- Liquidation path: A protocol with clear, executable liquidation pathways is easier to trust than one that assumes someone else will provide exit liquidity.
- Governance clarity: If changes to caps, listings, or emergency settings are hard to follow, operational risk rises even when the product looks polished.
- User durability: Repeat borrowers and lenders matter more than temporary farming bursts when you are judging staying power.
Red Flags
- Borrow demand that exists mainly because looping and incentives make the number look pretty.
- Listings that outpace governance, risk review, or liquidation infrastructure.
- A surface where one asset or one strategy explains most of the activity.
- A protocol whose communications make stress events look simpler than they really are.
Decision Loop
- Start with collateral mix and ask what actually backs the borrowed dollars.
- Move to liquidation design and imagine the market under a fast 15% drawdown.
- Check governance pace, not just governance branding.
- Then decide whether usage looks durable enough to survive without promotional support.
A good comparison framework slows you down in the right places. If the protocol still looks attractive after these checks, then the interest is more likely to be durable instead of purely cosmetic.
Continue this cluster
Stay inside the same cluster so the logic compounds instead of resetting on the next click.
