Kava USDX: one toxic collateral leg

Kava Mint let users lock crypto collateral and draw USDX against it as overcollateralized debt, with liquidations retiring positions that fell below their required ratio. The accepted-collateral set included TerraUSD (UST) alongside KAVA, ATOM, wrapped Bitcoin, and Ether. That whitelist was a governance decision, and it is where the failure originated.

When UST collapsed in May 2022, the USDX positions it backed lost their collateral value almost overnight. Liquidations fired to defend the system, but they were selling a collapsing asset into a market that no longer wanted it, so they recovered little and did nothing to reassure holders. USDX broke to about $0.47, bounced to roughly $0.89, and slid back near $0.56 — the signature of a peg being propped up rather than structurally restored.

Kava governance voted UST out of Kava Mint and the team predicted a clean repeg, on the logic that USDX was collateralized rather than algorithmic. The prediction was wrong. The mechanism was indeed sound; the collateral list was not, and the loss of confidence outlived the bad asset that caused it.

A CDP stablecoin does not choose its risk by mechanism alone — it chooses it by what it accepts as collateral. Every asset added to the whitelist is a potential failure path, because the stablecoin inherits that asset's worst-case behavior the moment it is locked against new debt. Adding UST imported the full fragility of an uncollateralized algorithmic peg into a system whose whole pitch was that it was not algorithmic.

This is the part the headline design obscures. USDX's own contract behaved exactly as intended: positions that breached their ratio were liquidated. But correctness of the liquidation engine is irrelevant if the collateral it is unwinding has no floor. The whitelist, not the code, set the true risk surface, and a single entry on it was enough to be terminal.

Pharos treats collateral composition as a first-class risk input rather than reading overcollateralization as a uniform strength. A CDP coin's resilience depends on the quality and correlation of what backs it, which is why collateral quality and dependency exposure are scored separately from the nominal collateral ratio.

Overcollateralization protects against ordinary price moves: if collateral drifts down a few tens of percent, liquidations close positions while there is still surplus value to cover the debt. It is a buffer sized for volatility, not for a collateral asset going to near-zero faster than the system can unwind it.

UST did not drift; it gapped. The collateral backing a slice of USDX lost more than 90% of its value in days, blowing through any reasonable liquidation ratio before the auctions could clear. When a buffer is overwhelmed by a discontinuous move in the collateral itself, the surplus that was supposed to make the stablecoin whole simply is not there — and the liquidation process converts a collateral failure directly into a peg failure.

The deeper lesson is that overcollateralization and collateral quality are different risks. A position can be 150% collateralized and still be unsafe if that 150% is denominated in an asset capable of becoming worthless. The ratio measures how much collateral there is; it says nothing about whether that collateral will still exist tomorrow.

Acute depegs usually resolve one of two ways: the coin recovers to par, or it is wound down. USDX did neither cleanly. The toxic collateral was removed and the contagion stopped, but the peg never came back, and the token was not retired. It settled into a long half-life as a discounted legacy claim.

By later years USDX traded persistently below a dollar, with a recorded all-time low near $0.10 and a market value that drifted down with its credibility. As of mid-2026 it changes hands around $0.65 with a market cap near $73 million and daily volume measured in the low thousands of dollars — present in the data, but no longer functioning as a stablecoin in any meaningful sense.

This is why the case is filed as died rather than wounded. A wounded stablecoin takes structural damage and goes on operating near par; USDX lost the one property that defined it and never got it back. The CDP machinery still runs, but what it produces is no longer a dollar.

One toxic collateral asset can be terminal even under nominal overcollateralization. The mechanism's soundness is bounded by the worst name on the whitelist, so collateral hygiene — what is accepted, in what size, and how correlated it is — is the real safety control, not the headline collateral ratio.

Liquidation systems defend a peg only against collateral that retains a market. When a whitelisted asset gaps toward zero, the same auctions that normally protect the stablecoin become the channel that transmits the collateral's collapse straight into the peg. A CDP coin should be read through the fragility of its collateral set, not the elegance of its liquidation design.

Recovery is about confidence, not just plumbing. Removing the bad collateral repaired the mechanism but not the market's belief in it, and a stablecoin that has visibly broken once can fade indefinitely below par even after the proximate cause is gone. Restoring the code is necessary but not sufficient; restoring trust is the harder, slower problem.