The problem with lending against new tokens

New tokens on Elysium are easy to buy and hard to borrow against. This page explains why, and why a price feed with a fixed LTV cannot fix it.

How a token lives on Elysium

On Elysium, the Hyperliquid L2 built by Kinetiq, a token usually moves through four venues:

1. It launches on a curve.
2. It moves to a pool.
3. It graduates to a HyperCore spot order book.
4. It may get a HIP-3 perp.

At every stage, liquidity is thinner than the token's market cap suggests. Market cap is supply × last price. What a holder can actually get is whatever the bids absorb before the price falls away.

The holder's position

Holders of these tokens sit on positions they cannot easily use. If they need HYPE, the only option is to sell, into a thin market, at a worse price than the screen shows. Selling also gives up the position they wanted to keep.

Borrowing HYPE against the token would solve this. In practice, nobody offers it.

Why nobody lends against these tokens

A thin-pool TWAP can be pumped

A lending protocol needs a price. The usual answer for a long-tail token is a time-weighted average price (TWAP) from its pool. On a thin pool that average is cheap to move. An attacker buys the token up, holds the price for the length of the TWAP, deposits, borrows HYPE at the inflated price, and walks away. The lenders are left holding collateral worth a fraction of what was borrowed.

When a liquidation comes, there is no buyer

Even with an honest price, a liquidation has to sell the collateral. If the book is thin, selling a large position crashes the price. The liquidator either cannot fill the sale or fills it far below the oracle price. The gap between the debt and what the sale raised is bad debt, and the lender eats it.

Governance-set LTVs cannot see any of this

The standard fix is a conservative LTV picked by a vote. A single number cannot describe a market. It does not know how deep the book is today. It treats a 1,000-token deposit and a deposit of 5% of supply the same way, even though selling the second one would move the price far more. It does not change when liquidity leaves.

What Bythos changes

Bythos stops asking "what is the price?" and asks "how much HYPE would selling this quantity actually raise, right now?" That number is Depth Value (DV): the HYPE obtainable for a quantity, selling into the bids and stopping at 15% below the best bid.

• The limit follows the book. Your borrow limit is 70% of Depth Value. If the book is thin, the limit is low. If it deepens, the limit rises.
• Size is priced in. A whale depositing 5% of supply gets a lower effective LTV automatically. The book sets it, not a vote.
• The liquidation venue is the pricing venue. A Trawl sells into the same venue Sounding read. The value the protocol lent against is the value it can recover.
• Pumps are filtered. Depth Value used for borrowing is the minimum over the last 15 minutes, capped at 0.9 × the 30-minute TWAP, and new collateral only counts after 50 blocks. See Anti-manipulation.

A worked comparison

Take an example pool with 1,000,000 tokens and 100 HYPE, a spot price of 0.0001 HYPE per token, and a 30-minute TWAP steady at that price.

A hypothetical protocol using price × 50% LTV would let a holder of 500,000 tokens borrow 500,000 × 0.0001 × 50% = 25 HYPE. Selling those 500,000 tokens into this pool in one go would raise far less than the screen value, and the sale would wipe out most of the price.

Bythos stops the walk where the marginal price is 15% below spot, after about 84,907 tokens, which raises about 7.80 HYPE. The Depth Value of 500,000 tokens is therefore 7.80 HYPE, and the limit is 70% × 7.80 = 5.46 HYPE.

That is a much smaller limit, and it is the honest one. It is the amount the market can repay.

What this means in practice

Bythos does not make thin tokens liquid. It lends exactly as much as their liquidity supports, and no more. Low limits on thin books are the design, not a flaw. Read Sounding for how Depth Value is computed and Risks for what can still go wrong.