The Next Generation of THORChain Liquidity

THORChain has always prioritised passive liquidity. This is the biggest market — make no doubt.

Dual-LPs in THORChain deposit capital and hope to make yield. However, they are exposed to two assets on two chains and this was the UX-killer. The dev team attempted to solve this first by allowing assym-deposits. However the LP was still exposed to price shifts, which counted the efforts.

Impermanent Loss Protection (ILP) was delivered to address this, and affords dual-LPs the ability to always buy back their original deposit amount. This solved the price-anxiety, but did not address the UX.

What is ILP? If a legacy THORChain LP deposited 1.0 BTC in an infinitely deep BTC pool, they would be issued a deposit value of 0.5 BTC and 0.5 BTC in RUNE. This is recorded and insured. When they go to withdraw, the redemption values are compared with the deposit values and the shortfall made up. This shortfall would guarantee they leave with enough assets to buy back 0.5 BTC and 0.5 BTC worth of RUNE at the deposit prices.

What is this? This is almost like single-sided deposit and yield, but via a convoluted User Experience (UX) involving another asset and not quite true single-sided exposure.

Enter THORChain Synths

THORChain Synths are synthetic versions of L1 assets collateralised by liquidity units. 1.0 btc/btc is made up of 0.5 BTC.BTC and 0.5 BTC.BTC worth of RUNE liquidity on a best-effort basis. However, despite synthetic assets looking like they are always 100% collateralised — collateral is *not* how they are redeemed. If a synth is presented to its L1 pool, it is always given 1:1 swap rights on the L1, no matter the health of the collateral.

To say in a different way — THORChain allocates synthetic asset holders on a best-effort basis up to 100% of the collateral needed to back the synth, but synth holders are treated like L1 holders when they go to swap across the pool. While the synths exist they look like collateralised assets (<100%), but when they are redeemed they are treated as pegged assets (1:1)

Since the collateral exists on a price-bonded curve, it retains much stronger inertia to its underlying. The loss of value from price X to price Y is the Impermanent Loss:

divergence_loss = 2 * sqrt(price_ratio) / (1+price_ratio) — 1

Whereas if the collateral was 100% the indigenous asset, the loss of value from price X to Y is literally the price loss. Here are some numbers:

-20% price change results in a -0.6% IL
-33% price change results in a -2.0% IL
-50% price change results in a -5.7% IL
-75% price change results in a -20.0% IL
-80% price change results in a -25.5% IL

A 200% collateralised asset would have to be liquidated at a -50% price change to stay solvent, but a THORChain Synth only demands a 5.7% increase of more pool capital at a -50% drawdown.

This is a breakthrough in synthetic assets. Firstly, THORChain Synth collateral never has to be liquidated to preserve the 1:1 redemption quality of its supply and secondly the capital requirements are magnitudes smaller.

Risks

What are the risks and who is providing the IL capital buffer? To date that has been other dual-LPs in the pool. When the price draws down -50%, they cough up the 5.7% capital required. If the synths are only 15% of the pool, the dual-LPs only tender 0.057/(1/0.15) = 0.8% .

However this requires Synths to be capped because the risk on LPs should be capped. However, LPs are given ILP, so they don’t bear the risk — the protocol does. They simply accrue a liability the protocol ultimately pays out.

PoL

So what if the protocol was the entire dual-LP? This means it is directly providing the capital required for synths, not through a round-about manner with ILP. All it needs to do is add more liquidity when a price shift happens, to target a LP ownership rate. In other words, PoL just needs to ensure Synths are never approaching 100% of pool liquidity.

So PoL could be set to be 1% of the pool, and the SynthCap to be 99%. This means for every 20% change of price, PoL is adding 0.6% of the liquidity. This is extremely capital efficient — $1 in PoL can manage almost $100 in liability.

What if PoL was 99% of pool liquidity and the SynthCaps were 1%? It’s flipped the other way. It takes ~$100 of PoL to protect $1 in liability.

So is the magic number 50%? At this point, $1 in PoL is managing $1 in liability. Take a $100m pool, this would require PoL to effectively be $50m. The author believes this to be the absolute minimum — there is no reflexivity created here.

PoL can make money

If PoL is not being used to underwrite Synths (ie, the price shift is favourable), it is actually making money. It’s ultimately an insurance fund that is participating in the market by buying strong assets and selling weak assets. If it can make money, then it can weather downturns better. This is in contrast with ILP (which ultimately did exactly the same thing), which was passive — it was a silent liability that never participated in the market.

This is why PoL should not be set to 1% — an opportunity is being missed out by being so capital efficient.

The correct answer to how big PoL should be is somewhere between 50% and 1% (50% and 99% SynthCaps). Perhaps PoL can be ratcheted up as the legacy dual-LPs leave the system to become the Savers they always wished they were.

Potentially 80% SynthCaps is the sweet spot — 20% of the liquidity is PoL that protects the 80%.

Pareto would agree.