really fast curve updates per LP
1
11
Conversation
Can you unpack this?
1
its a central limit order book. a curve is a big pile of bids and asks at a continuous function. updating the curve per LP would require a price time priority queue for order fills.
1
3
Limit orders are granular, & expressive of MM's price preference! But to run a full OB matching engine is expensive even for the likes of @CoinbasePro (let alone on-chain w gas).
I wrote on AMM and OB equivalence. Incl. a possible spectrum of designs: professorjey.com/assets/papers/
1
1
One of the most powerful byproducts of the AMM is the LP token - it allows DeFi to be leveraged (in defi lingo, "composable") - can you think of how we can replicate such a construct in decentralized CLOB?
1
4
To begin, let's think about central MM. The MM is a service provider to the exchange. The LP token in this framework is a representation of share of service provided (work).
1
The AMM is nice, because LP participation share (as in ) is determined by asset balances in the AMM state.
1
Fittingly, LP token construct democratizes the level of service provided by all LPs, to their commensurate share of the pool reserves.
In a LOB, MMs whose orders at top are exposed to big market move "crests", but they are credited for contributing most for making the spread.
1
In LOB exchange, you need to incentivize a lineup of diverse MMs.
Exchanges show love to their traders, but their best MMs are family.
Arguably there is an exclusive "LP token" club for these next of kin.
1
So to answer 's thought, I think we need to consider slicing another dimension(s) not just share of asset balance, for assessing the LP's contributions to the AMM when it becomes hybrid AMM/OB style.
Perhaps this may look effectively like quantile/tiers of LP tokens.
2
1
I guess what I'm having trouble grasping is the LP token balance expresses the holder's contribution to liquidity while not withdrawn
With a LOB construct, it's v hard to reason about how much a set of limit orders are contributing to the liquidity of a book
I've shown that any given LOB has an equivalent AMM state. But I agree to crack the new LP concept begins first with defining the metric of contribution. This is new territory!
1