固定されたツイート
Fermi Labs
33 posts
Fermi Labs
@FermiLabs
Bulding the infrastructure layer for on-chain capital markets. Building @FermiTrade and @ContinuumChain https://t.co/s5nzgXxxIg
Singapore 参加日 Şubat 2023
18 フォロー中966 フォロワー
Fermi Labs がリツイート
Fermi Trade v1 is here.
Fully onchain execution. Guaranteed FIFO transaction ordering. Tradfi grade microstructure. <2ms optimistic finality. It’s all possible.
Invite-only Mainnet rolling out this week.
Fermi Labs@FermiLabs
English
"The expected value (EV) of being first in the queue is ~2X-3X of being last, for otherwise identical orders- based on a combination of fill probability and expected return."
0xSeldon | Fermi@yettisheep4
English
Problem: All leading perps DEXes use a centralised/trusted sequencer.
Solution: Proof of Sequence
Continuum@ContinuumChain
English
9/10 of the top Perps DEXes today rely on a trusted, centralised sequencer. Is this really the future of France?
English
A great overview of the approaches at play. Fermi Labs is betting that verifiable FIFO will be the optimal market structure that will win out in the end. Let the games begin.
0xSeldon | Fermi@yettisheep4
English
@toly the aliens are waiting for us to build onchain nasdaq before revealing themselves
English
@toly All this reflects is increased retail participation in equities, on platforms like robinhood which use PFOF. it's not a defence of FBA in any way.
Here's the graph showing increase in retail paticipation over the same timeline:
English
That’s an odd definition of “true fifo”
Most serious actors based elsewhere (say Tokyo), would prefer to transmit their desire to trade via private fibre, then sign and submit a tx from a collocated node in nyc next to the nasdaq server. There is no faster way for a Tokyo based signal to travel to nyc.
If the signal itself arises in NY, the placement of the trader is immaterial- they probably have an edge node in NYC designed to react to the information and place a trade without human involvement. The edge case of human in the loop + caring about +-100ms inclusion in a fifo queue rarely if ever happens imo.
And the tradeoffs of slower state updates seem like a price discovery killer
English
@moonshiesty It is not pure fifo. If i sign a tx at time T locally in location X, and nasdaq is in nyc someone who signs in nyc at T+t will be scheduled ahead of me.
toly 🇺🇸@toly
There is no true “fifo”. True fifo would be each user gets a perfect timestamp as soon as they sign their tx and txs cannot be dropped and are executed in timestamp order. All the deviations from this introduce some exploit vector. Mcp + fba is actually a closer approximation to true fair FIFO then what people call fifo.
English
Long term, Jito’s, or any fair market scheduler’s revenue is hurt more by adversarial nodes than by fair competitors.
If these nodes are scheduled at the same time vs in sequence:
1) default and hides all info
2) Jito with rebates to wallets
3) adversarial node
Jito makes more when 3 is starved, not when some users like ellipsis or drift send an oracle update to 1. Because if 3 cannot be starved the amount of risk and effective volume onchain will be capped.
moon shiesty@moonshiesty
MCP would eliminate the @jito_sol @harmonic_gg business models: BAM and open block building are incompatible with MCP is this a good or bad result? discuss
English
Continuum is now live on testnet.
You can check out the explorer here: continuum.wtf
And learn more from the whitepaper: fermilabs.xyz/whitepaper.pdf
(This is the last post in the thread)
English
Continuum will be the base layer powering @FermiTrade, the first perps DEX offering FIFO ordering guarantees.
English
Introducing @ContinuumChain - the first blockchain with fair ordering guarantees at microsecond resolution.
English
Fermi Labs がリツイート
>To have price discovery on a CLOB you must have FIFO: yes, absolutely. A key point that many miss.
>but FIFO is incompatible with permissionless sequencing: no? It is indeed possible to enforce FIFO at millisecond (ms) resolution and make any exclusion or re-ordering detectable, even in an untrusted sequencer model.
Armani Ferrante@armaniferrante
Consider the following. To have price discovery on a CLOB you must have FIFO, but FIFO is incompatible with permissionless sequencing. Therefore you can’t have price discovery on a permissionless CLOB. You can have extremely liquid markets. But not price discovery.
English
MEV will persist as long as block builders retain the privilege to decide ordering & inclusion within a block.
A question that's not asked often enough: is it possible to cryptographically enforce strict FIFO ordering & inclusion at ms resolution?
If so, that would finally give us feature-parity with tradfi.
Hasu⚡️🤖@hasufl
Thesis Drop: MEV has become the dominant limit to scaling blockchains. Our data reveals that the dominant paradigm of MEV extraction has moved to on-chain searching, and is starting to consume most of the capacity of most high throughput blockchains. 🧵 👇
English
Fermi Labs がリツイート
3/ 🔍 Highlight of the Week:
Onchain trading has developed from fixed AMMs to customized AMMs and order books.
Now, e.g., @eulerfinance & @FermiLabs are pushing capital efficiency beyond existing limits.
Open questions remain on blockchain scaling & ordering guarantees.
English
@toly does it correspond with increasing max tx size, total CUs, and/or stack depth?
English