Try   HackMD

TBD

  • What are some good heuristics to identify "arb flow"

    • inspect can pickup arb bots that execute atomic complete swap cycles, compile their addresses for any given pool and filter that out as arb flow?
    • but what about stat arb? are single swaps that take advantage of deviation from external "true price" to be accounted for as well?
    • how does this change for assets that don't exist on CEXs or just have a single pool on-chain (all flow is non-arb flow?)

  • What does arb/non-arb volume breakdown for any given pool tell us?

    • do the tradfi toxic flow assumptions hold true?
    • are LP returns strictly proportional to the amount of non-arb flow they get or is it more complicated?
      • regardless, is it still the biggest factor?
    • how does this differ for x*y=k pools vs concentrated liquidity pools?
      • Do v3 pools get a lot more non-arb volume?
      • How does JIT liquidity change things? does this make JIT driven DEXs simply better than ones that actively mitigate it?
      • Are unsophisticated LPs (say on x*y=k pool for an asset where there's another concentrated with JIT liq) simply signing up to lose money bc they get exclusively arb flow?

  • On price execution of RFQs

    • do RFQs/Cowswap style batch DEXs truly offer better execution than say v2/v3 pools at all times?
      • intuitively feels like no and there's a lot of adverse selection (intuition: solvers only fill an order where there's $ on the table for them, that only happens if they're filling under true price)
      • is "gasless" trades a bad deal if you're paying it in execution quality?
    • does technically being a limit order make this not a problem? (since user technically agreed to the price)

Possible starting points:

  • Look at ETHUSDC 5bps UniV3 pool (since its volume rich, we've decent arb data) for a week, conduct the analysis, then extend it to all pairs, all pools across all protocols.
  • Scrape cowswap trades from their API (from Jan 21st for the amount of volume it had), see execution price and compare it to a simulated trade on uni v2/v3 (making sure the timestamp is the same as when it was filled on cowswap)
Last changed by 
 
Taarush V
0
107

Read more

inspect listener error

ERROR:__main__:{'code': -32000, 'message': 'first run for txIndex 275 error: insufficient funds for gas * price + value: address 0x9D8e2dc5615c674F329d18786D52A F10a65Af08b have 241395854351228173 want 241436981652380070'} Halts at block: https://etherscan.io/block/15538730 Tx causing the break (payment from block builder to fee receipient): https://etherscan.io/tx/0xdc8a049e7a122eebd33833627bcb1c2dc0fa57515d7273d70b73a742753f9d71

Nov 16, 2022
mev-inspect collaboration

Given the limited capacity we have, it’s best to push mev-inspect-py and the dashboards depending on it to a contributor / community maintained work stream. We would like to give a grant for the grantee to complete the following tasks in the next 2-3 months (in order of priority): Debug and fix critical bugs that stop the listener from moving forward Currently, we modified it (only in prod) to skip over blocks with errors as a hacky fix but ideally we fix issues that cause it to halt in the first place Ensure mev-inspect data <> dune export tool works Currently, as long as the listener is running, it exports that data into dune’s s3 but like mentioned above, if it stops so does the export tool Create a dashboard on Dune that looks similar to mev-explore as an example to showcase how the community can use the data to create dashboard. (we will deprecate mev-explore after this, and point any post-merge graphs to this Dune dashboard)

Nov 16, 2022
Potential yield pools / Ukraine design

Design goal: Let people delegate their funds (stables/ETH) and donate the yield Plug into as many yield sources w/ sufficient TVL Something like https://rdai.money/, more flexible Proposed Architecture: accept arbitrary token -> swap for stablecoin -> tokenize stablecoin 1:1 -> deposit stablecoin into vault -> yToken held by proxy -> bot harvests yield and sends to donation add For ETH, BTC and stables

Feb 28, 2022
Tracing + token flow analysis

Primer Process Aside from the EOA initiating the transaction itself, we check for balance changes for the contract (in the to field of the tx) that executes the arb/liquidation but also proxy contracts that can be identified by looking at the delegate calls (if any) in the tx traces. Types of traces (by action_type): Call, which happens when a method on the same contract or a different one is executed. We can identify the input parameters in each instance by looking at this trace. Self-destruct, when a arbitrage contract destroys the code at its address and transfers the ETH held in the contract to an EOA. Common pattern among searchers given gas refunds. Create, when a contract deploys another contract and transfers assets to it. Reward, pertaining to the block reward and uncle reward, not relevant here.

Jul 27, 2021
Read more from Taarush V
Published on HackMD