Add more inspectors

[gakonst]

https://github.com/flashbots/mev-inspect-ts/issues?q=is%3Aissue+is%3Aopen+%22inspector%22+

[obadiaa]

Single-tx inspectors to add

• Kyber Network (https://etherscan.io/tx/0x3e54f6d046e3c3d5441590d5204cb9ff726b18f99220a654fd6ad5388da12d27)
• Bancor (https://etherscan.io/tx/0xda5cad0c628dca3bf70b61984bf2e0621d27dc94e4c6e120c4cc2490fdd6592e)
• bZx initial Uniswap offering (frontrunning) (https://defirate.com/bzrx-initial-dex-offering)
• UMA initial Uniswap offering, same kinda stuff (https://defirate.com/uma-uniswap-listing/)
• DPI/underlying arb
• Augur/Catnip arb

Multi-tx inspectors to add

• Uniswap sandwiching (see this paper: https://arxiv.org/abs/2009.14021)
• Backrunning an oracle updates (DyDx liquidations) (Random ordering of equally-priced transactions incentivises competitive spam  ethereum/go-ethereum#21350)
• NXM price rebalances (backrunning / sandwiching)
  The MCR increments every four hours, so the price drops right wen the MCR update happens on chain. The price of NXM would decrease until funds held is < 1.3 * MCR or MCR at 100k ETH, whichever came first. https://nexusmutual.io/token-model.html So one could essentially sell right before predictable price decrease and buy back right after, in large size

[gakonst]

Agreed with everything, clarifying:

• What does long tail arb mean?
• What is a multi-legged arb? Arb1 -> room for some other txs by other players -> Arb2?

[obadiaa]

Added some extra resources for the inspectors above and happy to brainstorm some more, although I feel @jparyani's work on taking the other approach of looking at what the top bots are doing and using that as a compass for which inspectors to write might be a lil better

@gakonst I added both of these points from the original ts repo issue you linked, I'm not sure what long-tail arb means in this context, maybe @thegostep remembers?

Multi-legged arb just means when you go through several price pairs instead of just 2, but it can still be within the same transaction!

[thegostep]

I believe long tail means small arbs not profitable under normal execution risk conditions

[gakonst]

Yeah @obadiaa now that we have support for the "big" protocols, I agree using the big bots as a compass.

[obadiaa]

We could also use Gervais et. al's paper https://arxiv.org/pdf/2101.05511.pdf for some direction

Generally, protocols I can think of to cover:

• Swerve (?)
• Mooniswap
• Bancor
• Kyber

Seems like our inspectors already cover a large market share though (thanks @sui414 for the pic)