We know that good empirical estimations of MEV by type (liquidation, sandwich, backrun etc) can be given by mev-inspect-py, however has anyone done any analysis on a platform basis?
I would love to see what percentage of each DEX’s flow is actually being frontrun/sandwiched/backrun and if there are any significant discrepancies between DEXes. What could explain them? Fees? Other protocol mechanisms?
Are some protocols more vulnerable than others to toxic MEV? Why?
This is helpful to understand from a user perspective but also from a protocol designer perspective.
Cool idea! It would also be interesting to look for interactions between DEX design and chain architecture: For example, DEX design A might attract more back-runs than design B on rollup 1, but B attracts more than A on rollup 2—perhaps because of properties of rollup 2’s sequencer, etc.
Alternatively, if DEX design A attracts disproportionate MEV across all chains it’s deployed on, that’s decent evidence there’s something about that design per se that’s inviting the MEV.
I suppose a lot of thought should go into controlling for DEX-level factors like average order size if your interest is in drawing conclusions about DEX designs themselves rather than incidental characteristics.