First up, I’ve made a minor schema change by replacing the swap volume fields with user swap volume fields.
(eg: user_swap_volume_usd = swap_volume_usd - extractor_volume_usd)
I’m doing this to make reporting less error prone, as I think it makes people less likely to include extractor swap volumes when reporting on user volumes.
I’ve completed writing an API usage document (link). I’ve also written this data sources & limitations document which I recommend you read (link).
For a long time I’ve been aware of issues with Flashbots calculation of sandwich attacks. As you can see in the document I’ve been successful in recalculating those when measured against the comparison data (Eigenphi).
However, I do not currently reclassify Flashbots arbitrages. Recent data comparison tests show that these are well below the volume detected by the comparison set.
Improving Flashbots arbitrage detection is beyond the scope of this or any existing funding we have (in fact, re-calculating sandwiches was as well and I personally subsidized this).
Unfortunately, Flashbots are not actively developing mev-inspect-py at the moment. If this continues, I will seek further funding to re-classify arbitrages as I have with sandwiches.
While I want to be transparent about these limitations, ours is still by far the most comprehensive and accurate public MEV dataset available, and I have no reservations about launching as is.