Content pfp
Content
@
0 reply
0 recast
0 reaction

Nastya pfp
Nastya
@nastya
So I'm trying to figure out who the best traders on Farcaster are - out of hundreds of thousands of users. What are the main challenges? First - the data. I mostly use Dune since it's the biggest source of blockchain data with rich context - but it's far from plug-and-play. Blockchain data is messy, and Dune queries often need cleanup or deeper knowledge to make sense. Even something that sounds simple - like getting DEX swaps for a wallet - turns into a whole project if you want to be accurate. Second - figuring out what metrics actually matter. How do you calculate PnL, and over what timeline, tokens, chains etc. A high total PnL might come from a one-time airdrop - not consistent skill. From what I've seen, past big wins don’t mean much - what really matters is whether the trader's still active and knows when to take profits. I've gone through many iterations of the algo, and I'm still refining it. But so far, the most useful approach is looking on the freshest metrics. Anyone cracked this already?
4 replies
4 recasts
15 reactions

jivansh.eth pfp
jivansh.eth
@parzuko
we spent a lot of time trying to solve this problem at /castmoney unfortunately, indexing is a hard problem when you get into the nuances. values are not consistent across data aggregators/vendors and computing everything yourself is expensive and tedious (it is not impossible tho 😅) biggest fix for a lot of logic is to have opinionated design. for example: airdrops gains maybe don’t count towards pnl, specifically according to your algorithm also, important metrics to rank traders are usually not singular events. “good” traders have multiple wins, have active wallets, also lose sometimes, usually terminally onchain etc
1 reply
0 recast
4 reactions

Nastya pfp
Nastya
@nastya
In-house indexing seems such a headache... Yeah, opinionated design seems a way to go. Initially, I was quite surprised that literally every service shows different numbers for the same wallet
1 reply
0 recast
1 reaction

Ape/rture pfp
Ape/rture
@aperture
Hence why you don't have to do the indexing in-house. You can let The Indexing Company handle that, so you get the data inhouse in your own database. Then you can focus on building the logic.
0 reply
0 recast
1 reaction