Henrik 🦅
@chaplino
After today’s DeFi meeting at the knowledge well (a16z)
1 reply
0 recast
2 reactions
Sid
@sidshekhar
The ML research one right. any key takeaways?
1 reply
0 recast
0 reaction
Henrik 🦅
@chaplino
very heavily tilted towards trade execution/simulation/trading products three teams presented, all sought to make complex trade strategies (eg a flashloan arbitrage) available as a product to the 'masses' w/ very low code efforts, one other team was doing MEV search & execution as a product -- with tokens slapped on top a bit like alternative versions of the cash carry trade Ethena does
2 replies
0 recast
1 reaction
Sid
@sidshekhar
Oh very interesting (cc @ericjuta). Wonder if any of the team members are on FC
3 replies
0 recast
1 reaction
Ξric Juta
@ericjuta
+1 on ML based solutions but then you run into trust and verifiability still requires a ton of work integrating a core risk engine though as surrounding infra to the core fitted model
1 reply
0 recast
1 reaction
Ξric Juta
@ericjuta
would say ML just cause reinforcement learning agents aren't truly capital efficient yet (gas costs + nature of the market volatility curve pool peg analysis for forecasting entropy distortions could help but again that's factored into the core risk engine to generate a signal to prep for volatility IMO
0 reply
0 recast
1 reaction
