this paper is f*cking insane
real quant trading isn't about predicting the next move, it's about stripping away everything that looks like signal until only what actually works survives.
the framework: kill the noise, then isolate the residual, then stress-test the backtest, then size for survival.
randomness fakes skill easily - a smooth curve can be overfit, a hot streak can be luck, a 30-stock "diversified" book can be one giant correlated bet.
real quants don't ask if it looks good. they ask if it survives every attempt to break it.
bookmark before this gets buried.
@jukan05 A-shares = “monkey market” — no bull/bear 🐒 Story stocks rip or dump on “convenient” news, but real businesses can go insane too. Honghe Tech is up ~40x since the Sept ’24 rally (~2x SK Hynix). Yet volatility rinses retail traders out; most are lucky to capture even 10%.
Anthropic engineers just showed how they build a full app from scratch, using a loop of agents
40 minutes from the team behind Claude Code
they used three agents: one to plan, one to build, one to judge, cycling until the app actually works
the winners won't have the smartest model, they'll have the best loop
watch it, then read the full guide on how to actually use loops below
Honestly, I feel a lot of disappointment and frustration.
In the full version of my CPO article, I tried to capture what I have been hearing from people actually working in the field. I met with multiple optical industry professionals, interviewed them, researched the topic, and organized those conversations because I believed the article could be helpful to many people.
But many of the responses came back as, “Someone said this,” or “SemiAnalysis said CPO is delayed.”
Of course, I understand. I do not have the same level of credit as those well known names yet.
Still, it is disappointing to feel that the time I spent writing this piece and the voices from the field were dismissed so easily.
I am not asking people to blindly trust me. I just hope people can take the repeated voices from the actual field as seriously as one sentence from a famous source.
Full article: https://t.co/SBRHsXIXr8
When do you reach for other models instead of Codex? What can we do better? Hit me with all of your frustrations. dms open.
If you can give me detail (e.g. specifics/transcipts) - it'll help a lot in finding out exactly what we need to do to improve the next model