@macrocephalopod Hey ceph . I was wondering do you think it’s possible to systematize stock picking (longer term) investing methods made by discretionary investors ? For example there are some tech investors who are extremely good at tech trends and understanding how it impacts supple chains…
@HangukQuant Just curious , but if you let opus 5.5 ( or whatever is the latest model ) build the whole thing without too much supervision but only a broad architecture , how good or bad do you think the final result would be ? I’m still wondering how much hand holding one needs .
@HangukQuant But the agents alone would not have been able to come up with the strategies you came up with to if you only gave them your dataset and backtester alone right? Or it would have been super overfitted.
@Danielsrotaryo1@pplsartofwar I don’t understand how the vibes and the statistics/numbers are so different . America still has the largest gdp, fast growth for a developed state, still has the reserve currency and the strongest military with 800 bases and the reserve currency.
@HangukQuant Hey . Are you offering any kind of student discount for the Substack? I think I saw a post a while ago but I don’t think it’s valid anymore
@HangukQuant Are you planning or interested in doing anything pure equities ? Or do you think the equity markets are just way too efficient for a solo quant ?
@GoshawkTrades The only way to learn the best quants’ dev and research process is just working at one of those tier 1 firms for a few years . Of course it’s not easy , but there is no substitute for being able to learn from and talk to the best in the world and the institutional knowledge.
@HangukQuant Hi. I’m just wondering , how has the improvements of llm’s coding abilities changed your workflow if any or how does it fit into your workflow how as a solo quant dev/researcher ? Of course I assume you still have to read every single line of LLM generated code .
@Scifospace /2 I believe the typical workflow is to train on the first 15 year, validate on the past five year, and then select a few good models with reasonable metrics, and then port them to a simulated strategy, then tweak buy and sell rules. Is that roughly correct?
@Scifospace /1 Hi, I am a relatively new P123 user. I was just wondering, what is your typical workflow to go from a ML model to a simulated strategy model?
@RankEquity What is your validation process before live deployment ( like what do you want to see in a simulated backtest ) and how do you decide on which ai model to choose once you see the validation result ? Are you just choosing the model with the highest (highest bucket %return?)
@KrisVerma88 Hey Kris ! Love your work man! Just wanna ask do you think locate cost has gone up noticeably ever since the elimination of the pdt rule ?