Entropnt

yup right here: https://github.com/PlathsOven/jupydash have you given it a go / how useful have you found it? would love to chat about what you'd like use it for

looks like the llm i'm using is outdated rip

it should, let me know if you have any issues

good catch lol, fixed

you'd be surprised how inefficient the markets actually are in practice. for example, while i was trading indian options, implied vols were consistently higher than realised vol for the very simple fact that retail loved to buy options and there wasn't enough capital among the risk-tolerant market makers to absorb all that flow. so a sufficiently informed retail trader could profit by simply selling options in those overpriced names.

it really is similar to the argument for why startups exist at all -- if the opportunity was real, why hasn't big tech already absorbed it? in practice, seems like it's either because the market is just so large that everyone can have a slice of the pie, or the pie is so small that the titans just cbs (e.g. cross-exchange arbitrage in crypto was a lucrative opportunity for a while -- an obvious and easily engineerable strategy).

hey appreciate the feedback! how useful was it for you?

Hey, co-founder of Halcyon here. Thanks for your feedback!

Our plan is for the AI agent to create an initial iteration of the bot, but make its logic visible and tweakable without seeing or touching a single line of code. That way, you can correct for any inaccuracies in its first pass.

We'd love to hear more of what you'd like to see from Halcyon! We're happy to chat over video as well: https://calendly.com/sean-gong/halcyon-founding-users-x-foun...

Hey, co-founder of Halcyon here. Thanks for your feedback! We want to help traders build their own automated strategies without touching a line of code. AI tools nowadays are great for building an initial functional strategy. But building visibility around the strategy and iterating further on it takes not only time and effort, but also a continuous understanding of all the moving components in the strategy, which only gets more difficult as the strategy becomes more complex.

We want to make the entire feedback loop easy: building, monitoring, iterating.

We'd love to hear more of your thoughts! We're happy to chat over video as well: https://calendly.com/sean-gong/halcyon-founding-users-x-foun...

Hey, co-founder of Halcyon here. We want to help traders build automated trading strategies without touching a line of code. The AI agent will talk to you about your strategy to gather enough information for us to build it for you.

If you have any other thoughts or requests for Halcyon, we're happy to chat over video as well: https://calendly.com/sean-gong/halcyon-founding-users-x-foun...

Hey, co-founder of Halcyon here. Thanks for asking! Curious to hear more about what you currently use for backtesting and what your process is, so we can build around you.

If it's easier, we're happy to chat over video as well: https://calendly.com/sean-gong/halcyon-founding-users-x-foun...

You’re right that prediction markets in their current form are closer to a casino than a truth machine — that’s literally the thesis of the article. The question is whether that’s inherent to the mechanism, or a product of broken system design. The article argues it’s the latter, and proposes specific structural fixes. The mechanism itself — weighting estimates by conviction — is well-established. The implementation is what’s failed.