Have you gotten much traction on the subscriptions? I made some financial simulations for my own use and was thinking about ways to monetize, but I dismissed subscriptions as I didn't feel there would be a big market. I'd love to be wrong about that.
Anyway, I don't plan to actually make a product out of mine, but if you're interested in speeding up Monte Carlo financial simulation by literally a million times, let me know :)
It does seem a bit niche, but there's been modest traction so far. And sure, I'd be curious to hear how your Monte Carlo implementation might differ. Feel free to reach out on email/discord if you'd like to chat more.
> And sure, I'd be curious to hear how your Monte Carlo implementation might differ.
Excel is just too slow/not the right tool for MC (or most serious statistics). From my experience with mcmc, compiled languages come in pretty handy to speed up any loop that just can't be vectorized with some linear algebra. So I'd use any tool I'm comfortable with to set up the simulation but delegate the hardcore simulation loops to something like c++.
Agreed on Excel not being a suitable tool for MC; that's one of the many things that pushed me in the direction of building ProjectiFi. Within ProjectiFi, I'm sure there are some things that could be done to speed them up further (e.g. offloading them to a backend where they can run faster), but there could also be some trade-offs with that -- for one, it would violate the "everything stays client-side by default" principle mentioned in another comment.
> I'm sure there are some things that could be done to speed them up further (e.g. offloading them to a backend where they can run faster), but there could also be some trade-offs with that -- for one, it would violate the "everything stays client-side by default" principle
I guess it's just me who hasn't yet seen any serious mc simulation running on a client I guess :/
The way I see this done whenever it runs fast is using c/c++/java. Another important thing with statistics is the mathematical validity of your results, regardless of whether they look right. For validation, most literature I come across uses c/c++/r.
Anyway, I don't plan to actually make a product out of mine, but if you're interested in speeding up Monte Carlo financial simulation by literally a million times, let me know :)