The reason this would work is that automated trading systems/algorithmic trading, including highly secretive proprietary ones, are absolutely normal in the industry. By 2006 nearly a third of all trades in the US and EU were conducted by such systems (
cite).
Ours would just be ridiculously better than any other option because we can cheat*. Cyber devils know everything on their host systems and seem to develop immediate skills around using their capabilities, such as driving a car despite having been bodiless internet peons right up to that moment. Plus basically being strong AI in the sense that they're as intelligent and intuitive as a human but operate at computer speeds**.
Buy a big server cluster, set up a facility near an exchange, purchase analytic software and textbooks on various aspects of mathematics/economics to load onto the system, and stuff it to the brim with the best minions / idea the thousand hells ever had.
At first it'd just be our money in the fund, but once it shows consistently good results others would buy in. Even once we have lots of money such a tool would still have value, so it has staying power as project that stealing assets doesn't.
Farming favors is a lot harder though. We should try collecting lower grade favors from medium fish at quantity sometime, since we need 'cash' flow to keep up healthy influence growth, but a backfill project for after we're more of a medium fish organizationally.
In my eyes all of this doesn't invalidate what you're saying, but it does demonstrate that our optimal gain in situations like is only rarely going to be financial, primarily due to way our opportunity costs and barriers to entry line up.
* Not just the cyber devil stuff either. Divining the future is hard, but the present is much simpler. With enchanting we could feasibly skip the internet connection and get zero lag data from all over the world just watching the public data feeds. Which isn't cheating; the exchanges outright sell faster access as a service and we can claim "advanced machine learning based predictive algorithms" if anyone asks.
Or alchemy things; if we can research recipes from the other properties listed in ExWoD then stuff like this from Mummy the Resurrection would be useful:
- Lesser Quiddity - With the use of Sekhem, the Mummy activates the quiddityof an object (its symbolic spiritual attributes)
cite
Not sure what the bonus would be for a demon computer, but any reasonable symbolic association should enhance something useful to the inherently data processing heavy task of stock nonsense.
** It wouldn't surprise me if they were capable of picking up on the 2008 financial crisis and housing issues related to that before humans can. Which we could turn into ethical profit - just hear me out on this.
We buy up those mortgages and use them as normal until the bubble bursts. Then instead of foreclosing we keep buying while everyone else sells for pennies on the dollar. We only answer to ourselves in a corporate sense, so going negative isn't necessarily an issue.
Then we end up as landlord for who knows how large a portion of the country. If we target the buys correctly we could even "coincidentally" focus our money on politically significant areas of the country.
Then all we have to do is not be a gaping asshole about it and we're heroes to huge groups of people. From there we can refinance to something reasonable to trade money for influence, forgive debt outright as part of pushing some political agenda, or talk our way into getting Uncle Sam to effectively pay the tab at near sticker price by dressing it up as a bailout followed by forgiveness on all the debts we own.
People get to keep their homes, we get lots of money, and the people who created the situation get screwed financially while being held to whatever standard we set. Win-win-win.
There'd be complications in all of this, but by that point we should be scaled up enough to manage it.
It's also a modern-ish issue we can interact with in the mortal world that doesn't necessarily take us into politically charged waters OOC, which isn't exactly a common opportunity.