Created: August 21, 2020
Modified: August 28, 2020

finance

This page is from my personal notes, and has not been specifically reviewed for public consumption. It might be incomplete, wrong, outdated, or stupid. Caveat lector.

Yale course: https://oyc.yale.edu/economics/econ-251
MIT course: https://www.youtube.com/playlist?list=PLUl4u3cNGP63B2lDhyKOsImI7FjCf6eDW
Several things can be true:
- The game of finance attracts, and creates, people with antisocial personalities.
- Financiers often abuse their positions to extract more value than they individually deserve.
- Allocating capital is one of the most consequential, and hence most socially valuable if done well, lines of work.
- The decentralized incentive systems of modern finance are incredibly more effective than central control would be.
There is a sense in which financiers are everybody's boss. They are the upper management of the economy. Ultimately, CEOs have to do what they say.
- Maybe this is why everyone at Goldman is a 'vice president'.
- Or it could just be that they're trying to project the image of being highly qualified.
I think it would be fun to play with finance a little bit. It's a fun set of ML problems and could even be addictive. I'd learn a lot about the world. I might get excited about time series models (which after all, are really AI, or at least would have to be in order to be any good).
It'd also be a chance to learn about a new area and develop the skill of thinking about companies. How do you value a company? What should I be thinking about?
I'd get to exercise the SuccessfulFriend-like parts of my brain. And it's a subject he could actually teach me.
Suppose I were to get into finance. What would I want to invest in?
- First: what do I believe in? (includijng but not limited to things I believe that no one else believes)
  - Are there good companies with good financials?
- I believe in fake meat and dairy. It's highly moral, it's environmentally friendly, it's ultimately going to be very cheap. It should take over a large fraction of the world's current meat consumption.
- I believe in drug development. It's expensive and insanely difficult but it's also incredibly valuable. The tools are getting better. If it's insanely valuable now, think of the value when we can do it 10x as well (when do I think that will be?).
- I do believe in the relatively rapid advance of AI. Industry does mostly incremental work and can't factor in unknowable future tech advances to their plans. But I believe there will be such advances. Training bigger and bigger transformers will become possible and we'll be able to condition on a fuckload of text and image and video and audio data. We might not understand what we're doing, but we don't need to. The thing we're showing is that fundamentally the problem wasn't all that hard. Most people are at least subconsciously duallist and will tend to overbelieve in the idea that humans are special and can't be replaced.
  - It could actually be really fun to become a finance-industry expert on AI, like Nathan Benaich. I'd get to think about big ideas, people would care about my thoughts, my PhD would carry a lot of cred, and ultimately, it is an area I believe in (even if it's full of a lot of salesmanship today).
  - writing inbox: do I still believe that AGI might happen in my lifetime?
- TODO: come back to this after I learn a bit of nuts-and-bolts valuation and trading basics.
- Second: what is my trading strategy? What are some common strategies?
Separately: is there a probabilistic-programming-for-finance story? That I want to pursue? Maybe integrating epistemic uncertainty / RL / etc, or maybe just modeling?
- Obviously time series prediction is a thing.
- Quantopian did some stuff. Thomas Wiecki, etc.
- Generally, mixed-effect modeling is a thing. Can I explain it?
- Ultimately one should fine-tine GPT3 on the entire stock market correlated with the entire internet and see what it does.
  - Or at least with all financial news.

finance

Links to this note

diversification

Meta