What does Veeky Forums think of CAPM?

What does Veeky Forums think of CAPM?

Attached: AT-CAPM.gif (500x455, 19K)

Other urls found in this thread:

reddit.com/r/CryptoCurrency/comments/815rhx/quantamize_modeled_cryptocurrencies/
kamny.com/load/publications/p03_eng
twitter.com/SFWRedditGifs

i think it's easier to just buy index funds and not try to figure any of that stuff out

I think the interesting question whether you could move the frontier up with a convex combination of crypto + regular portfolio.

p sure hedge funds do research on this.

Placing Crypto on the efficient Frontier would be bafflingly hard to do. But it would break the scale. Since there is no "low risk" crypto and it seems to have no effect on other markets in not so sure it would shift it up as much as remove part of the diminishing returns characteristic. Although again I feel like expected return on crypto is just a giant ????? Calculating beta values for coins would be quite useful though.
Well yeah. But why not discuss finance on the business and finance board for once?

>efficient markets
>Crypto
Pick one.

Well, mathematically it wouldn't be, using ex post values. There's one company that has calculated correlation matrix for the top-20 coins, it'd be a small step to figure out the covariance matrix.

But ofc, to calculate _expected_ profits would be very hard, as you said. I guess one way to tackle this would be to partition the existing data and try to compute a minimum variance portfolio and see how that would perform out of sample with the other half of the data.

efficient frontier and efficient markets are conceptually different, though

basically with cryptos you are just looking another asset class that is decoupled from the (uncorrelated with) general performance of the stock market

Currently doing a finance 2 course at uni, the teacher said today that its pretty impossible to predict returns and risks for stocks etc you can only assume it and argue for the point. Isnt it kinda dumb to learn this kind of stuff if that is the case when the market portfolio beats every other combination almost every time long term.

Link on the correlation? That would actually be insanely helpful on building a portfolio. I have access to journals if you know the name. And running your model would still find nothing concrete because depending on your window you could have literally any result going in any direction. Your standard error would be fuck huge and you would probably fail any significance test.

I was thinking of bootstrapping or something, I've got background in econ, not so much finance.

-- A quick look at the literature shows that it's been attempted with varied (very modest) success.

I'll dig the correlation matrix link shortly