"And some are not"

-- Anyone who has watched people apply technical analysis to Bitcoin charts as a source of amusement over the last decade

Thank you. I have a very close friend that has been doing this for some time, and it’s really painful to see him lose money to those “forecasts” that resemble more astrology than science. I see from your profile you’re a Quant. Is there any guidelines I could give him? Any recommendations to go to the more “scientific” side of finance, and stop looking at charts and draw arbitrary lines to try to come up with patterns.

Sure, I can try. My thoughts on this actually come more from my experience in prediction market betting than my work, which is hard to apply at an individual level.

I think just properly accounting for wins and losses can be good at instilling a sense of humility; I'd recommend he calculate his track record if he hasn't already. I enjoyed the book Thinking in Bets by Annie Duke, which is about the psychology of developing some habits around making bets and the cognitive biases we have. A Random Walk Down Wall St. is also a classic that is good at instilling a sense of humility in you as an individual investor.

You could add that even winning strategies can be unceremoniously smashed against the rocks of market execution failures, especially in an unregulated space like crypto. I had to learn from personal experience: roughly a years’ worth of hard-earned arbitrage profits were wiped out in the aftermath of the 50% drop back in March, despite holding offsetting positions, due to uncontrolled liquidations plus a DOS attack on an exchange.

Basic economic theory and econometrics. Finance theory is just a thin layer on top of that.

What is a "quant", when referring to a person?

Quantitative analyst https://en.wikipedia.org/wiki/Quantitative_analysis_(finance...

Not that much more crazy than what you read about other stocks in terms of technical analysis.

I would say that depends. The macro trends have been pretty obvious.

most are not. =)

Predicting past performance with error bars measured in parsecs doesn't sound super useful to me.