Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> If someone says their algorithm relies on some sort of TA, run for the hills.

Care to explain? I'm genuinely curious as I've had some success in this area. Curious if I should be aware of something that I'm not...



Fundamentally, the history of a price has nothing to do with its future price. The laws of nature do not care if you are on a bull run. So TA is completely bunk in that regard.

However, obviously many people do rely on TA, which means TA is influencing the price (you can beat their TA algo with your own TA algo). The problem is, you never really know what everyone else is doing. Relying on TA amounts to playing rock-paper-scissors, blindly, with 1000 opponents, and hoping you choose the winning move against most of them.

Keep in mind there are only 2 outcomes, win or lose, and they occur randomly, so it’s really easy to fall victim to selection bias and think you have a winning TA-based algorithm.


> Fundamentally, the history of a price has nothing to do with its future price.

This is not true at all. Price movements show auto-correlation, for example.


Makes sense, thanks for the explanation.


TA indicators have number of flaws.

One of the biggest flaws is that TA indicators tend to repaint. So an awesome winning MA crossover in hindsight might never really execute during real trading.

Additionally using TA for trading also involves self-fulfilling prophecies. And if there are any markets which follow this prophetic tendencies - it is cryptocurrency.


Excuse me for being ignorant, but what does TA mean in this context?


Trend analysis. Kind of the first thing they teach you in tutorials, I think mostly because it's easy to convey. I could explain it here, but you're better off reading the Investopedia article.

https://www.investopedia.com/terms/t/trendanalysis.asp


They do mean technical analysis.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: