Thread by @AgustinLebron3

1/ A short thread about priors.

- What are they?
- Why do we need them?
- A few specific priors I hold.

2/ Priors are supposed to be "beliefs you hold before seeing the data".

Taken at face value this is crazy. You've been collecting data since the day you were born.

3/ But what we usually mean by priors (in a non-technical sense) is "what is your default assumption".

You may object "Ah, I don't hold any assumptions until I see the data."

That's dumb. (A) You do actually, and (B) you couldn't function in the world if you didn't.

4/ Especially in trading, if you waited for the data to tell you everything you needed to know, you'd act like a madman:

- You'd study *everything* because you wouldn't know where to focus your efforts.
- You'd never find anything. Once something is obvious in data, it's gone.

5/ So you need priors, informed by years of study and learning and people teaching you, etc.

Of course, don't go off the reservation the other way.

Great decision-making in trading and in life is about holding good prior beliefs, then updating them the right amount w/ new data.

6/ Some (obvious to some) priors I hold about markets:

- Volumes are higher on Fridays.
- Almost all US equity up-drift happens overnight.
- Options MMs don't *try* to pin stocks to strikes.

7/ But crucially, enough evidence could easily change my mind about those priors. How much data? Depends on the prior.

Some more priors I have, this time more general/sociological:

8/ Levels of belief:
A - I really believe this and it would take a lot of data to convince me otherwise.
B - I'm pretty sure this is true, but it wouldn't shock me if I believed differently in 5 years.
C - I kinda think this is true.

9/
A: If you're selling a subscription to something, you're not a trader. Or at least, trading isn't the *real* way you're making money.

A: Technical analysis as practiced on #fintwit is worthless.

10/
C: Crypto isn't a long-term solution to almost anything important.

B: The signal-to-noise problem makes it almost impossible for even the smartest most-well-motivated retail person to make $ trading.

11/
B: Retail interest in trading circa 2020 is like poker interest in 2004: most people will lose their shirts, but those who remain will be *fierce* competitors that will make markets more efficient.

12/
Ok, I told you some of mine.

Tell me yours! Use the A,B,C scale.