I've hit followers! This would have been utterly shocking to me three years ago, so thanks to everyone. As a thank you, here goes a thread on a question I get asked a lot: How do I come up with good trading strategies?

1/ Sadly, my answer is usually "I don't know." And that's because I really don't. A. I don't know your background. B. I don't know what you're good at. C. Most importantly, I don't know what interests you.

2/ As a substitute, in this thread I’ll show you a process for how I come up with trading ideas. Hopefully you can pick out aspects and ideas that you can apply to your approach.

3/ STEP 1: Live In The World This sounds like a joke right? Aside from some random orbiting astronauts, we all live in the world right? No.

4/ It’s actually a constant challenge to live in the real world. To see it clearly and to avoid the protective cocoon of the familiar and the safe. I wrote a whole thread about this. https://x.com/AgustinLebron3/status/1434963956599193601…

5/ But finding new trades means learning about unfamiliar things, or maybe about unfamiliar aspects of familiar things. I actively curate my information diet, and I work hard to pay attention to *what* I pay attention to.

6/ Rick and Morty quick hits on YouTube may be fun but it’s the worst kind of mindless time-wasting. Learn to use these buttons:

7/ Of course, you can’t go overboard and create a filter bubble for yourself. There’s an optimal amount of randomness that’s not zero, and the task is to shape the randomness in useful ways. For example:

8/ Arxiv on my RSS reader. I wrote a thread about how to read academic finance papers. https://x.com/AgustinLebron3/status/1443951372580229123…

9/ - ACX open threads - Thoughtfully-selected DM groups - Financial and non-financial news from countries you don’t live in - High signal-to-noise ratio follows

10/ STEP 2: Put Yourself In Position To Notice Things Useful trading ideas never come to me during periods when I’m not paying attention to markets. That doesn’t mean that ideas always come to me while I’m staring at markets, but it has to be “in the air”.

11/ The world of trading is broad, so figure out which parts interest and motivate you. Then spend time doing those things. In my case, I like staring at ticks. My idea of a good time is looking at this.

12/ Minutes easily turn into tens of minutes. Sometimes an hour. Often I’ll look at a few different ones together. I do it because I enjoy it, but I’m also actively putting myself in position to notice interesting things.

13/ Of course, interesting is in the eye of the beholder. Maybe staring at ticks makes your brain hurt. You don’t have to like what I like. But you need to find the things that interest you, and then put yourself in positions to notice things about those things.

14/ STEP 3: The Magic Happens At some point, an idea emerges. How does it happen? I don’t know. Sometimes it happens at your desk, but sometimes it’s in the shower or in a dream.

15/ But when you’ve been mulling over enough interesting things for enough time, eventually the pieces seem to want to self-organize in your mind:

16/ - “If X is true and Y is true, I wonder if Z is true.” where X is an idea you might be able to study. - “Is there ever a time when X and Y happen together?” - “This reminds me of [weird thing from 5 years ago]. I wonder if the same thing applies here.”

17/ These thoughts aren’t trades, at least not yet. But they’re the kind of thing that could become trades, or components of trades. Call them trades or signals or edges or something else. Doesn’t matter. The point is that you’ve thought of an idea you want to investigate.

18/ STEP 4: Make A Plan I bet you expected this next step to be “Go study the idea”. But that’s usually a big mistake at this stage, and common one too. I’ve often made the mistake of prematurely jumping into writing code or looking at data.

19/ Sometimes I still make this mistake! But why is it an error? Don’t underestimate the importance of priors! The thing about markets is that (a) sure there’s a lot of data, but (b) almost all of it is noise.

20/ What I mean by noise is “movements and activity which are orthogonal to the idea you had.” One person’s noise is another person’s signal, but from one specific person’s perspective, almost all of the apparent actionable information in historical data is irreducible noise.

21/ This means that if you’re trying to pull out a small signal out of that noise, you can’t just look at the data. It’s going to lead you astray because you’re not going to be focused enough to avoid getting swamped by the noise.

22/ You can think of this like overfitting, but on steroids. Overfitting is a specific stats/machine learning term that refers to model-building. But the same issues crop up at a meta-level.

23/ That means you need to put together a plan that you’re going follow before looking at data. That plan needs to have:

24/ - The specific question or set of questions you’re going to try to answer. - An estimate about which data (and how much of it) you’re going to need to look at. Data is a scarce resource and we can’t afford to waste it.

25/ - Expectations you have about the results of your study. Write this down! This will help you remember and figure out which results are surprising when you do the study.

26/ - Bounds on value. + Lower: “This needs to be worth at least X in order to be worth looking at.” + Higher: “I don’t expect this will ever be worth more than Y.”

27/ Armed with this plan, you can finally proceed to the next stage: digging in. Ok, that's good for today. Tune in tomorrow for part 2 where I talk about how I try to dig in.