There was so much to to absorb there. I think we might at the end, we might come back to how do traders get started and all of that that you've said. Because I mean, if they begin a traders or maybe traders who are trading for a while, but they, you know, they want to, you know, adjust what they're doing might hear that and go, well, that's so overwhelming. How do I do that? But it's really, I think, what I've figured out from you by following your content is you can take a lot of these complex ideas and really simplify them. So I'm going to come back to that at the end and ask you for a simple answer to how people can get started there. But we've got a bunch of questions in the chat as well, which I'm going to raise to you in a minute. But first, I wanted to kind of switch gears a little bit more towards finding signal, especially in the noise. We've got some very noisy markets. And you've talked about finding signal a lot. First of all, what do you even mean by that? What is finding signal? Signal? Yeah. I'm

going to try to give you a non three minute answer. Let me try to give you my perspective on what this means. So if you think of the market as a completely random process, all the data are random, random walk. If there's a repeating pattern, then you might consider that a break from the normal randomness of the market. And that's what you're looking for is this consistent pattern that appears in an otherwise random process. And that is what I consider a market inefficiency. It's some structural issue in the market that's causing a consistent pattern to repeat itself. And it's our job to amplify and detect that pattern, and then figure out how to exploit it for profit. And then of course, measuring if it's actually random, or if it's actually repeating. So with that, with that kind of on the on the table, a lot of the concepts that folks are used to hearing about, you could consider market inefficiency. So think about a mean reversion process, we all know that volatility means reverse. And if you consider volatility as a random process, and you witness this mean reversion all the time, well, the question then is, how do we exploit, we're seeing the signal. So how do we measure this and how do we exploit it? So a very, very simple example could be like, take the Z score of your volatility. And if that Z score goes above or below three or minus three, it's at some extreme. So this should catch your attention. Now, whether that's going to work consistently or not, I'm making no comments on that. But that's just an example. Another way you could do that is try to forecast volatility in a way that is better than the market. So there's these arch models, there's modified arch models, there's been thousands and thousands of pages of academic research on forecasting volatility. But if you can model volatility in a way that's improved over the market, then you can price your options more accurately, and you can find yourself in a better position than the rest of the market. So these are the types of things that you think about. Now, it's not easy, right? I'm not trying to make it sound easy. I think there's often a lot of signal in these economic relationships. Pairs trading is very popular. Pairs trading is, you can imagine, two assets that are linked, economically linked, think like Apple and Foxconn, and that relationship breaks. There's a dislocation in that relationship. Well, that relationship tends to be together. So can we exploit that temporary break for an opportunity profit? That could be a signal for you. So these are the kind of spaces where you want to look. These economic, you find the pairs that are economically linked, geopolitical shocks, everything that's happening in Eastern Europe right now causes all sorts of opportunity to look for market moves based on these geopolitical shocks, supply and demand, etc. Yeah. As you were explaining that,

you said the word structural inefficiency, I think it was structural. What about behavioral markets or market participants act certain ways around maybe specific price levels, or maybe some kind of indicator or price action, does it have to be structural, or would you even consider that to be structural as well like these behavioral aspects?

It's a great one. I think of technical analysis when I think of technical analysis, I'm not a huge fan. So forgive me for those of you that are. But I think one of the only redeeming qualities that I find in technical analysis is exactly that. Is that if you've got this golden cross, well, every retail trader and their mother's brother and their sister is going to be looking at this level and doing something about it. So there's some counter positioning that you can take advantage of there. Now, of course, the question is, can you consistently do that in a way to consistently make money non randomly? I think that's that next level question. But to your point, yeah, absolutely. There's this hurting behavior. We've seen this and most trends are because of this behavioral aspect, right? Like new information isn't priced in fast enough, then all of a sudden it is, and then you've got this herd behavior like Nvidia, right? All of a sudden the world woke up and realized that every single generative AI application is based on GPUs and guess who's the best provider of GPUs in video? The whole world woke up to that. I mean, Nvidia's gone up what a trillion dollars in market cap in like 12 months. Well, clearly, there's some hurting going on there, but some behavioral thing. Yeah, so there's definitely that as well. Yeah.

Okay, I'm going to ask you a quite a broad question and let you answer it how you see fit and then we'll get into some of the questions in the chat. Where are the market inefficiencies

today? Where would you look? The back spread is a good one. Yeah. You can start to find some mean reversion happening here now that the kind of trend of the overall market has started to become exhausted. You can find some names that are overextended and starting to mean revert. I trade a momentum factor and that's been behaving really poorly lately, which typically tells me that the market regime is shifting. Now it's obvious to see that, but if you can build a portfolio of assets that are exhibiting lots of mean reversion, then you can absolutely capture some edge. Now, this is a portfolio based strategy as opposed to necessarily a single asset or spread strategy. I think the pairs trade is always a good one. It just depends on how big of a universe you can search. And then I'll say it again, the crack spread is still working. There's a uranium trade that's been quite good. So if you can find all the uranium names, there's some regulatory constraints that impact certain uranium producers from exporting to the US that you might be able to exploit. So we've seen that trade work pretty well over the past couple of months too.

Sounds like you do a lot of keeping in touch with the news, looking at developments in geopolitical developments. So I guess that that may bring up different opportunities as history reveals itself. Yeah, I think so. I guess you don't run out of ideas to test. Is that an assumption?

Yeah, I mean, that's kind of the problem. It's like you've got this backlog of stuff to test and only so much time to test it. What I found super effective is you find one strategy and you just express your view in many different ways. So the crack spread, I keep going back to this because it's like taking money from a baby. But you can trade the futures, you can trade the options. Obviously, if you're an options trader, you know that there's an infinite number of ways you can express your view and the options market. That's such an underrated way to continuously capture edge on an idea. For sure. You know, somebody said something, I read it somewhere. It's this concept of a luck surface area. Like the more you read, the more you consume, the more you know, in an intentional way, the broader the surface area that you have that helps you get lucky with stuff. And I kind of see it with trading the same way. The more you know, the more opportunities you'll be able to put together, you'll be able to piece the puzzle together, put two and two together in more ways than your competition. And that's what edge is, right? It's that it's making money on the margins where you have some informational advantage for some reason. So that's kind of the perspective I take. The more I can consume in an intentional way, the more opportunities will present themselves through time.

Yeah, yeah, I really liked that. I read a book. I'm trying to remember the name of it now, but it was talking about creativity. And you know, the whole idea was to consume a lot of broad knowledge, even in, you know, outside of trading and just synthesize it or let your brain, you know, think about it, and you come up with all these unique ideas. And if you look through history, there's quite a few examples of if someone taking an idea from one, you know, what's the word? Not sector, but yeah, one domain and applying it to another and, you know, revolutionizing that. So I really liked that you, what did you call it? Luck surface. Luck surface area. Luck surface area. I really like that. It's pretty neat.

Not mine. But I'll, yeah, I'm not going to buy, but I use it. Yeah, yeah. Now we've got

a couple of questions here in the chat. Excuse me. So I'll have a crack at some of these. Cool. There's a long one here from Charlie. First of all, Iliya said quantopia didn't they shut down long ago? Maybe I was, was it quantopia? No, someone who, who did you have the view in?

Quantopian shut down. They got acquired by Robin Hood. John Faucet bought the right to the email list. And he's kind of resurfaced quantopian in the last couple months as a, as a community and as a, as an email distribution. So it's more of an educational content. You don't get a cool platform and all that. But you go to quantopian.com. I'm pretty sure that's the domain and you'll, you'll be able to sign up for the newsletter. It's like long reads once a week. Like, let's look at a strategy and go 5,000 words on that strategy. But he's, he's back. Okay.

Cool. I wasn't aware of that. So thanks for explaining that. I'm going to put this question on the screen. I'm not exactly sure what they're asking. Maybe you, you know, if not, then we can just skip it. But I'll, I have a question. Can someone answer please? If my balance curve is growing smoothly, over time and my equity curves having trouble to meet the balance curve, what does that mean? Any ideas to smooth

it? Don't know what balance

curve is.

Okay. Yeah, sorry. Sorry.

Yeah. If you, if you, whoever posted that, if you want to just clarify a bit further, that'll be helpful. Okay. This one's a long one. And we kind of touched on this a little bit, but let's, let's have a crack at this one from Charlie. Thank you, Charlie. So I'm going to read it. So three part. Hey, Jason. I fully appreciate that the correct way of doing quant trading is having a trading strategy based on some economic relationship and then improving using machine learning or AI. My curiosity got the best of me. And I've tried using data mining to build trading models using economic features without providing a economic relationship. All validation tests are passed and the models keep working during live trading. My question is, if the models cannot be fully understood, but they make money, does that invalidate them? It's, it's kind of a

question, old as time, Charlie. I mean, ultimately it comes down to two things, right? If your model is making money, great. Put on some very good risk management, right? In all cases, you want risk management. I'm not just talking about stop losses, but whatever your style is, make sure that you're implementing the appropriate level of risk management just to protect the downside. I don't want, I talked about data mining and I don't want folks to get so caught up on definitions, right? Like ultimately, if you've got a model that is statistically significant and you've got enough data points to show over the long run, it can consistently make money on a non-random basis, then you're probably in good shape. Like who can explain how large language models work, not even the PhD scientists that invent that, right? So there is always going to be an element of this black box where you put data in, it does its thing and the stuff comes out. But going back to first principles, it's just important that you understand the theory and most importantly, the assumptions behind that model. So for example, does your model assume a normal distribution of asset returns? Well, if the answer is yes, then you're violating that model because your asset returns are not going to be normally distributed. I mean, that's a pretty known thing. That's why a lot of time series don't work in the markets and that's why the markets are hard because they're non-stationary, right? And all these things, you just need to be careful that you're not violating the assumptions of the model that you're using and make sure that you're doing the appropriate testing on the back end to show statistical significance and put on your risk management and who really cares if you can explain what this column of data means, if your model is performing and you're sure that it's correctly, accurately representing what you think it is.

Thank you, Jason. Well, said, hopefully you found that helpful, Charlie. Great question. Thanks for posting that one. Thanks, Charlie. World of Thor would like to know a starting point for a newcomer. So this goes back, I guess, to our first principal's discussion. Yeah. But yeah, what do you recommend for someone who's starting

out? That's great. So there's, I think the mental model, I'll talk about the mental model and then I'll give you the answer. I think the mental model is here, is this going to be a part time thing or is this going to be a full time series thing? And it means two very different things. If it's a part time hobby thing, then you've got to manage your expectations that, you know, you've got to carve out the time. If it's a full time thing, then you got to treat it like a business. Okay. So that's kind of like the very first conclusion that you have to job for yourself. Now, more practically, I think the best way to do this is to get the market intuition by literally staring at the markets for like a month or two months, get yourself a paper trading account and just stare at the limit book on these things and actually enter limit orders and enter market orders and, you know, in your paper trading account, buy the offer and sell the bid and see what happens and see how the market reacts and see how like, you know, all of a sudden there'll be a five tick spread in a market because some news comes out and the liquidity dries up. Well, if you had a position on what are you going to do about it? The more intuition that you can get on how the market behaves, the more comfortably you will be. Okay. So after that, I think you got to learn the tooling. And again, if you want to get into touchable analysis, then figure out like how these things are calculated, talk to people, understand what works, understand what a breakout means, understand what consolidation means, understand what patterns to look for, a lot of rote memorization, a lot of pattern recognition. If you're trying to go to the quant route, learn Python, full stop, learn Python, like, that's it. Take a, take a $19 Udemy course, you know, you can read my newsletter, I publish, I think I've got 80 weekly newsletters out now, you can read them all. You can join the newsletter and get them in your inbox. You can take a $19 Udemy course, you can do whatever you want, but get yourself very comfortable with the Python syntax, loads of free content out there, free courses, paid courses, paid content, spend three months getting good at Python, spend three months getting good at pandas, P-A-N-D-A-S. So pandas is the data manipulation library. It's at the heart of everything we do in quant, everything we do in data science is this pandas library. And then all the while, just read, you know, read about the markets, you know, listen to better system trader, listen to people with experience, listen to people who have done this, read about their exploits, read about their losses, just read, expand that luck surface area. And then after three or six months, you'll be in a pretty powerful position to start. Okay. Then you can kind of start thinking about how to apply all this knowledge to the actual markets.

Yeah.