Transcription
Trading didn't click for me until I started stealing strategies from the best traders in the world. Now, I tried trend lines. I tried support and resistance. I even tried complex indicators from academic papers and MIT. But it didn't really click for me until I stopped trying to reinvent the wheel and I started stealing strategies from the best traders in the world and then most importantly automating them.
Today, I'm going to be sharing my exact process for finding strategies that have made me over $200,000 and why most traders are wasting their time trying to be original when they could be just copying what already works.
Now, to start off, first we need to understand the main type of edges that exist in the market. So, primarily we have risk premiums. So, this is where you're getting paid to take risks others won't. And then you have inefficiencies. So, these are actual market mistakes you can exploit. Think of things where there's maybe something with the exchange. Maybe if you're trading crypto and there's an issue with the exchange, or there's maybe arbitrage, all those sorts of examples.
Now, most traders think they're trading inefficiencies. All the videos you see out there is everyone saying, "Hey, this, you know, this strategy is exploiting the market." But most of the time, they're just harvesting risk premiums. And there's nothing wrong with that. But let's Sinclair, a quantitative researcher and portfolio manager, break this down into how to look at risk premiums and inefficiencies.
"Inefficiency and a risk premium."
"An inefficiency is literally a wrinkle that exists. There's profit there because not enough people have noticed that it's there or not enough people can actually do the trade. So it's it's literally like an oversight. Whereas a risk premium is literally you getting paid to take a risk someone else doesn't want to do. The example I've used before is if you see that there's $20 sitting on the street. Okay. And for your younger viewers, there is this thing called money which used to consist of little bits of paper that you would occasionally find on the street. So according to the like an efficient market theorist, the old joke, right, is well, there can't be money on the street because otherwise someone would have picked it up. But traders know that that's not true. So we will occasionally see money on the street. If you see the money on the street and you always see it on Sunday morning outside of a bar, okay, that's an inefficiency because what's happened is someone's walked out of the bar the night before. They were reaching into their pocket to get their phone out. They dropped the money. They didn't notice it. So that's why that money's there. It's an inefficiency. It's literally shouldn't be there. It's there because someone else made a mistake and you're the first one to pick it up. If you see the money on the street because it happens to be sitting in the middle of a freeway, that's a risk premium. Other people have seen that money there and they've just decided that they're not going to go and pick it up. So risk premiums can be there for forever and people can agree that it's there and just disagree because of their risk tolerance whether it's worth going to get it or not. So a lot of things that people think are inefficiencies like that really aren't. They just..."
Now, why does this actually matter? Now, inefficiencies often disappear faster. They have limited capacity and they're harder to find, but they do give higher returns for the most part. Now, risk premiums normally persist for decades or they at least have such as trend following, for example. They scale with capital very well and they can be a great foundation for any trader portfolio of strategies, especially if you're starting to manage more, say, six figures or something like that. Or if your plan is to have more stable returns and then slowly add more capital, whether that be from your job or just from trading profits, it's just going to take a bit of a longer period, then going for something such as inefficiencies. But then you have all the cons of inefficiencies that they disappear a lot faster and also they're harder to find. So for a long period of time, maybe you don't find an inefficiency and you're not really making any money.
Now, what do I mainly do? I mainly do risk premiums. But as I kind of mentioned here, it's the foundation of my portfolio. I have risk premiums that kind of pay the bills. Let's say they're just, you know, the the main core returns, and then I put inefficiencies on top of those. So when I find an inefficiency, whether that be in the crypto markets or any other market, then I can exploit those, but they often, you know, dissipate kind of fast. So they can be a nice, you know, really good adder to the portfolio return, but they are not going to be the thing that I am relying on for month-to-month, uh, sort of returns or better yet, yearly to yearly returns. Let me say.
Now, this moves me into what should be the first thing you look at for actually finding these strategies. Now, this is going to sound a bit boring, but it's going to be starting with books. Uh, some of the great ones I've read are "Advanced Futures Trading Strategies," "Quantitative Momentum," and then "Option Volatility and Pricing." This is normally the easiest way to find solid strategies to start from. Now, I will say there is a lot of crap books out there. So, don't get me wrong that, you know, every book you buy is going to be a great book, but there are some amazing books like the ones I kind of listed, and there's there's more definitely as well. But the great thing about a book is they aren't limited to just a YouTube video or a very short YouTube video where someone sort of explains it. Normally in a book, they go through the whole thing, and then if it's a systematic book, they'll normally provide like backtest results or at least for that time period or their methodology behind it. Like, for example, I know a lot of profitable traders using similar approaches to Robert Carver. I even use some approaches from him. Now, are they going to be approaches like I've mentioned before that are going to make you 200% returns in a year? Probably not, because a lot of them are kind of risk premium based. So they're going to be very nice, stable returns, uh, or at least things that could actually work. Are they going to be free Sharpe ratio or two Sharpe ratio? No. It, it's, it's primarily going to be a risk premium, which is still a great way to get started in trading. Also, this book just has a really nice, uh, kind of cover. So, I think it's quite a nice looking book.
Now, the next one is going to be watching others. So, this is where it can get a bit more, I guess, tactical. It really depends on what you're trading. Uh, but one, you know, way you can definitely do this is you can look on-chain. So, if you're trading crypto, you could look at what are other profitable traders doing. Now, I'm not saying by this is just to copy trade them, but try and understand what are they exploiting? What is their sort of strategy? Are they seeing something that you're not? Now, obviously, this takes quite a lot of time. It's hard work, but this is going to really apply to anything in trading. So, if you're not ready for hard work, trading is definitely, uh, not the thing for you most likely, especially in the earlier years.
Now, two is Twitter. Twitter has a lot of gems, but the same with on-chain trading, there's a lot of noise and stupid stuff out there you have to filter through, and some of the time you will fall for more of the stupid kind of content. I know I definitely did earlier on. And let me actually rephrase that. I, there isn't as much stupid content on Twitter, but there is a thing where it may not apply directly to your approach, right? So, I'm not a discretionary trader, so there's a lot of great advice out there on Twitter around discretionary trading, but is it going to apply to me? Most likely not, because I'm just, I don't really do that sort of trading. So, you will have to filter nevertheless.
And then lastly, I would say podcasts. Uh, I pretty much went through the whole Chat With Traders library. It's a great podcast. I will say the older episodes are normally better. Um, but there's a lot of other, you know, popping up podcasts where they'll have, uh, another guest on again, maybe that you've seen on another podcast in the history. For example, like Ernest Chan normally goes on new podcasts, Robert Carver normally does as well. And sometimes they drop really nice hidden gems. A lot of times it's going to be things maybe you've already heard, but the point of podcasts nowadays is really to try and see when maybe someone slips up or they reveal just more information and they're more candid around what sort of approach they're doing, or is there something particular that they're looking at. And then additionally, if you're consuming these a lot, sometimes you'll relisten to the same podcast, but something will just click for you for that time. And that happened to me a lot of times. So personally, I would say podcast is really up there. Um, but probably the most tactical thing is like on-chain and maybe Twitter, and then podcasts I think can be a really good, uh, thing to also consume. Also nowadays everyone's quite lucky, like we've got AI, so you can scrape blogs and you can go through transcripts in minutes, but it just does take a bit of work. You're going to have to go through a lot of these, and there's going to be a lot of filtering, but just as this lesson is, I'm going to actually show you a clip from Ernest Chan, basically saying to similar things, sources of ideas, and I actually written down so many of them in in the book. So it is very hard to just sit there and just daydream and say, oh, you know, I think that, uh, this trade will work, right? It seldom happens that way. Usually it's inspired by what you read. There are many, many academic papers on different trading strategies. There are many, many blog posts on simple trading strategies. Now, the, so the sources of ideas, many books also mine as well, and others. So the, the actual, um, I would say, um, trick of the trade is not the original idea generation because there are already thousands of them out there. You all you need is to read and and and and listen. The, the really, the trick on trade is how to filter them down.
Now, number three is testing everything. I use solid data providers like Polygon and Kinetic because bad data equals useless results. And that's why I'm not really, uh, a promoter of manual backtesting or trading replay. It's not really a scientific, let's say, method to actually doing proper data analysis and, you know, statistical significance. And the most important thing there is because if you have bad data or if you've stored bad data or whatever, it's going to make your results useless. There's, there's no reason really to get any accurate results from it, right? So try and use a solid data provider when you can. Now, I will say there are fields to this, meaning that more specific data, let's say tape data or tick data, that's a lot harder to get quality data for or it's going to be more expensive. Whereas daily data, you could probably get that from most places and it's going to be quite decent. It won't normally be perfect from some of the completely free sources, but it will do a pretty good job. So, it depends on what your needs are and where you're starting, but for the most part, try and prioritize solid data and try and get it from either a data provider that provides you just like a CSV file, or try and actually maybe learn a bit of code or hire a coder, and then they can pull some data for you.
Now, ideally, you want to be testing across like five plus years of data. Uh, and you also want to include all your costs in these, uh, in these backtests. So, you want to make sure to do commissions, slippage, locates, whatever really applies to you on actually a live trade, you should be applying to your backtest. And the main idea of testing everything here is we're trying to filter out as many bad ideas as possible. And this is really only possible by testing it historically and seeing if it works historically at all. Now, obviously, this gets into a whole another realm of overfitting and overoptimizing. I have a lot of these kind of videos on on the channel. But to keep it, you know, very short and simple, simply test the idea that you came up with or the idea you saw. Test it once, see how it looks. If it's somewhat decent performance, maybe spend a small period of time, you know, adding maybe another filter or trying slightly different approach. If it looks completely horrible, just leave it. Just completely leave it. If you try and overoptimize that to start to start it making look good, uh, it's just going to be a terrible experience. You're going to be left with something that's really overfit. So try and just, you know, out of the gate, if it somewhat looks decent or somewhat has potential, then you can spend maybe a bit of time on it. Uh, but at least my filter is, if it looks terrible at the start, I just leave it.
Now, if you're not willing to pay really for a provider or you maybe also want to pay for one, uh, you can use Yahoo Finance, they're completely free. Uh, IB Insync that connects to Interactive Brokers, completely free as well to get some data. Polygon is paid, but I do think they have a free tier, but it's not the greatest. And then there's also Alpaca, which I think is paid as well.
Now, four is running robustness tests. So, a lot of you maybe have already heard of backtesting, but you probably haven't heard of like parameter sensitivity or out-of-sample testing or multi-color simulations. And these are probably the next step above backtesting. Technically, they are actually included in backtesting. It's normally what you should do. Uh, but a lot of times nowadays when someone says backtesting, they kind of just think, you know, one historical test, and that's backtesting it. Uh, but for the most part, you should be doing something like parameter sensitivity and out-of-sample. Those are my main two ones, and I also have videos on the channel, uh, around those. Now, of course, there's tons of other approaches you can do, and there's also probably new approaches you can kind of develop into it. Uh, but at least for the standard ones, I'd say parameter sensitivity, out-of-sample, or walk-forward optimization, and Monte Carlo, I, I guess would be the third one. And those are the kind of the main ones you want to look at. And for the most part, a simple backtest is not going to be enough for statistical confidence or significance really. So you want to have these extra tests layering on to see if you've maybe overfit or if the strategy just got lucky for a period of time. The idea here with backtesting is you're trying to get an idea for one, its behavior, and does it even sort of work? Because ultimately, you won't know if it, you know, continues to work until you take it live. But taking it live can take months, maybe even years depending on the sample frequency. So you want to get a decent idea beforehand, and then there's the whole other conversation that the better you backtest the strategy, the more conviction you have in it. Meaning you can hold it better, you can execute it better if you're not automating it. And also you can go through those drawdowns and not be too worried. So there's a lot of benefits to backtesting it overall. I'll leave you with a clip with Jim Simons of kind of what he says about other funds and how they tried to do backtesting, but they actually did like manual overrides, which just leaves it with not being a scientific approach.
"Uh, you know, investing firms say, oh, they have models, and what they typically mean is, you know, we have a model, and it, it advises the trader what to do, and if he likes the advice, he'll take it, and if he doesn't like the advice, he won't take it. Well, that's, you can't, that's not science. Uh, you can't simulate how you would do, how, how was, how were you feeling when you got out of bed, uh, you know, 13 years ago when you're looking at at historical simulation."
Now, number five is automation. It was an absolute game changer for me. The problem with manual trading is your emotions get in the way of extracting your edge, and now my algo can execute at 2 a.m. while I sleep, or if I'm, you know, going on a plane or whatever. There's no emotion, no hesitation, just execution, and that's it. Now, to automate, though, you do need to have exact criteria. You need to know what are your entry and exit conditions, position sizing formulas, what are your stop-loss rules, exit triggers, etc. And it can't be something that's broad. It needs to be something that's exact, like, you know, two if conditions. So, if this happens, then this happens. Uh, and a lot of times the subcriteria there. So, let me give you a quick example. A lot of people will say it needs to touch an order block, or touch an FVG, or it needs to, um, touch a trend line. But how do you define those things? And there's a lot of different ways of defining those things that it's going to execute with in your algo. So you need to understand, okay, to define a trend line, I'm going to use this X method, right? Um, now, a lot of ways you can go around this is you can look up indicators that people have quantified those sort of approaches already and get ideas from them, or you can come up with new ideas yourself. Um, both have pros and cons, but that is two methods you can do. Also, if you've never touched any automation before, maybe just start working in Excel, because Excel formulas will give you a really good idea of like, how does this sort of if and or kind of logic work? And then from there, you can maybe define them more and then hire a coder, or you could learn code yourself as well and trying to automate those. Just to give one final thing here, automation, while I did say like, you, you do need to have it in exact criteria, you can do sub or like parts automation, meaning that maybe you handle all of the, you know, scanning to the tickers, you find what tickers it should trade. Um, but then the actual entry rules that the algo does that for you. For example, I know a lot of kind of hybrid traders that I've also worked with, and they'll normally do the entries, and then the algo will handle everything on the exit side of things. Uh, so they don't have to worry about trade management.
Now, second, you will need to pick a method to do automation. So that can be direct API connection to your brokerage or exchange, for example. So straight to Binance or straight to IB. It could go through a third party like Alpaca or QuantConnect. Or it can be inbuilt into a platform like Ninja Trader 8, for example. Now, each one of these have its pros and cons. I couldn't go over them all in this video. Um, but personally, at least for me, I use direct API or I use a platform solution sort of like Ninja Trader 8 for futures.
Now, to sum this all up, after six years of trading, the main thing I try and focus on nowadays is proven concepts or stealing proven concepts plus rigorous testing and automation above trying to invent unique strategies. Trying to invent things that are original to you can feel great. It's a, it's a great mental exercise. You feel very smart, but ultimately on P&L and results you get out there, it's going to be very limited, at least in my experience. So, try and at least start with taking concepts that already work, and then you can also apply them to just more inefficient markets. A lot of huge alpha in crypto was that people that were already in traditional finance when crypto was first kind of getting established, they could just pretty much copy-paste those strategies into crypto, and they would work very well because crypto was a more inefficient market. And then to sum up where I find my ideas, as I kind of mentioned in this video, it's books, podcasts, and then also a lot of Twitter as well. And then lastly, test it properly. Do walk-forward optimizations, do real costs, you know, include for slippage, etc. And then let machines execute for you. And it sums up how I've become profitable.