0:00 I made more than 30 million euro for the bank. 90% of institutional traders were going like in one place to get their trading strategies ideas. The place guys is the social science research network and I want you guys that you steal this like steal as much as you can because this one was something that really impressed me when I started. He traded over $15 billion in total volume at one of Europe's largest banks. Introducing the one and only Matteo [music] Conte. There is an industry paper called VWOF, the holy grail of day trading [music] systems. The fact that the vast majority of institutional traders are using this type of algo amplifies these directional moves. While trading at the banks, Matteo ran over 400 plus strategies at one time. And in this episode,
0:48 he reveals the exact place that institutions find their edges. And this is in a nutshell the big difference between retail and institutional trading. What is the pipeline every single institutional traders follow? Every single trading strategy needs to have these three boxes filled. So you need to know why you're opening. You need to know where to close. And a third condition is having like these exact edges and strategies have been used to get consistent prop firm payouts [music] by moving towards automation. Matteo reveals all of this and so much more in this special episode of Chart Fanatics. Every single institutional trader that I know at least once if not every single day they send an execution order to the market. They often use
1:36 I guess starting off like where should we begin? What are we covering today? Yeah. So let me start with um some facts like obviously uh I started trading just like the majority of the people that are watching is most likely like being 16 17 years old wondering what trading actually is. How do you trade like an institutional trader? But like online is very hard to find proper information, especially like fresh information straight out of the trading floor. And um to get to know what is the actual process, the step that I went through was going to university, studying 12, 14 hours a day because like it's an extremely competitive type of job. So,
2:24 you really need to have top grades just to book yourself an interview with banks or hedge funds. And if you're lucky enough like I was at the end of my five years of university, I got a seat on the trading floor. Uh I joined the bank in 2018 getting my own books in 2019 and since there like there's been a process of learning how institutional traders actually trade and it is very much different from what you usually see online. Um the very first thing that I want to say is that the first difference between retail and institutional trading is that retail
3:13 chase trades. So all retail traders go after trades. On the other end, institutional traders go after validated trading strategies trading strategy. And this is in a nutshell the big difference between retail institutional trading. And this is as well the reason why there is inconsistency between the profitability of retail traders versus like having institutional traders that make money basically every single month. We are going through what is the pipeline that
4:05 every single institutional traders follow. Mhm. which so pipeline and this pipeline starts from getting the idea for a trading strategy. And this is just the beginning of this long journey to end up to actually having um our trading strategies trading life because once we have the idea, the next step is defining a set of rules. This set of rules needs to be so specific that you can feed it to a machine and to feed it a machine it needs to be encoded.
4:54 Now I imagine at this point the audience might be thinking oh two things. One is what's the difference between chasing trades and validated trading strategy? To them that might sound like the same thing. Yeah. Well, chasing trades, you might see like one of the usually like they point at one specific setup. They say this is a setup where like this uh can work out well for you, can make you a lot of money 70% of the times, right? But if you look for validated trading strategies means that is a strategy that if you apply it systematically every single time is going to win 70% of the trades. Do you see the difference? So more so rather than the idea of
5:39 a trade, it's actually validated with data with actual specifics that you can look at and go and have not certainty because don't want to give across the wrong impression but almost it's the closest thing you can probably get to certainty within the markets. Yeah, you can you can understand what are the probabilities of success and this is the main difference. It's not just about because like a good strategy can have 70% win rate, but if you only trades that strategy as a single trade, it doesn't guarantee you that you're going to be right that that you're going to be on the right side of that trade in that specific moment. On the end, if you apply that strategy over 100 trades, statistically you will win
6:26 70% of them. And this step of going from a set of well- definfined rules to a piece of code, it has been for a long time the biggest barrier to entry for retail traders to start trading like an institutional trader. Lucky for you guys uh in 2023 something u magnificant happened which was like having Chpt and the other large language models going mstream which completely remove this massive barrier to entry and now that you have the ideas or you can get the ideas we will get there where to get the ideas like an institutional trader you will know shortly how to that the
7:12 rules that every single training strategy needs to have, you can encode it leveraging the power of your CHP code or your favorite large language model. And then the next step is to back test. So here what is back testing? Back testing is taking your set of rules and replaying history. So every time your entry conditions were satisfied, you would have like a simulated trade back in time. And obviously you don't back test it manually like I've seen doing it online every now and then, but you leverage the fact that you translated it into a piece of code to just replay history. And this one is going to
8:02 give you one single equity curve which can be a pricing then rising if the strategy is not good enough. But as an output what you get out of it is a set of statistics as we were saying that uh can tell you if you are on the right path. Like for example, if you have a positive net profit, what is the win rate of your trading strategy over the past let's say five years? Um what is the average trade win? So how much your strategy is winning every time you place a trade? What is the draw down of your strategy? So which draw down is the where draw down is the peak to valley of
8:51 your communive P&L when you have a rough batch and all of these statistics are necessary to put you in the map. However, this one is not the last step because then we need to validate. Can you read it? Yes, of course. Yes. Okay, the validation is another extremely important step in institutional trading and usually is divided in two phases. One is the split test. Mhm. So splitting your data between in sample and out of sample. So in sample versus out of sample.
9:38 Why is this one important? This one is important because when you work on your set of rules when you are developing your trading strategy putting code you should always work with in sample data. So let's say that you have 10 years of data here. You should purely focus like let's say we split this data in 80% in sample and then you have a remaining 20% out of sample in your in sample data is where you're going to work on your rules adding removing parameters fine-tuning the parameters until you find a strategy that looks like it's behaving correctly in sample. Yeah, let's take a break for a minute there, guys, cuz a quick word from our official platform sponsor,
10:26 Ninja Trader. If you've been following along on this channel, you know how much time we spend breaking down charts and market structure. And Ninja Trader is an awardwinning futures trading platform built for exactly that. You get advanced charting, depth of market tools, and market replay so you can work through historical sessions and practice in a fully simulated environment, all without risking real money. Here's how to get started. If you're new to the platform, use the link in the description below to download Ninja Trader and explore it in the simulator. You get two full weeks for free to test everything out. And if you're already trading prop, choose Ninja Trader as your platform at checkout. It's the best platform for the job. So, go check it out and
11:12 let's get back to the episode. And afterwards you feed the same rules like frozen in time to unseen data. So that if your strategy is still performing good out of sample means that you didn't overfit it. Mhm. Because overfitting is when you add too many rules or parameters just to make your equity curve look nice in sample. But as you apply to unseen data, you will see something like this, like an equity curve that gets destroyed. And this one is going to be the first one. If you have a good result both in sample and out of sample, it means that you didn't overfeit and there is a high chance that your strategy will work well with live data as well. And this one is really
12:01 to put it a very high level, we uh just to give a little bit of context. And the second step is the Monte Carlo analysis. Mhm. Monte Carlo analysis is something that like when you look at the internet looks extremely difficult to understand but uh what is really about is just taking all the traits of your back test and then either reshuffleling the order of the traits which is what is known as Monte Carlo reshuffle and basically like in this way you create multiple version of the past multiple equity parts so that you can understand if your equity curve in your back test was looking
12:49 good just because of the sequence of trades or is that because you really have an underlying edge and the so let's just show let's say that this one was the equity curve of your yeah the original one and then you start changing the order of trades right But you will always start from the same point and you will always end up to the end because the traits are always the same. However, you the path to get there is going to be different and then you repeat this process basically thousands of times [snorts] and the bigger is the dispersion from the top
13:35 equity line to the bottom or the weakest is your edge. Yeah. on the dread like when you see that reshuffleling the order of the trades you have a very tight distribution around the average it means that no matter what is the order to which your trades took place you still were going to have a nice tight distribution of the outcomes you're showing the edge is very strong exactly regardless of how the trades are shuffled so when we say shuffled to for the audience in terms of that it's not changing the strategy or anything. It's literally a case of let's say the original uh data was a thousand trades and of those thousand trades like the average losing
14:22 streak let's say was seven and the average winning streak was five but it shows you the peaks as well. So the peak losing you know number of losing trades in a row was 15 in the original test and then the most wins in a row was 15 just for example random numbers. But then the the reshuffle was essentially putting all those trades into a random order and it might be a case where you know the the peak losing number of trades in a row was now 45. Exactly. And therefore you know to your point the further that's how when the further away those lines will get. Exactly. uh which means then the edge is maybe isn't as strong versus let's say if that number was 16 instead of 15 on on the reshaw or 14 that's when they're going to be closer together. So showing you that regardless of how that data is redistributed and all those trades are mixed up it's actually
15:11 seeing a consistent data. Yeah 100%. And this one like I really like the reshuffle because it's like something very simple that literally doesn't require a lot of computing power but already like helps you to visualize like how dependent were you to that order of trades and gives you a first idea of the distribution of possible draw downs because like this the second uh um common version of the Monte Carlo is the bootstrapping and resampling. So, which is the classic spaghetti charts. I'm not saying because I'm Italian, but it's like it is the classic spaghetti charts, right? Where you have multiple equity lines. And um the difference between the
16:05 uh reshuffling and the resampling is that the ending point of the equity curve it is different. Right? And the reason why is that is because you don't simply just reshuffle the trades but some of the trades appears twice. So they can appear multiple times. Some others don't appear at all. And that's how you reproduce multiple version 10,000 20,000 simulations of the past. So that's where it comes back to your validation. So it's not just a trading strategy, an idea is then validated through these stress tests if you will. Correct. And what is cool as well is that on top of that doing something like this type of Monte Carlo you do get as an output like a distribution
16:53 of outcomes. So like you can understand what is the expected profit of the strategy? What is the expected return if I run this strategy 20,000 times? What is the probability of getting a draw down larger than $10,000? Mhm. And all of that all these possible scenarios they are given by the distribution of outcomes of the simulation which can be used as well like during live trading for example if I know that after 10 trades okay after 10 trades I should expect as maximum draw down
17:44 $10,000 if If I see that there was only a 5% probability of getting a draw down larger than $10,000, like that one would be a red flag for me saying, "Okay, maybe I need to step back and stop the strategy because it's very unlikely that after 10 trades, I get a larger draw down than $10,000." And all of these are yard sticks that are provided by this validation phase that you can use to give you awareness while you're trading live. And once all of these steps are completed, so you checked that you didn't overfeit it with uh where the simplest version is running like a comparison between insample results versus
18:31 out of sample. You did your Monte Carlo to see how strong your edge actually is running 10,000 20,000 simulation. Only at that point we reach the final phase which is live trading. And note that with live trading is not that distution trader sits in front of the monitors waiting for the entry conditions to be satisfied. Just like as we did like we encoded the strategy. So like the live trading is actually all about automation
19:16 where you still leverage your machine to just monitor the data that are fed to it and every time that your conditions are met opening or closing your position on your behalf. So at that point like the question might be like so all this work to have your machine trading and indeed like what is your job right like your job as an institutional trader it is doing this research to make sure that the strategies that you're trading live they do have a positive expected returns they do have a positive value they most likely than not they're going to have a positive P&L at the end of the Okay. And your job is only one monitoring the performance
20:10 and two monitoring the risk. So this is what you're doing while you're trading strategy are trading live. And given the fact that they are automated, you can go back and restart this process of going from trading strategy idea rules and code back test validation and eventually having the strategy trading life. Futures traders, it's time to hear about Apex Trader Funding, the largest futures firm in the industry. They have completely changed the game with their new evaluations. So far, they've already paid out over $600 million to traders around the world [music] just like me and you. Now, not only that, Apex actually holds the record for the largest payout to traders, which is of course with Jadecap, no stranger to chart
21:00 fanatics [music] with $2.5 million in a single payout. And the person before that was Trader Kane with $2 million in payouts. Now, what sets Apex apart? Apex, you can get up to 20 accounts which no other firm offers. Now, not only that, they have extremely clear rules with their revamp. There's no MAE rule, no 5:1 risk-to-reward rule, no hidden rules, no gray areas, clearly defined draw down models. So, you have two options now. end of day draw down and intraday trailing draw down which means no payout denials, no payout reviews, no discretionary decisions and no surprises. What sets them apart as well? Traders receive 100% of approved payout requests up to 20
21:48 accounts as I mentioned and one day to pass. Now the other thing that I love about Apex, not only can you get 20 accounts, but there's different account sizes. So they have 20k accounts all the way to 150k accounts. So you can get 20 of any of the account sizes that you choose. Not only that, you have various trading platforms from wealth charts to trade of to rhythmic. The choice is yours. So make sure you use the code CF [music] to get up to 90% off on evaluations right now using the link in the description below. Trader Apex today. Let's get back to the episode. Look at this. Like usually the re classic retail traders like just as the idea and trades live with real
22:38 money the idea right after on the other end like as an institutional trader you need to follow this pipeline. Yes. And once you have a strategy that has been validated, that is proven that can make you money in the future, you finally trade it live. So this as you to your point, like most people will learn an idea maybe from a YouTube video or just generally from some somewhere. Very rare nowadays that people are coming up with their own ideas, but that's possible too. And but straight away from there, they're looking at uh going live most of the time. At most someone might create a few rules. Someone might you know back test a tiny bit but you know not the level of detail that is necessary to really have confidence and edge within the markets. In terms of the first
23:27 two which I know we're going to focus on like the idea generation and rules but when it comes to the idea generation and those rules um and then getting encoded. So then the encoding side of things is where again as you mentioned that language models can you really help with that for the average retail trader. But in terms of that from that point on a lot of like the back test side of things is automated. Obviously you're you know giving the inputs to make it all happen but the back testing and the validation that's all an automated like being done for you process to give you that data for you to review. Is that correct? Yeah correct. like uh that that's the big advantage. Like nowadays even the average retail trader has access to so many welldesigned tools that allow you to run back test and run like simulations. My favorite one I'm not affiliated or
24:22 anything like that but my favorite one personally is multi charts because like the programming language is extremely simple. It's very easy to understand what the code is doing which is key to understand what you're doing if you want to make money in the market as an institutional trader and then like the back testing infrastructure is very solid. So you can rely on the back test. you have the possibility of run um simulations using your back test like classic Monte Carlos and really like is not a matter that you as a retail trader you don't have the tools it is a matter of really like having access to this process getting to know which are the steps and the
25:09 boxes that you need to check before trading your strategy life and you say that I I love that you said that it is nowadays is harder that retail traders come up with their ideas. uh because um today I really want to focus on the first two steps like uh where do the idea comes from as an institutional trader and I want you guys that you steal this like steal as much as you can because uh this one was something that um really impressed me when I started that uh so many institutional traders were going like in one place to get their trading strategies ideas. The place guys is the social science research network. interest. Okay. Why does this matter? Because 90% of institutional
26:00 trading strategies or strategies applied by institutional traders on any single trading floor are coming from research papers. Mh. And uh the feedbacks that I got like when I started like this journey of showing like the behind the curtains obviously social trading to the broader public was that yeah but you cannot use research paper because it's hard to read them or like the edge is already decayed. Fair enough. But you don't copy the paper. You look at its findings and you look for the underlying reason. the why that strategy can make you or is supposed to make you money in the future only triggers the idea that allows you to frame the rules for your trading
26:49 strategy and this is very important I will write it down that the goal is you don't copy the paper but what you do is analyzing the findings and you look for the why, the reason why your strategy is supposed to make you money in the future. Um I brought you four examples that uh I thought was a good like exercise to go from research paper
27:36 to an actual testable strategy. Mhm. Uh which we can cover one by one. So let's write it down again that 90 to 95% of the trading strategies ideas for institutional traders are coming from research. And in particular like there is two types of research. There is academic research. Mhm. which is research conducted by university professors or researchers within the academia or industry
28:22 which might be hedge fund managers or traders which conduct some specific research around trading strategies like momentum mean reversion or strategies on gold. um all of that and they publish them online. Website like the social science research network is like this giant collection of all this research which is 100% for free where anyone can just go and their job is to dig and trying to find something interesting. M again it's not a matter of reading 40 pages because just like for the encoding you can leverage your machine leverage the help of large language models to crack this
29:12 uh these uh papers trying to understand okay what is this research about what are the findings why is it supposed to make money you don't copy the research one to one but it is an extreme important starting point to define your trading strategies rules. Which are the rules? Let's move there and then let's have a look at the examples. Every single trading strategy needs to have three things. Mhm. An entry. So the conditions on why you should open a position, an exit, why you should close the position after it has been open. Mhm. And position sizing.
30:08 Every single trading strategy needs to have needs to have these three boxes filled. So you need to know why you're opening. You need to know where to close and generally is having a take profit stop-loss. And a third condition, one that I find very often across trading strategies is having like an exit linked to time. So, for example, closing a position after 1 hour or at a specific point in time like 3 p.m. And this one is satisfied if you don't hit your take profit nor your stop-loss. Okay. And the third component, position sizing, which is something that I feel
30:55 like is not discussed enough across uh uh retail traders because it's really like knowing how you size your trading strategy, your position can make a huge difference in both reducing the risk of blowing up your account. Yeah. And uh improve the risk adjusted return of your trading strategy. We will have a look at some very simple examples on that one as well. Tradzella is the all-in-one platform built to turn you into a profitable trader. And at the center of it is an AI co-pilot [music] that does the heavy lifting for you. It watches every session. It learns exactly how you trade and your strengths and your weaknesses, the patterns you keep repeating without even knowing it. And it tells you in plain language what's working [music] and what
31:42 needs to change. The moment you close a trade, your AI captures it automatically. Fully sync to your broker or prop firm. Zero manual entry. You can replay any trade tick by tick, pip by pip, to see the full picture. Back test your strategy on years of real data before you ever put up real money behind it. Your AI co-pilot ties all of it together, surfacing insights you'd never find on your own and guiding every decision you make going forward. Hundreds of thousands of traders, one AI powered platform, [music] real lasting results. Now, if you want that AI co-pilot and see your trading excel, the link for Tradzella is in the description below. Use the code CF20 for 20% off your yearly subscription or CF10 for 10% off your monthly subscription. If you're a trader, you
32:31 have to know about Chart Academy. Chart Academy is the first all-in-one free trading education platform. That's right, absolutely free. No catch, no upsells, no links to sell you something. All free. proven world-class traders providing an A to Z masterclass on their strategies, their concepts, and their processes that make them the best. From futures, options, crypto, stocks, and forex, everything is covered on chartacademy.com. Whether you want to fix your trading psychology or learn from a market wizard directly, you can do it all on Chart Academy. Not only that, by just spending time on Chart Academy and bettering yourself as a trader, you can get points that unlock exclusive discounts, free trials, and free accounts with your favorite prop firms and trading softwares.
33:18 Don't take my word for it. Chart Academy is live. The link is in the description below. Now, let's get back to the episode. But now that we have all of this, now that we know that the rules needs to have an entry, exit position sizing, we need to have a reason why our strategy is supposed to make us money. We can move forward and have a look at the first example. Definitely. Let's do that. Should we do that? Yeah, sure. So, let's start from the first paper. And the idea here, I really want to show you how to go from a research paper to an actual tradable strategy. So let's start from the first paper which is market intraday momentum. So market intraday
34:11 momentum and this one is a paper from 2018 a paper that you can find on the social science research network 100% for free and the reason why I wanted to take this one as first example is because at the end of this example You should be able to understand why one of the most popular retail trading strategies, the opening range breakout actually worked so well over the past six years. As I said, like the first thing that we need to know about any single paper is the findings
34:59 and the why. why this strategy is working, why they found these specific findings within the paper. So the findings of this paper is that overnight gap plus the first 30 minutes of trading from 9:30 to 10:00 a.m. It reveals the imbalance between buyers and sellers and it predicts the performance of the last 30 minutes of the regular trading hour. What does it mean? It means that if you had a return a positive returns between 4 p.m. to 10,
35:50 you should have a positive return even in the last 30 minutes of trading. This is what they analyzed in the paper. This is the finding of the paper. This is true. And then you need to ask yourself, okay, but why? The reason why is because after the market is closed, there is all a new set of information that gets released. Yeah. Like uh news, earnings, global positioning. Mhm. So news plus earnings plus global positioning which creates all of these orders between buyers and sellers and these
36:39 orders collides in the first 30 minutes of trading revealing the imbalance. Yeah. And this imbalance if there is a lot of buyers over sellers the market is not able to digest it just in the first 30 minutes and this imbalance is carried on in the remaining part of the day and that's why you have an intraday momentum. Yes. Is it clear? Yeah. Yeah. And uh now that we know the findings, we know what they did like okay you check what is the performance
37:24 in the first 30 minutes if there is more buyers more sellers and that one allows you to predict the last 30 minutes. You might be tempted to try to develop a strategy taking the entry, the exit, and the position sizing of the strategy and replicate it. But if you try to replicate it, you're going to have a very nice equity curve up to 2018 followed by a very poor equity curve because you would be victim of the alpha decay. Yes. However, we don't copy the paper. But we can take the fact that the imbalance gets revealed a finding
38:13 that they have in the first 30 minutes of trading setting up an entry of a classic opening range breakout. But now we know why it is supposed to work because overnight there is all the set of informations and not all the players all the market participants are going to trade until the market is open. Yeah. And we can say that entry condition open a long if the price closes above the high of the range between 9:30 and 10:00 a.m. Right. If the price closes above the high of the range, we open a long position. Mhm. Exit. You could say
39:05 for the long the same like putting a stop loss at the low of the range. So would be somewhere down here. Right. So here we have our stop loss and for the takerit a classic reward to risk ratio either between one or two. Okay, this needs to stay simple because we need to have the control over what we're doing. take profit. Let's say we have here the eye of the range. So would be let's say if we take one to one
39:56 the TP would be up here. Mhm. And we know that we are going to open a position only if the price will reach or will close above the high of this specific range. And we can add even an additional condition like closing the position if we don't hit nor the stop loss nor the take profit let's say at 3:30 Eastern time. So a very simple set of rules but very well defined. Yeah. That allows us to translate it into a tradable trading strategy. Last one that is missing is position sizing. Yeah. What I always suggest during the development phase is to use a fixed amount of contracts of one. Why?
40:47 Because makes it simple to develop the strategy. However, this is not what I would suggest to trade live. Because when you have a fixed amount of contracts like one, if you have a very wide range, Yeah. So high volatile days like you're going to put at risk something like $30,000. Yeah. On the end in narrow days you're going only going to put at risk let's say $1,000 and academia research again with the two very important papers and this is something like it is done so often on the trading floor is to make your position sizing conditional to the volatility of the instrument.
41:32 Yes. The two papers I'm talking about is volatility managed portfolios and the second one is the impact of volatility targeting. What is the volatility targeting? Volatility targeting is to have your position sizing where you always put at risk the same amount of money. Let's say $10,000 per trade. Mhm. [clears throat] But the amount of contracts that you're going to trade with is going to be based on the volatility of the underlying. So like if you have a very wide range. Yeah. So a highly volatile day, you're going to use a lower number of contracts still risking 10,000. Yes. Okay. M on that trend like if you have low volatility day with a very narrow range you still
42:24 put at risk $10,000 betting or like trading with a higher number of contracts. And in this way, you don't have that highly volatile days are going to dictate the outcome of your trading strategy, but overall, they're going to smooth out the outcome of your trading strategy, improving by definition, the risk adjusted performance of your trading strategy. And this is like something like that is never discussed across uh retail traders but it's such a small simple change that can highly impact the performance of your trading strategy. And this one is just one version of the strategy. But given the fact that right now we're just brainstorming on what is the right strategy that we should test.
43:10 We could have a version where we have a short version. So to go short if we break the low of the range. Yes. However, this one I tell it from the experience is that when you develop a strategy based on the imbalance between buyers and sellers in the first 30 minutes, that's works better on long go only strategies because once you start breaking down the level, yeah, you have a lot of buyers stepping in, removing that imbalance so easily on the trend like on trending case like it's hard that more buyers steps in and you have this drifting effect making this strategy fairly effective. Mhm. We didn't copy the paper. It's not that just because the paper saw that uh we if
44:02 we have a good performance in the first 30 minutes we're going to be long in the last 30 minutes. But we took the concept of taking this imbalance gets reveals itself in the first 30 minutes to start defining properly the rules of a strategy that we can actually translate into a piece of code. Mhm. And that's like key on this process of going from research to a testable strategy because now like even if you're trading from home an opening range breakout now you know that the reason why your breakout is actually has been working fairly well over the past five years six years and not that this paper is from 2018 like it is because you have informations accumulating overnight
44:53 Yeah. Imbalance revealing in the first 30 minutes and that's what is pushing the price in your direction. So now you start getting like more knowledge, more awareness around the why your strategy should work and the why you're doing what you're doing. Mhm. This is what I really find fascinating about this process of justifying what you're doing based on leveraging the work of people that are much smarter than you at the spent. So, well, they say like your why is so important, right? And just in life, let alone in terms of trading like knowing why you're putting a strategy to use or why you're planning to trade a strategy is so important. You know, like you said originally, a lot of people have maybe an idea, but usually that idea is just
45:40 something they found, you know, come across and there's no validation, no trying. But this is not only got a thesis behind it like an actual academic thesis but taking the understanding of what that's representing as you mentioned you know with the uh overnight action and all these orders and participants waiting to step in and because they're waiting that then creates as you say an imbalance in price and our job a lot of the time in terms of edge and profitability is because you notice an imbalance in price and so you know it's showing you that full process step by step as you say and [snorts] as you say the opening range breakout is very popular. I'd say over recent years in particular. But one thing that is cool about um this and over opening range breakout like in general is that they are part of a structural limitation of the financial world in a way that
46:30 there is so many players that are only allowed to trade after 9:30 Eastern time. And that's why they don't place orders before because they can't because if they were allowed to place the order before like this market anomaly the fact that there is like the reveal of this imbalance in the first 30 minutes would disappear. So knowing why it is working like let's say things changes and this uh usage fund or some pension funds they start trading after or like before 9:30 a.m. the effect of this one might disappears and knowing that it's working because of that they lose that
47:16 inefficiency. Exactly. M so like uh gives you a complete different perspective over what you're doing and why you're doing and why you're making the money. Next example. Do you have any other question on this? No no this one this one makes sense 100%. you know, like I said, because it's a a very widely used strategy, maybe not properly tested or data collected, but um that's I guess one of the interesting things is that yeah, like uh if it's a if it is a validated strategy um even if you didn't I guess that's a strange one, right? Maybe an idea is validated not by you, but someone's obviously validated it to be be able to publish or someone potentially has validated
48:01 it, but you're just using the idea of it. Uh just the concept that you've learned online, you could still be profitable, but you might not have the confidence that you need. Yeah. Exactly. Long long term. Exactly. Like if you on top of that like you get the confidence over what you're doing like it changes your perspective on trading completely because like you might get like let's say we develop the strategy. By the way guys the academia is saying that the first 30 minutes are the range that is optimal to take advantage of for any opening range breakout. Yeah. So and that one is like it's not monte they saying that I invite you to develop a strategy based on the rules that we discussed before to confirm that obviously the long only type of strategy works
48:52 better and it's not just a matter because of the beta of the market that obviously if you have long strategies and the market is just going higher you get a better performance because that strategy performed well even in 2022 when the market was bleeding and the other thing is like the risk toreward ratio to use is one one or one to2 like that one is like the optimal that you can find for for this strategy yeah for this strategy for academia but again the the the goal is to gain the independence of going and testing it yourself like trying to get the idea frame it with entry
49:39 exit and position sizing and then use large language models to get versions of your strategy and try to test it yourself because then like you gain independence from anyone like you can go on the social science research network. Spend the full day finding all the strategies that do you want testing as many strategies as you want and if you keep digging you find gold. Of course that's how this game works. not only at home like in every single trading floor on earth. The second strategy or better the second paper I wanted to still be somehow close to something that is still understandable across all the retail traders which is a paper from the industry this time.
50:33 So the previous one was purely academic. Mhm. This one there is an industry paper called VWOP, the holy grail of day trading systems. The biggest difference between this one and the other one again is that this one is based on research and there is a beauty in that as well like is very practical. Mhm. Because you have the price, right? So we have our price and our time and from the price and the trading activity at any given point in time you can get the VWOP which is the volume weighted average
51:24 price. Yes. Say differently, the VWOP is the average price of everything traded up to any specific point weighted by the volume. Mhm. Is it clear? Yes. Yes. Cool. So this one is our VWOP and this one is our price or the stock or of the index. Again we need to go back to which were the findings of the paper and we will need the why. The findings of the paper was that if
52:12 you go long when the price is above the VWOP and you go short when the price is below the VWOP, closing the position when the price crosses back the VWOP. So as exit we will have and we will repeat this later but as exit we will have that when the price is equal the VW whoop we close the position we can get very good returns like we can get systematically good returns and what the research that they did was on QQQ which is an exchange traded fund. Mhm. tracking the performance of NASDAQ 100
53:07 and they used a time frame of 1 minute. So instrument they use QQQ on a one minute time frame. Every time that a one minute bar was closing above the VWOP was getting longer and they were closing the position when the price was closing back below the VWOP. Yeah, exactly. With the VWOP. So exit the position. Let's put it with a and the same like going short on the flip. Yeah, exactly. on the flip when um one minute bar was closing below the VWOP and closing the position. This one would have been actually a losing trade but closing the position when it
53:54 was crossing back to the to the VWOP line. The returns that the fund was actually like quite impressive. You can find it in the paper. Something like in the order of 671% over 5 years. But we need to go beyond the headlines. We don't care about the headlines. What we care is like understanding what they've been doing and why. Why is it possible that when the price crosses below or above, we have like these directional moves that allow them to generate 600% over five years. The reason why is behind the VWOP Argos.
54:43 So every single institutional traders that I know at least once if not every single day they send an execution algorith or they send an execution order to the market. They often use VWOP algos which is a type of algorithm that tries to track and get the same execution or as close as possible to the VWOP price. Yes. Okay. So it's like a so it is a volume participation algo. If there is a lot of volume traded at a specific point, the algo will send more orders at that specific point in time and that one causes some heavy directional moves because you have a lot of buyers or a lot of sellers that try to track the VO price and in the moment like
55:35 you have this crosses between above or below the view price. The fact that the vast majority of institutional traders are using this type of algo creates like amplifies these directional moves. So now we know that it might make sense to follow a logic that is related to this um uh vivop logic. They did the research. They already tested a strategy a version of a strategy using a one minute time frame. Yes. So we could do exactly the same. What I would change as first thing to get a little bit of context and a little bit of different ideas is that we don't need to deploy the same strategy on QQQ but we could might decide to test the strategy of ENQ so NASDAQ 100 futures
56:27 if we want to use the same underlying so NASDAQ 100 but you could test it you could test the same on ES you could so S&P 500 future You could test the same on crude oil. You could test the same on single stocks. Like again, you're writing the code once and then you can apply that code to whatever instrument you want, right? As long as you know that you have institutional traders using this specific algos on that instrument. Yeah, like this logic is supposed to work across all the instruments, right? This is the first important takeaway. And then like obviously you need to define your entry. So again it's going to be to go long when the price closes above the VWOP
57:18 to go short when the price closes below the VWOP. Mhm. exit. We can decide to simply have one exit condition. I'm guessing the VWAP is uh plotted on from the open. Yeah. Yeah. That one is another thing that you might decide like you can decide to use uh the VWOP from the open or you can decide to use the previous day VWOP incorporate as well. Those are like the type of parameters that is up to you to use. What I strongly suggest is to use it from the open. Okay? So to start uh uh considering like tracking from 9:30 a.m. If you're trading US or some markets here in Europe
58:11 like starting tracking from 9 and then like uh after you get some data in start like checking what is the good uh level at which uh um or better and and then like trying to understand even during the back test what is the number of minutes or hours that you need to have which is the optimal that allows you like to benefit the most out of the most volume I imagine because it's from the start of the day. Exactly. Exactly. And uh lastly like obviously like uh the position sizing [snorts] and just like we discussed before always use one contract in development process. Uh if you
59:03 are testing for stocks, you can might decide to use a percentage of your portfolio. Let's say you have a $100,000 portfolio to allocate 10% of it just to see how it would behave or again like uh having like uh your position sizing conditional to the volatility of the underlying instrument just we have seen before. But this is again in the very first phase development phase always one contract and then you can test different version. Yeah, different version like afterwards once you make sure that the strategy that you Yeah, exactly. that you have on end is actually pointing to the right direction of profitability. Next one. Yeah, we can. Yeah. So now we have seen
59:54 two examples and both of them I decided to take them because is topics that the classic retail traders release a lot and on that right now we start like going like with the last two examples um with stuff that is one of them more academic which is the one that we're covering now. Yeah. And the difference between this one and the previous one is that this one is the results of a collection of research papers. Okay? So like multiple interviews. Yes. Don't know if some of you already heard of it, but this one is about the post earnings announcement drift. The logic of the post earnings announcement drift
1:00:42 is that you have the release of the earnings. Yep. Which is set by this vertical line earnings. And what has been found over literally like 60 years of research. This one is one of the very first inefficiencies properly documented. Okay. And still alive today. So it's like to the ones I say ah alpha decay. Yes there is alpha decay but there is always a different angle you can approach with always basing your your knowledge from research. You have to get creative essentially. Yeah a little bit like you need to that's why the advantage of testing different things and say okay this one is not working on this specific type of stock. Yeah.
1:01:31 but maybe can work on a different type of stock. Okay, which we will see with this one as well where there is like a very good insight. But like you have the earnings so until the earnings day the stock can do whatever let's say that is rising in price then there is the earnings release. What the post earnings announcement drift is saying on a very high level is that if there is a positive surprise in the earnings and as a consequence a positive price reaction. Okay. Yeah. For the following days and weeks the stock keeps drifting higher. Would it be the same in reverse? Exactly. And the same in reverse. If there is a missend earnings
1:02:21 or a negative price reaction the day after of the earnings, there is a negative drift that keeps going for the following days or weeks. Mhm. And here are [snorts] the findings of the research which split it across three papers. So the first paper is from 1968 and is an empirical evaluation of accounting income numbers. Empirical evaluation say [laughter] numbers. Okay. And uh this one was the very first um observation that uh where they saw that if there was a
1:03:09 positive increase in value the following days and weeks the price was keep drifting higher and this one was like its purest form of the post earnings announcement drift or drifting lower when there was a negative price reaction. Yeah. And this one was the first contribution to this uh uh price anomaly. Yeah. The second one was a paper from 1989 which is delayed price response or risk premium in full was pied the delayed price response or risk premium question mark. delayed price response or risk premium. And the biggest contribution of this paper was noticing
1:04:00 that this drift Yeah. was lasting for around 60 days. Okay. So 60 trading days which is approximately exactly 3 months Okay. Yes. So including weekends. Exactly. Exactly. Um the third paper was a paper from 2006 called comparing the post earnings announcement drift for surprises calculated from analyst and time series for that's a very long yeah but but the main point of this one was the introduction of the analyst consensus and the general obviously we'll probably have links in the description and we've shown them on screen but for just the general general idea
1:04:46 of how much detail these papers have. Like one of these papers, how how long and detailed are they? They are like 30 to 60 pages. But that's the key of taking the paper, downloading it, feeding to the machine and then extracting exactly what is the finding. Okay, what is the reason why? Because for some people that might not sound like a lot, but in reality, you know, that's on one topic. There's literally one one thesis on one topic that's a 30 to 60 page. But but that's and that's like and this particular one has multiple papers. Yeah. And you need to you need to start changing like the perspective that it fits a lot is because many people spend so many hours studying that that
1:05:36 the finding is reliable and you don't need to spend months on just to understand the finding. So in reality, you don't need to because of these papers and and their availability as well is that you don't need to trust it in the or you don't have to worry about trusting it because the research has been done. Yeah. To your point, it's like okay, you just got to find the angle, you know. Exactly. Like what where it's at now. Yeah. 100%. Like research has been done. Often these papers are published in financial journals. So like they are peer-reviewed across different academics. So like they they they say if it turns up being published on a journal means yeah what they found is actually true. And uh then like it's up to you like to take like each one of these
1:06:26 elements. Yeah. And combining it on a working trading strategy like again on the first one we have the price reaction. So if we have a gap up on the following day Yeah. from open to close. Mhm. um you might have like the first layer of the post earnings announcement drift. The second one you know that it lasts for 60 days. The third one is introducing the concept of the analyst. Yes, the analyst what they do is before the earnings are released they give an estimate of the earnings per share. The insight that they had was okay let's not compare just on the price but let's
1:07:13 see what is like the drift if is stronger the drift if the actual earning per share is larger than the estimate of the analyst. Mhm. So if we have a larger or stronger positive drift in this case and if the earnings per share is below the estimates from the analyst. So if there is a miss in the earnings is the negative worse. So it is more strong the negative reaction. And today this is like the standard. So not just using the price reaction but keeping consideration
1:07:58 what is the actual earning per share versus the expected earning pressure as well. Okay, these the insights from the papers. Now the why why is it that not the informations are reflected immediately on the price? Because we need to keep in mind that if you go to any university course they will talk to you about deficient market hypothesis. Yes, the efficient market hypothesis is saying that as the news are released, all the informations are reflected immediately on the price. Mhm. But this one like in front of us,
1:08:44 we have 60 years of research. They say no, it's not the case because if that uh was the case, you wouldn't have this drift over the following days interest or weeks, right? Mhm. And the reason why is because one there might be a low coverage of the stocks. Okay. So not all the analyst might cover properly all the stocks. So we can say that information travel slowly. Yes. And lack of awareness almost. Yeah. Exactly. like if especially like if we focus on small cap or
1:09:32 midcap it's not that these stocks are as much as follow as mega cap so Tesla invidia Apple because they're like the information is out it is reflected right away indeed like newer paper from 2021 show that if you try to apply I post earnings announcement drift strategies to mega cap you basically have no edge. However, if you apply it to small cap or midcap, the price is not reflected immediately because of the analyst having a smaller coverage across the stocks. And the second reason is liquidity constraints.
1:10:23 constraints. What do I mean with that? Yes, maybe there is a stock that is followed by a fund manager that would like to build a larger position, but maybe it's not able to create or allocate all the capital that wants to allocate on a single day because there isn't enough liquidity. So it decides to spread the execution of the order over the following 10 days. Yeah. Or over the following couple of weeks. And that's what generates the drift. This drift on the upside or the downside. If we now try to take all of this and translate it into a trading strategy, we need our entry
1:11:15 which again can be if we have a positive price reaction. So the price increases and we have a bits on the earnings. So the actual earnings per share is larger than the expected earning per share. Then we open a long position at the opening of the following day. Okay. Exit again we take it from research suggesting that 60 days is optimal. So let's use 60 days position taking them as starting point. Yeah. as suggested starting point and as position sizing.
1:12:08 You could do something like either a percentage of your portfolio. So allocating for every single stocks that you're monitoring let's say one to 2% of your portfolio or a specific dollar amount or again like in this specific case you could do as a position size conditional to either the price movement or like the deviation between the earning per share the actual earning per share with the from the expected earning per share. [snorts] And you could try to check and and you could try to test these different variation to understand which one has the best risk adjusted returns. We in this particular one cuz like with the VWAP uh it was un it's quite easy to understand where would your
1:12:58 stop be right and where would you look to exit that trade as we already saw with the diagram. Uh and same with the opening range breakout with this one in particular. What would that look like? because you would essentially maybe be the low of the previous day. No, not necessarily like um that one is needs to be tested like uh for that purpose like given the fact that here we have um cross-sectional type of trade and monitoring let's say we're monitoring 20 to 30 stocks. The simplest starting point for an exit is just focusing on time. So you take the time which was suggested from the research and focus on that. Afterwards you might add an additional layer for risk management purposes where you say okay let's put like let's try to put like 5% lower
1:13:51 than the opening of the day or like if we have a gap up like using the closing before the earnings got released. Okay, as a stop-loss for example. But these are just like examples that you would you would you think that these are things that you would when you get to the sort of encoding stage and so on you would have these sort of rules so you can do your back test and everything. Yeah I would start for the big test I would start exactly like with this simplified version. to just using a time exit of 60 days. And then like I would once I divide between in sample data and out of sample in sample is where I would add different rules like for example what would happen if I
1:14:36 include a stop-loss. Okay. Would improve their risk adjusted performance. Okay. There is where like you test people. Exactly. Because um why do you do that? Why do you do this split? Because you don't want that you just c fit your strategy. Then you notice that if you add a 1% stop loss, it is the best possible results that you can get in sample but then like you apply it out of sample on unseen data and in that way like you understand did I over fit or was actually the best possible version. Understood. And so it really is all a process that you need to follow to make sure that uh um you're not fooling yourself, you see. And um yeah and one one interesting point as
1:15:28 well like obviously as I was saying this one is a strategy that nowadays is working the best on small cap and midcap but not necessarily like you need to focus on small cap and midcap in US. You could take the same strategy and applying on the European stock market, right? Because you might find that in Europe there is a better performance across I can imagine maybe the analysts are even less maybe. Yeah exactly like there is less coverage there is less interest there is less um maybe even the liquidity aspects in terms of the constraints. Exactly. So like see you're funny like getting there. You're getting there. You're getting there. Like all of that like is
1:16:16 all points that you really start. This is the new way of how an institutional trader thinks of like connecting these dons thinking exactly just like you did right now thinking yeah there there is less liquidity in Europe so maybe I could apply this strategy on European names rather than just focusing on US names and um and um on the one thing that I want to mention if you want to try to develop strategies on the postix announcement drift that the short side is becoming less effective for one simple reason that when a company has negative results Yes. they tend to pre-annon announce it. Yes. Mhm. So that one is making a huge difference nowadays
1:17:05 that you have the CEO saying yeah the numbers are going to be poor. [laughter] So like you want to get ahead of it. Yeah. doing doing a lot of risk management EPR before end and that's why you do have try and soften the blow. Yeah. And that's why you do have already like already some down a negative price reaction before the earnings is announced and that's why you so those gaps won't be as dramatic potential. Exactly. Exactly. And even the consequence might be not as dramatic. Yeah. It might be a 30-day thing or or just sideways because it's already priced. Exactly. So essentially you're trying to the thesis of this the foundation is that it's something that's in a surprise to the market almost. Exactly. versus uh if they're announcing
1:17:52 it then it's already going to start getting priced into them. if they start putting some announcement of yeah it's going to be a bad so like doing some management of the expectations that's where like the p tends to die out and this is a tendency that the cos tends to do a lot before negative numbers I've seen some cases even in positive numbers that they started like saying yeah not the numbers are going to be much better than what we're so then it wouldn't be valid for that and that one exactly when you will have um so I guess like even though it's the CEO it's almost in the same category as the analysts right if the analysts are covering that oh you know this is a well one they're covering it a lot and they might be even covering it saying hey this is going to be
1:18:38 negative it's going to be this giving their thesis uh again it will start to get priced in it will start to be expected uh so similar thesis if the CEO is talking about it on the positive side as well it's not going to be as much of a a shock if you will in the market and you probably won't see as strong of a gap. Maybe we still get a gap, but not as strong. Sometimes you might not even get a gap because it's already been spread out there to the masses and Exactly. Exactly. But again, on the positive side, we see it rarely, but can happen. It is something that you just need to be aware of. It's almost like because the CEOs don't want a very drastic negative uh impact on their stock because that won't come across well. But they love no doubt a a surprise positive announcement because then you know it's all in the headlines. Oh, the stock's up 10% today.
1:19:26 100%. 100%. And it is just something that it's a good to know if you're developing strategies like that. What's interesting as well though is uh these strategies so far, they're not complex, you know, they're not like uh something that's overly hard to understand. It's not even hard to understand necessarily. And a lot of the time when people talk about institutions and institutional trading, they automatically probably assume that there has to have crazy technology,
…
