The Big Picture
Recently (but always too late) I saw this wonderful scam course on Udemy: https://www.udemy.com/course/cryptocurrency-algorithmic-trading-with-python-and-binance/
The first question, of course, is why don’t you just run your code and become a millionarie? Oh, you love teaching, I see.
The same question applies to all the authors of books like https://www.amazon.com/Understanding-Crypto-Fundamentals-Cryptoassets-Management/dp/1484293088
Why don’t you just apply the “wisdom” you are selling?
Here is a non-bullshit, less wrong answer.
Have you ever seen any of these boxing or Muay Thai teaching videos? Some very competent and well-versed guys explain slowly and methodically what you do when a jab or a hook comes at you!
The best of them even use the notions from physics to justify their tactics, and of course, they telling you that only through practice and automatic responces (whithout thinking) one could master these techniques.
How, why don’t they just go into the ring and win a title?
Well, it seems that there is some gap between the theory and “teaching” and the actual performance inside the ring.
It seems like very different risk/reward ratios are involved and that it is not so straigforward as merely “do what I said”.
The course that “teaches” you how to load data in Pandas and “analyze” it never mention that once you get the data it is already “the past”, and that the past performace is almost unrelated to the current market conditions.
(Well, the patterns on the charts obviosly it influence the decision making of individual players and thus contribute to the trend formations, but these are not the primary marker forces).
So, the first thing to understand is that you are going to go inside the ring and meet the actual jabs abd hooks, while all the “teachers” and “coaches” would stay aside and watch you got bruised.
This is the main point. Actual trading is fundamentally differnt from all the courses and all the books, which, at best, explain the underlying principles, social dynamics and show some useful techiques. That is all.
Now it is time to realize that the chances and the efforts required to succeed in trading is, in principle, comparable with those required to win some title in boxing or Muay Thai, and for that you better have some genetic advantage.
The only good news is that one could be really clever and use computer automation to prerform the mundate tasks, which has to be performed right on time and flawlessly.
Automation is the key.
One could use ready-made commercial on-shelf tools or make one’s own. Arguably, the second approach would make you an expert in understanding (from the principles and the high-level concepts down to the lines of code) of what you do.
This, however, requires substantial investmet of time (and therefore money – bills has to be paid) and actual, serious intellectual effort. Basically, you have to program your own Trading Assistant (avoiding all the current AI hype, bullshit and memes).
This is an eqivalent to winning of a title.
The deep pockets
Just as it is techically impossible to feed oneself form an two acre plot of land, it is impossible to successfully trade with a small account.
The name of the game is to win more than you lose on average and to be prepared for a rare luck (Fortune favours the prepared). You also have to be prepared to sustain ocassional substantioal loss - a scam wick, a market manipulation, a sudden FOMO euphoria or a panic selling.
“But once in a while it is a dildo”. You will be liquidated quite often and you have to have the funds so that you woun’t be washed out of the market. There is no alternative to have a deep pockets.
There is another principle. To “get even” you have to win substantially more than you have lost, due to the fees, slippage and being stopped out of a position (forced to have a shitty re-entry, which will be stopped-out again).
In other words, you have to have the ability to take the blows (all these jabs and hooks), quickly recover from them and restart again and again. This is how one wins a fight - by being able to sustain the blow, quickly recover and restart in hope to get lucky and hit it just right.
The Luck
This is the game of luck, and it favours the prepared, well-trained, properly skilled and personally experienced. the coaches and trainers will lose.
Unskilled and unprepared will waste a rare luck (a chance) and this is why 90% of the new traders being washed away from the markets. Even pure luck is not enough in this game.
On bullshit
Bullshit, bullshit everywhere. The most common kind of bullshit is due to applying abstract concepts in inappropriate contexts, especially from the probability theory and other abstract branches of mathematics.
Yes, there are the notions of a volume, /speed and momentum, and therefore some derived notions of a slope (derivatives), pressure, and what not, but the causaliy of these notions are different from “stable” and “consistent with itself” physical world.
What causes the movement of the price is differnt every time. There are, however, some recurrent patterns which eventually emerge again and again due to similarities of the actual causes. All the causes are social, after all.
So, what we have is an immence, overwhelming, dynamic complexity. Not just that, markets are keep evolving with each and every change or event. It is a evolving adaptive social phenomena, which we could observe only “looking backwards” (as if sitting on the train in a backward seat).
This is the fundamental principle - you only able to actually observe what just happened (the events which are already occured and their immediate consequences). These are the properties of the environment you are locked in.
What to program.
A reactive system (which reacts on emerging patterns and events), never predective (which trying to predict the future).
Another fundamental principle is that there is no way to do anything meaningful in a /discrete but partially observable, stohastic environment, which keeps changing and evolving. This is not a typical fully-observable idealised environment of a board game or an oversimplified theoretical model.
The only way, however, is to use an over-simplified, incomplete but less wrong model, as if it adequately captures all the relevant aspects of the actual environment (which is not and never will).
This is what we have to program. A model and an agent (an assiatant) with acts on the model.
Notice that all the Deep Learning algorithms would fail due to a non-stable, dynamic environment, as if the language you trying to understand keeps changing in the real time, including rules of the grammar. Sometimes a proper philosophy beats abstract bullshit.