This is the real process to learn the trading from which you are going through . Trading is not that easy bro , you have to work hard for this . It happened with me also. After reading a book of 250 pages and after watching many videos of 1 hour or 2 hours on stock market or trading ,when you found it worthless . It increases the frustration and it also lowers the level of confidence . But you continuously have to work on it , then you will reach at valuable content . Trading or Stock market is a total game of mindset . Keep working on your mind also . You will see the unexpected results very soon . BEST OF LUCK
Why start with algo trading? Algo trading should be the last step, the finishing touch. What algorithm are you going to use? You need first to understand the market, and find out what works and what doesn't. Algo trading should be used when the trading system is complete, consistent in any market, and profitable. Only then you can start to automate everything in an algo. I compare what you do with building a car. You start with constructing the automation of building the car, but you even have no clue what you need to get a car first. You first build a prototype, test it and if everything is successful you can start the automation (which is the algorithm in trading). What will you include in your algo? Why? What's the logic?
i wish i were a master coder if i were starting out though id probably just try to code up simple strategies without caring if they work or not and build from there.
Hey guys. Thank you for all the responses. I greatly appreciate it. Here are some of the questions people have asked me so far: 1. Why do you want to learn algo trading? Shouldn't you learn discretionary trading first? Quite simply, algo trading seems more interesting to me. I am more interested in building a product that works for me instead of being the product that works. Discretionary trading is fine but I think that a quant strategy that is constantly optimized by an experienced developer should beat a discretionary trader. Some people will disagree with that and point me to studies where quant firms underperform discretionary firms. That is fine but there are successful quant firms and there must be a reason why so many companies still invest in quant. There are successful quant traders and quant books. Ernest P Chan exists for a reason. I also like to write code and I love python. 2. What are your goals? a. Choose a data provider, be confident in it, learn the api, and be able to download some historical data. I've several solutions, such as OANDA, Alpaca, IB, iex cloud, Yahoo Finance, etc. Either the support/community is lacking, the documentation is lacking, the amount of instruments offered is lacking, or some other reason. I'm looking into TwelveData and Analyzing Alpha now and those 2 solutions look promising. b. Create a basic strategy, back test with a popular framework like Backtrader, modify the strategy, repeat. c. Continue developing different strategies. When ready to deploy into production, create a method for determining which exchange to execute orders. There are many exchanges and there must be differences between them. What are the differences, and how are they relevant to my strategy? d. Use strategies in production. Continue to optimize strategy and create new ones. Retire strategies if necessary.
You have much too high expectations and you don't seem to understand that an algo is only like an excel. What you put in it will define the result, excel cannot invent a system on his own. It is just a tool. The same applies to algos. You follow what I call "the lazy approach". You do nothing but wait till an algo will produce millions for you. My guess is that you will have to wait many years, or maybe even never get anywhere close to aprofitable system. Also don't compare yourself with the successful funds. You have no idea how much work they have to get where they are. There is a reason why so many funds have bad returns. The reason is not that they are stupid. You do realize that optimizing means using past (known) data to "predict" the future behavior of the market? The fact that you will always have to optimize confirms that the system does not work in future. If it would work, you would not need any optimization. You are always running behind the facts. The optimization needs to adapt the failures of the algo.
A baby learns to crawl, stand, walk then, run. You are just jumping in heads in without being ready for it. Learn to be a competent trader, at a minimum before, designing any algorithms on trading? Or if you are any good at coding, focus on working on coding and make a living out of it. That would make more sense.
In order to code a trading system, you first need to have a trading system ie. know how to trade. You're putting the cart before the horse. Also, take a step back and consider that millions of people like yourself are trying to pull cash from the market, by running simple algos and "optimized" backtests using exactly the same public retail data feeds, tools, execution engines and so on. To make profits you need an edge - what's yours?
Yeah, being able to code is not that much of an asset when you don't know what to design. Coding and development are the least of your worries, and probably a distraction. Trading by code, as a baby step, means getting a FIRM grasp of orders, order management and execution. This says nothing of how to use those tools. Steps should be: 1) Learn to manually execute some strategy, down to the tactics to make it profitable for a few trades. 2) Spec it out, and code it up to see how deficient a couple a trades are at handling 2+ years of situations and outcomes. 3) See if you can overcome those deficiencies with a better, realistically implementable, maintainable, debug-able, strategy and tactics. Repeat. You will discover a rabbit hole of possible tools to measure and test those steps above. Still not even talking about any trading skill yet. Then you will need to wear two hats, developer and trader. Currently it sounds like you are neither. You might be able to write code, but developing productions systems is another whole deal. BTW: stop complaining about old UI and systems. Many people used those tools effectively when they were useful to the markets at that time. If you were around during those times, it would be easier to see how they fit in, UI aside*. Stop focusing on what you know about and focus on what you do not have knowledge about. Trading. Best of luck * Bracket trader circa 1990 is an example.
Most discretionary traders are not successful. Neither are most algos. (that's my story and I'm sticking to it.)