How to research powerful systems for Algorithmic Trading
When we talk about algorithmic trading, we talk about trading with a fixed set of rules, and those rules should make profitable trades over time.
Sounds easy? Well, it isn’t.
Research is the most important part of algorithmic trading, is where we learn new concepts, possible new rules and how to analyze time series in order to find edges.
There’s some kind of not written rule over the Internet, and that is: Most of the information is false or only partially true. This is due a lot of factors such as SEO (Positioning a website creating content), CPA (Commissions paid for doing an action) and mostly ignorance, really.
He have more than 7 billion persons living on the earth, half of them have access to the Internet, and there’s a huge portion of people that makes a living on it. And that’s great, anyone should live how they want or can, but the downside is for the user, the spread of fake news is huge, and that means, strategies that doesn’t work and will drain your account’s balance.
If you are a discretionary trader, and you don’t do some research on your strategies, about what technical indicators, price patterns, tests and such works on the market, you are going to have a bad time.
And if you know how to code, you are lucky, coding and algorithmic trading are not the key to become a millionaire (no, trading won’t make you rich), but at least you have a great head start when it comes to doing research.
Why research is so important
Let’s move to the facts, why research is important and why algorithmic trading is a great way to help you doing that.
Candlestick patterns research
First of all, i need to ask you something. Is Price Action profitable? I mean, candlestick patterns. Are those profitable on the market? Can you answer that?
If your answer is yes, in most cases you have been fooled by the survivorship bias.
I have to say this, i love this example, Candlestick patterns are very common in most of the popular trading books, in theory, they reflect the behavior of institutional traders. But in practice…
Practice is really different and markets have evolved since those books were written twenty years ago. We have Machine Learning (Artificial Intelligence), High Frequency Trading, Dark Pools. Everything has changed since then, so we need to validate every single hypothesis.
Coding and Backtesting Candlestick patterns
I’m going to use Tradestation, a simple platform oriented to Futures and Stocks trading with easy-to-learn code to test some common patterns out there.
As we can see, using some assets selected at random, in most cases we suffer from a random walk process, since those equity curves are completely random.
But why? I have a personal theory, it’s the disruption of High Frecuency Trading and algorithmic trading strategies like statistical arbitrage, where we use to have a reversion pattern now it can be a lot of things, maybe its just an algorithm consuming liquidity, maybe at that point the distance of a certain spread was too high and algorithms were buying one asset and selling the other forming those patterns… etc
My theory is backed by this:
This is the equity curve of a hammer pattern on a non-tradeable index, meaning that no one can buy or sell this index, just use it as an information source of the Nuclear Energy market, since this cannot be traded with anything, the price is not manipulated.
When it comes to trading, specially algorithmic trading, we test every hypothesis possible for some pattern or model to see if it worked in the past, at least, to check if it has an statistical significance at least in order to trade it.
As you can see, some things may work, other don’t. It’s a fact.
If you don’t have evidence about something, you can be fooled by randomness, meaning that, even in a losing equity curve they are some winning trades.
Want to learn how to code for free?
You can check out my Youtube channel and some of my videos, including how to code in MQL4 (Metatrader 4), some algorithmic trading strategies and there’s much more coming.
Sources for Trading Strategies
Let’s start with the point of this article after a long prelude, let’s say something like, ‘Hey, i know how to code, but what i’m supposed to do now?’
Great question, trading strategies über alles.
Besides all the disinformation they are still some good resources on the Internet to find decent algorithmic trading strategies, but be aware, you will need to work them on your own.
Not all websites are crooked with bad losing strategies, but, half of the basket contains bad apples. That means, some strategies will work, others don’t. To this date, i still don’t know any site with a 100% of profitable trading strategies. But that doesn’t mean that some sites aren’t good enough for us.
This one, for me, its the best website, since it contains a compilation of more than 1,000 strategies, it takes a while to test them, but its worth it.
Besides that you can check some trading strategies on this site.
Probably my favorite source when it comes to algorithmic trading.
Most algorithmic trading books are way to old and they need some “minor fixes and improvements” when it comes to strategies, i will talk later about that.
I intend to do some book reviews and summaries in further articles and videos.
Anyway, this books are mandatory for any trader who wants to at least, survive in the market.
Practical speculation by Victor Niederhoffer
Victor Niederhoffer is one of the forgotten fathers of Algorithmic trading and statistical arbitrage, he wrote a couple of books after losing everything in the market.
Wait! I’m going to recommend a book about someone who lost everything? Yes, indeed.
His wisdom not only comes by his knowledge but for acknowledging his own mistakes and learning from them.
This is a very interesting book if you are new to trading.
In practical speculation we can learn about why conventional technical analysis doesn’t work and some interesting ways to approach the market.
Trading Systems and Methods by Perry Kaufman
This one is a classic, a full encyclopedia about price formations and indicators to create trading systems, i can’t tell much more about this one.
It’s hard to describe it, probably the most complete book about algorithmic trading out there.
How to make any strategy profitable
When i talked about “minor fixes and improvements” to conventional trading strategies, i mean to apply a simple concept.
Try everything, play with the data, change the rules, you make them.
The fact that the RSI indicators is below 30 doesn’t mean that we have an ‘oversell’ position or anything like that until you prove it.
What happens if i short sell once the RSI is below 30? In some cases it may work.
Play with the data, create new filters, test ideas, that’s why coding is important. The market doesn’t work like 30 years ago and it’s a delusional idea to think it does. Data never lies.
As always, hope you enjoyed this article.
Victor – Follow the Edge.