Even though I shamelessly used the Christmas holidays as an excuse not to post anything for the last three weeks, the reality is that I failed at my first attempt to reach profitability.
Since I lost no money or credibility, this failure is a formidable opportunity to explain what happened and understand better how I can create conditions favorable to success.
Stock Market: A market without any customer
During my research, I identified three categories of businesses :
- The retail or service, where you must fight for every customer. Once your product is ready, your main focus is to create an acquisition channel to ensure steady customer (and, therefore, money) flow.
- The passive businesses, aka real estate, where someone else builds your product and demand is not an issue. The real problem is the starting cash and loans. In other words, it's "just" a matter of money and cash flow management, which is not that tricky but too slow for my needs.
- The no product and no customer businesses, like the trading market. You don't need a product nor any customer, nor a massive amount of cash. It's a high risk, high reward game accessible to anyone willing to take its chance.
Guess which one I was eager to start with?
Trading is easy, right?
Even though I am aware that most casual traders lose money, I figure that I have skills that most of them lack.
First, I am an INSA trained engineer and programmer with ten years of high profile companies experiences. I am also a former algorithm competitor, and I am familiar with machine learning. Finally, I am helped by a former private bank engineer that knows everything there is to know about trading.
I allocated 1000€ and four weeks to this project, and my confidence was through the roof!
I spent part of the first week understanding the stock exchanges' mechanisms. I wanted to ensure that trading met my ethical requirements and did not contribute to our civilization's demise. It was no easy task since I started with a considerable bias against the stock market. The more I understood it, the less I could see why it was terrible. Fundamentally, it is a way for companies to get funding. Coming from an investor-backed startup, it was no surprise.
Once confident that I am not messing with a shady mechanism, I spent the rest of the week learning how to use a software called Backtrader.
This software allows you to create your trading strategy in python and execute it on historical data. It's a good starting point to ensure that your strategy is viable (i.e., profitable) and optimizes it for your target market.
I spent the second and third weeks creating and training an ML model (particularly LSTM & CNN) to perform time-series predictions. It was a tremendously exciting journey, and I learn a ton, but after some success, I bump into a harsh reality.
The classic rush to action
As I would gradually realize, my initial hypotheses were flawed.
I made several choices after my "analysis" week:
- My initial investment will be 10k€.
- My target market is Euronext growth companies since their stocks are European, cheap (under 30€), and volatile enough to scalp them in a short time-span.
- My target ROI will be 8% per month. To be honest, I chose this number a bit arbitrarily (spoiler, it bit me in the ass later on).
After weeks of research and implementation and a profitable strategy based on machine learning and sentiment analysis, I fired up excel and computed my earning projections.
My average weekly ROI was ~1%, i.e., 4% per month, half my target. The first year's projection with a 10k€ initial investment totals to 4000€, which is way below my 2000€ target per month!
I also discovered that the taxation of this particular stocks are very expensive! With commission and taxation, the year's profits total to ~2500€.
Side note: Yes, I am aware of compound interests and that this strategy will make me millionaire in less than 20 years (taxes and commissions considered) but these strategies don't last forever and also I don't want to be millionaire. I want to generate base revenue this year.
A harsh realization
I took several weeks to create a whole machine learning pipeline to predict a time series where 30 min on Excel showed me that my best case scenario was not enough. I felt like a rookie, and it took me nearly two weeks to recover.
Since then, I created a backup strategy, but I don't want to miss the opportunity to establish some requirements before going into full programming mode.
Before starting on an idea, I will create a detailed business plan – Me after my first failure
It will allow me to precisely establish (and tweak) the investment (in time & money) and the expected return and target. Also, I will present the business plan to one of my advisors to enforce full honesty mode. I hope it will require me to better set my targets based on real competitors and market reality.
Finally, I will perform real-world tests more often (ideally every week) to fail (or pivot) faster.
Ready for the next step
Even though it was tough, I learned a ton about trading, machine learning, neural network, data visualization, and finance. And even though it was clearly not the shortest path to profitability, it led me to the level of understanding required to create a real-world algo-trading setup. Finally, I did created a profitable trading algorithm! Just not profitable enough...
Retrospectively however, I found it very funny (and a bit presumptuous) that it took me so much time to accept this failure. I thought that I would create a profitable business at my first shot in a couple of weeks—silly me.