How I Made $500K With Machine Learning And High Frequency Trading?

Machine Learning is the new thing in the town. Daily we hear of a new breakthrough having being made by a machine learning algorithm. We are living in the age of algorithms. The race is on for building better and better and algorithms. In this post: How I Made $500K With Machine Learning And High Frequency Trading, Jesse Spaulding claims that he made approximately $500K from 2009 t0 2010 ( around 1 year) with high frequency trading. We believe his claim to be true and don’t dispute it even if he hasn’t posted any solid proof. The video below explains high frequency trading:

You must have understood one thing; high frequency trading is highly technology intensive. You and me as retail trader don’t have the technological skills to implement it. High Frequency Trading requires cutting edge technology. It is a race against your competitors. You can only beat them with technology. Below is another video in which a high frequency trader talks about his secrets:

Let’s get back to the Jesse Spaulding and his post. We totally agree with him that this is not pure luck. Statistically your algorithm needs to have an edge over the market in order to make a profit. Now this edge can be weak or strong. Jesse Spaulding explains that he wanted a system that was reasonably confident of making money even before living trading. Of course he knew how to code. So he used it to make a few indicators that could predict price movement over short term. A couple of indicators had to confirm each other. The crux was the risk management model. Then Machine Learning was used to optimize the algorithm.

Now in this post Christian Felde questions the post made by Jesse Spaulding. Christian Felde thinks that the post does not post anything interesting. Well, we find Jesse Spaulding post to be highly interesting. We are also into machine learning and algorithmic trading. We think he did the right think in not revealing too much details about his algorithm. This is what we also do. There is too much competition. You just need a clue to guess how to do a thing. In this discussion forum someone calls it gambling. Gambling can never give you the edge. You only make profit when you are winning more than losing. It is as simple as that. In this article Jesse Spaudling has posted his follow up questions on his trading algo success.

Knowledge about the market microstructure is very important for intraday trading. This knowledge of how the market microstructure works was utilized in developing the indicators. But in the end we are dealing with statistics and probability and we are never 100% sure about the predictions. So we use a robust risk management system that can deal with probabilities and avoid big losses.

We believe that the best trading systems are those that have big reward to risk ratios. In simple terms, a good system risks $10 and makes $100 which translates into a reward to risk ratio of 10:1. Now it may not be possible to hit such a big home run. Most of the time a system that makes on average $20 for each $10 risked is a good system. In other words a system that has a reward to risk ratio of 2:1 should in the end make you a fortune. This does not mean you make a few lucky trades than lose a couple of trade. Law of averages means that on average you should win more than lose. Let’s make it clear with an example. Suppose you have a trading system that has a reward to risk ratio of 2:1 and you win 80% of the time. This means in 10 trades, you win 8 and lose 2 trades on average. It can be 7 winning trades in just 10 trade. But in 100 trades you should be able to win close to 80 trades and lose only 20 trades. So where is luck? Luck factor does not count here.

Machine Learning Will Become More Popular

Machine Learning is a new discipline that is a sub field of computer science. Machine Learning algorithms try to find patterns in the data that can be used to make predictions. Python is a language that is considered as the best language for machine learning as it has got many libraries that implement most of the machine learning algorithms. Read the post that explains a python machine learning algorithm that can beat S&P 500. Machine learning algorithms can be used easily on the weekly and the daily timeframes. You don’t need to become a speed trader.

Machine Learning now a days is being used in many fields now. Another good language for machine learning is R. Read this post in which we explain how you can use R to analyze the stock market daily. We believe in the coming decade more and more of these machine learning algorithms will be used in the market. Which will only result in retail traders losing more. What you should do is learn Python and start using it in your trading.