Quantopian wants to create a crowd sourced hedge fund by inspiring freelance quants develop algorithmic trading strategies. This is what it says: “Quantopian inspires talented people everywhere to write investment algorithms. Select authors may license their algorithms to us and get paid based on performance.” Quantopian company was founded in 2011 by John Fawcett and is based in Boston MA. The business model comprises of letting freelance quants to use Quantopian platform to develop and test winning algorithmic trading strategies. Right now there are around 70,000 quants who are using this platform. These winning algorithmic trading strategies are then picked by investor members of this platform to trade their portfolios. Quant members get a royalty on the use of their algorithmic strategies.
The algorithms are written in Python. Python in the last few years has emerged as a leading quantitative analysis language. In the last few years many new libraries have been added to it that can implement cutting edge machine learning algorithms.
In the last post I showed you how you can predict the next candle using a neural network. I have worked on the neural network and now I am getting good predictions. In the last post I tried to predict OHLC (4 outputs) using lagged OHLC. R was taking a long time to complete calculations. So I broke down the outputs into single outputs of High, Low and Close and reduced the size of the training set which reduced the calculation time considerably. All the calculations were done using R language.
Python has overtaken R as the best language for doing machine learning and statistical learning. R is an old technology. It is slow and it faces problems when you try to handle a large data set with it. Python is a new language that was developed in the last 20 years. It is easy to learn. Python is being used by big companies that includes Google. Youtube has been written in python. There are many other big companies that are using python. So it will be a good idea if you learn python.
Markets Are Changing With Algorithmic Trading Taking Over
Yes this is true. In the next few years you will see algorithmic trading taking over completely. This will mean the end of tradition trading based on technical analysis. Technical indicators like MACD, RSI, Stochastic, CCI etc were developing in 1960s. Haven’t you observed most of these indicators are giving more and more false signals. Machine learning is the new thing. In the coming few years, these machine learning algorithms will be used more and more to develop winning trading strategies and these traditional indicators like MACD, RSI, CCI etc will become obsolete. If you will still stick with these indicators you will only lose one trade after another. In short markets are changing and will change more in the next few years. Haven’t you observed candlesticks patterns not working. Most of the time you will find insider bars and pin bars giving false signals. Now a days markets reverse without any clear candlestick signal.
Let me give you an example. Richard Dennis is a trading legend. He started trading the commodities market in early 1970s. He developed a trend trading system that he named the Turtle Trading System. Turtle Trading System made a fortune for Richard Dennis and his students. Richard Dennis is known to have turned a small amount of $400 into $200 million trading with his system. His Turtle Trading System worked very well in 1980s. But today it doesn’t work. Try trading with it. It will give a poor performance. Why? Markets have changed. The way trends used to develop in 1980s were different than the way trends develop in today’s market. Trends develop today as they used to develop in the past but their dynamics have changed. Today trends are starting suddenly and ending suddenly. Why? This is happening because more and more algo trading systems are being used which has changed the way markets behave now a days. So it will be a good idea to learn Python and Machine Learning and start using it in your trading strategies. Did you read the post on John Simmons a mathematician who became a billionaire hedge fund manager?
Neural Networks are being used extensively to model financial time series. If you had read the last post in which I showed you how to develop a neural network to predict price. It has been reported that most of the hedge funds and big financial institutions are using neural networks to predict the short term price fluctuations. Then there is the Support Vector Machine algorithm that is also been used a lot in making predictions. Markov Chains are also being used. Price prediction is still a pipe dream. However, as these algorithms get more and more popular you will see the price patterns changing. So my advice become a quant and start using Quantopian to develop winning algorithmic trading strategies.