by Talaikis | Mar 23, 2016 | Investing
OK, I'm continuing my dive into machine learning. I will make two indicators - RSI, which tries to represent (as believed), mean reversion (more acurrately probably - momentum) and momentum indicator in the form of relationship between price and moving average. I will...
by Talaikis | Mar 18, 2016 | Investing
2016/03/21 fix: "Corrected" neuron counts to more conventional 4+2 (out from 10). Proper in and out samples. Layers changed to Sigmoid. Additional hidden layer. Following is an example of recurrent learning artificial neural network for returns regression, for...
by Talaikis | Mar 14, 2016 | Investing
Problem with expected return, most often referred as or for geometric returns is that we never can be sure our belief about average return would perform as well in the future. Most often they don't. Markets are noisy, markets can change their behavior, like we can see...
by Talaikis | Mar 13, 2016 | Investing
Hidden Markov Models are generative, probabilistic models, in which a sequences of visible variables are generated by some unknown ("hidden") states. In other words, we'll try to find some unknown probabilistic generative functions for, let's say, S&P500. As this...
by Talaikis | Mar 11, 2016 | Investing
This is huge subject, so I'll try it cover very fast. I'll try to find relationship between some data (in this case signal would be XLF, financial sector ETF, delayed by 1 day) and target would be S&P500 futures in the CFD form. Probably we'll go long on zero...
by Talaikis | Feb 25, 2016 | Investing
This update solves one of problems with interrupting in long running database updates. We just: Collect a list of CSV file names and write to the list file. After each file insert into database will remove it from list and update the list file. If process broken, we...
by Talaikis | Feb 24, 2016 | Investing
Unlike event driven backtesting where we do calculations on each new arrived data element, we can do simple, fast, but very flexible backtest on the entire vector at once. First, we get the returns data (will be using MongoDB from now on): Python df =...
by Talaikis | Feb 18, 2016 | Investing
In the second part I've made a crude row by row insertion of ticks to MongoDB version. Its poor performance surely can be greatly improved just not by checking type of every element in the table, but constructing the order of columns just once and then try to insert...
by Talaikis | Feb 15, 2016 | Investing
In the first part we've downloaded all 20,479 (as of time of writing) zipped data files. Didn't even knew there were so many. It took almost entire day on approx. 2-4 Mbps: Python ('Done in ', 1039.5901666680973, ' minutes') 1 ('Done in ', 1039.5901666680973, '...
by Talaikis | Feb 12, 2016 | Investing
Here's another idea of mine. I've decided to download, process and make available through my brand new MongoDB 15 years of forex tick data from GAIN (Also possible from TrueFX/ Pepperstone). The procedure, all automatic (I'm too lazy to download by hand): Walk through...